ࡱ>  B` Rebjbjss S]ڵVZZZ d DL)))x*L-LUj0|2222I4.w44 #z%z%z%zY~z~$h'-'C'4"I4'C'Cpp22PPP'Clpl22#zP'C#zPPVkU,SX20 A 2)JX oj%0U#X0χM(χSXχSX498P:<B444;Pd444U'C'C'C'CLLWfDc_LLfLLLpppppp  Automatic Extraction of Titles from General Documents using Machine Learning Yunhua Hu, Hang Li, Yunbo Cao, Li Teng, Dmitriy Meyerzon, Qinghua Zheng Feb. 15, 2006 MSR-TR-2006-17 Microsoft Research Microsoft Corporation One Microsoft Way Redmond, WA 98052 Automatic Extraction of Titles from General Documents using Machine Learning Yunhua Hu1 Computer Science Department Xian Jiaotong University No 28, Xianning West Road Xi'an, China, 710049 yunhuahu@mail.xjtu.edu.cn Dmitriy Meyerzon Microsoft Corporation One Microsoft Way Redmond, WA, USA, 98052 dmitriym@microsoft.com  Hang Li, Yunbo Cao Microsoft Research Asia 5F Sigma Center, No. 49 Zhichun Road, Haidian, Beijing, China, 100080 {hangli,yucao}@microsoft.com Qinghua Zheng Computer Science Department Xian Jiaotong University No 28, Xianning West Road Xi'an, China, 710049 qhzheng@mail.xjtu.edu.cn Li Teng Computer Science and Engineering, Chinese University of Hong Kong, Shatin, N.T., Hong Kong, China  HYPERLINK "http://by14fd.bay14.hotmail.msn.com/cgi-bin/compose?curmbox=F000000001&a=76816560e9aacdd215c43639a924648095e57c27454f6c0233a7a7ba135bda76&mailto=1&to=lteng@cse.cuhk.edu.hk&msg=MSG1125992514.13&start=1650911&len=1935&src=&type=x" lteng@cse.cuhk.edu.hk ABSTRACT In this paper, we propose a machine learning approach to title extraction from general documents. By general documents, we mean documents that can belong to any one of a number of specific genres, including presentations, book chapters, technical papers, brochures, reports, and letters. Previously, methods have been proposed mainly for title extraction from research papers. It has not been clear whether it could be possible to conduct automatic title extraction from general documents. As a case study, we consider extraction from Office including Word and PowerPoint. In our approach, we annotate titles in sample documents (for Word and PowerPoint respectively) and take them as training data, train machine learning models, and perform title extraction using the trained models. Our method is unique in that we mainly utilize formatting information such as font size as features in the models. It turns out that the use of formatting information can lead to quite accurate extraction from general documents. Precision and recall for title extraction from Word are 0.810 and 0.837 respectively, and precision and recall for title extraction from PowerPoint are 0.875 and 0.895 respectively in an experiment on intranet data. Other important new findings in this work include that we can train models in one domain and apply them to other domains, and more surprisingly we can even train models in one language and apply them to other languages. Moreover, we can significantly improve search ranking results in document retrieval by using the extracted titles. Keywords: information extraction, metadata extraction, machine learning, search INTRODUCTION Metadata of documents is useful for many kinds of document processing such as search, browsing, and filtering. Ideally, metadata is defined by the authors of documents and is then used by various systems. However, people seldom define document metadata by themselves, even when they have convenient metadata definition tools (Crystal & Land, 2003). Thus, how to automatically extract metadata from the bodies of documents turns out to be an important research issue. Methods for performing the task have been proposed. However, the focus was mainly on extraction from research papers. For instance, Han et al. (Han, Giles, Manavoglu, Zha, Zhang, & Fox, 2003) proposed a machine learning based method to conduct extraction from research papers. They formalized the problem as that of classification and employed Support Vector Machines as the classifier. They mainly used linguistic features in the model. In this paper, we consider metadata extraction from general documents. By general documents, we mean documents that may belong to any one of a number of specific genres. General documents are more widely available in digital libraries, intranets and the internet, and thus investigation on extraction from them is sorely needed. Research papers usually have well-formed styles and noticeable characteristics. In contrast, the styles of general documents can vary greatly. It has not been clarified whether a machine learning based approach can work well for this task. There are many types of metadata: title, author, date of creation, etc. As a case study, we consider title extraction in this paper. General documents can be in many different file formats: Microsoft Office, PDF (PS), etc. As a case study, we consider extraction from Office including Word and PowerPoint. We take a machine learning approach. We annotate titles in sample documents (for Word and PowerPoint respectively) and take them as training data to train several types of models, and perform title extraction using any one type of the trained models. In the models, we mainly utilize formatting information such as font size as features. We employ the following models: Perceptron with Uneven Margins, Maximum Entropy (ME), Maximum Entropy Markov Model (MEMM), Voted Perceptron Model (VP), and Conditional Random Fields (CRF).. In this paper, we also investigate the following three problems, which did not seem to have been examined previously. (1) Comparison between models: among the models above, which model performs best for title extraction; (2) Generality of model: whether it is possible to train a model on one domain and apply it to another domain, and whether it is possible to train a model in one language and apply it to another language; (3) Usefulness of extracted titles: whether extracted titles can improve document processing such as search. Experimental results indicate that our approach works well for title extraction from general documents. Our method can significantly outperform the baselines: one that always uses the first lines as titles and the other that always uses the lines in the largest font sizes as titles. Precision and recall for title extraction from Word are 0.810 and 0.837 respectively, and precision and recall for title extraction from PowerPoint are 0.875 and 0.895 respectively. It turns out that the using of format features is the key to successful title extraction. (1) We have observed that Perceptron based models perform better in terms of extraction accuracies. (2) We have empirically verified that the models trained with our approach are generic in the sense that they can be trained on one domain and applied to another, and they can be trained in one language and applied to another. (3) We have found that using the extracted titles we can significantly improve precision of document retrieval (by 10%). We conclude that we can indeed conduct reliable title extraction from general documents and use the extracted results to improve real applications. The rest of the paper is organized as follows. In section 2, we introduce related work, and in section 3, we explain the motivation and problem setting of our work. In section 4, we describe our method of title extraction, and in section 5, we describe our method of document retrieval using extracted titles. Section 6 gives our experimental results. We make concluding remarks in section 7. RELATED WORK Document Metadata Extraction Methods have been proposed for performing automatic metadata extraction from documents; however, the main focus was on extraction from research papers. The proposed methods fall into two categories: the rule based approach and the machine learning based approach. Giuffrida et al. (Giuffrida, Shek, & Yang, 2000), for instance, developed a rule-based system for automatically extracting metadata from research papers in Postscript. They used rules like titles are usually located on the upper portions of the first pages and they are usually in the largest font sizes. Liddy et al. (Liddy, Sutton, Allen, Harwell, Corieri, Yilmazel, Ozgencil, Diekema, McCracken, & Silverstein, 2002) and Yilmazel el al. (Yilmazel, Finneran, & Liddy, 2004) performed metadata extraction from educational materials using rule-based natural language processing technologies. Mao et al. (Mao, Kim, & Thoma, 2004) also conducted automatic metadata extraction from research papers using rules on formatting information. The rule-based approach can achieve high performance. However, it also has disadvantages. It is less adaptive and robust when compared with the machine learning approach. Han et al. (Han, Giles, Manavoglu, Zha, Zhang, & Fox, 2003), for instance, conducted metadata extraction with the machine learning approach. They viewed the problem as that of classifying the lines in a document into the categories of metadata and proposed using Support Vector Machines (SVM) as the classifier. They mainly used linguistic information as features. They reported high extraction accuracy from research papers in terms of precision and recall. Peng et al. (Peng & McCallum, 2004) also conducted information extraction from research papers. They employed Conditional Random Fields (CRF) as model. Information Extraction Metadata extraction can be viewed as an application of information extraction, in which given a sequence of instances, we identify a subsequence that represents information in which we are interested. Hidden Markov Model (Ghahramani & Jordan, 1997), Maximum Entropy Model (Berger, Della Pietra, & Della Pietra, 1996; Chieu & Ng, 2002), Maximum Entropy Markov Model (McCallum, Freitag, & Pereira, 2000), Support Vector Machines (Cortes & Vapnik, 1995), Conditional Random Field (Lafferty, McCallum, & Pereira, 2001), and Voted Perceptron (Collins, 2002) are widely used information extraction models. Information extraction has been applied, for instance, to part-of-speech tagging (Ratnaparkhi, 1998), named entity recognition (Zhang, Pan, & Zhang, 2004) and table extraction (Pinto, McCallum, Wei, & Croft, 2003). Search Using Title Information Title information is useful for document retrieval. In the system Citeseer, for instance, Giles et al. managed to extract titles from research papers and make use of the extracted titles in metadata search of papers (Giles, Petinot, Teregowda, Han, Lawrence, Rangaswamy, & Pal, 2003). In web search, the title fields (i.e., file properties) and anchor texts of web pages (HTML documents) can be viewed as titles of the pages (Evans, Klavans, & McKeown, 2004). Many search engines seem to utilize them for web page retrieval (Gheel, Anderson, 1999; Kobayashi & Takeda, 2000; Murphy, 1998; Yi & Sundaresan, 2000]. Zhang et al. (Zhang & Dimitroff, 2004), found that web pages with well-defined metadata are more easily retrieved than those without well-defined metadata. To the best of our knowledge, no research has been conducted on using extracted titles from general documents (e.g., Office documents) for search of the documents. MOTIVATION AND PROBLEM SETTING We consider the issue of automatically extracting titles from general documents. By general documents, we mean documents that belong to one of any number of specific genres. The documents can be presentations, books, book chapters, technical papers, brochures, reports, memos, specifications, letters, announcements, or resumes. General documents are more widely available in digital libraries, intranets, and internet, and thus investigation on title extraction from them is sorely needed. Fig. 1 shows an estimate on distributions of file formats on intranet and internet (Littlefield, 2002). Office and PDF are the main file formats on the intranet. Even on the internet, the documents in the formats are still not negligible, given its extremely large size. In this paper, without loss of generality, we take Office documents as an example.  Fig. 1. Distributions of file formats in internet and intranet. For Office documents, users can define titles as file properties using a feature provided by Office. We found in an experiment, however, that users seldom use the feature and thus titles in file properties are usually very inaccurate. That is to say, titles in file properties are usually inconsistent with the true titles in the file bodies that are created by the authors and are visible to readers. We collected 6 000 Word and 6 000 PowerPoint documents from an intranet and the internet and examined how many titles in the file properties are correct. We found that surprisingly the accuracy was only 0.265 (cf., Section 6.3 for details). A number of reasons can be considered. For example, if one creates a new file by copying an old file, then the file property of the new file will also be copied from the old file. In another experiment, we found that Google uses the titles in file properties of Office documents in search and browsing, but the titles are not very accurate. We created 50 queries to search Word and PowerPoint documents and examined the top 15 results of each query returned by Google. We found that nearly all the titles presented in the search results were from the file properties of the documents. However, only 0.272 of them were correct. Actually, true titles usually exist at the beginnings of the bodies of documents. If we can accurately extract the titles from the bodies of documents, then we can exploit reliable title information in document processing. This is exactly the problem we address in this paper. More specifically, given a Word document, we are to extract the title from the top region of the first page. Given a PowerPoint document, we are to extract the title from the first slide. A title sometimes consists of a main title and one or two subtitles. We only consider extraction of the main title. As baselines for title extraction, we use that of always using the first lines as titles and that of always using the lines with largest font sizes as titles. Next, we define a specification for human judgments in title data annotation. The annotated data will be used in training and testing of the title extraction methods. Summary of the specification: The title of a document should be identified on the basis of common sense, if there is no difficulty in the identification. However, there are many cases in which the identification is not easy. There are some rules defined in the specification that guide identification for such cases. The rules include a title is usually in consecutive lines in the same format, a document can have no title, titles in images are not considered, a title should not contain words like draft, whitepaper, etc, if it is difficult to determine which is the title, select the one in the largest font size, and if it is still difficult to determine which is the title, select the first candidate. (The specification covers all the cases we have encountered in data annotation.)  Fig. 2. Title extraction from Word document.  Fig. 3. Title extraction from PowerPoint document. Figs. 2 and 3 show examples of Office documents from which we conduct title extraction. In Fig. 2, Differences in Win32 API Implementations among Windows Operating Systems is the title of the Word document. Microsoft Windows on the top of this page is a picture and thus is ignored. In Fig. 3, Building Competitive Advantages through an Agile Infrastructure is the title of the PowerPoint document (note that the phrase Agile Infrastructure is in a different font style). We have developed a tool for annotation of titles by human annotators. Fig. 4 shows a snapshot of the tool.  Fig. 4. Title annotation tool. TITLE EXTRACTION METHOD Outline Title extraction based on machine learning consists of training and extraction. The same pre-processing step occurs before training and extraction. During pre-processing, from the top region of the first page of a Word document or the first slide of a PowerPoint document a number of units for processing are extracted. If a line (lines are separated by return symbols) only has a single format, then the line will become a unit. If a line has several parts and each of them has its own format, then each part will become a unit. Each unit will be treated as an instance in learning. A unit contains not only content information (linguistic information) but also formatting information. The input to pre-processing is a document and the output of pre-processing is a sequence of units (instances). Fig. 5 shows the units obtained from the document in Fig. 2.  Fig. 5. Example of units. In learning, the input is sequences of units where each sequence corresponds to a document. We take labeled units (labeled as title_begin, title_end, or other) in the sequences as training data and construct models for identifying whether a unit is title_begin title_end, or other. We employ five types of models: Perceptron with Uneven Margins, Maximum Entropy (ME), Maximum Entropy Markov Model (MEMM), Voted Perceptron Model (VP), and Conditional Random Fields (CRF).. In extraction, the input is a sequence of units from one document. We employ one type of model to identify whether a unit is title_begin, title_end, or other. We then extract units from the unit labeled with title_begin to the unit labeled with title_end. The result is the extracted title of the document. The unique characteristic of our approach is that we mainly utilize formatting information for title extraction. Our assumption is that although general documents vary in styles, their formats have certain patterns and we can learn and utilize the patterns for title extraction. This is in contrast to the work by Han et al., in which only linguistic features are used for extraction from research papers. Models The five models actually can be considered in the same metadata extraction framework. That is why we apply them together to our current problem. Each input is a sequence of instances  EMBED Equation.3 together with a sequence of labels EMBED Equation.3 .  EMBED Equation.3 and EMBED Equation.3  represents an instance and its label, respectively ( EMBED Equation.3 ). Recall that an instance here represents a unit. A label represents title_begin, title_end, or other. Here, k is the number of units in a document. In learning, we train a model which can be generally denoted as a conditional probability distribution  EMBED Equation.3  where  EMBED Equation.3  and  EMBED Equation.3  denote random variables taking instance  EMBED Equation.3  and label  EMBED Equation.3 as values, respectively ( EMBED Equation.3 ).  Fig. 6. Metadata extraction model. We can make assumptions about the general model in order to make it simple enough for training. For example, we can assume that  EMBED Equation.3 are independent of each other given EMBED Equation.3 . Thus, we have  EMBED Equation.3  In this way, we decompose the model into a number of classifiers. We train the classifiers locally using the labeled data. As the classifier, we employ the Perceptron or Maximum Entropy model. In this paper, for Perceptron, we actually employ an improved variant of it, called Perceptron with Uneven Margin (Li, Zaragoza, Herbrich, Shawe-Taylor, & Kandola, 2002). This version of Perceptron can work well especially when the number of positive instances and the number of negative instances differ greatly, which is exactly the case in our problem. We can also assume that the first order Markov property holds for  EMBED Equation.3 given EMBED Equation.3 . Thus, we have  EMBED Equation.3  Again, we obtain a number of classifiers. However, the classifiers are conditioned on the previous label. When we employ the Maximum Entropy model as a classifier, the models become a Maximum Entropy Markov Model. That is to say, this models is more precise. If we do not make the above assumptions, we can make use of Conditional Random Fields (CRF), which is trained globally (i.e., at entire the sequence level). CRF usually performs better than the other models. Voted Perceptron (VP) proposed by Collins (Collins, 2002) can be viewed as a simplified version of CRF. We use both CRF and VP in this paper. In extraction, given a new sequence of instances, we resort to one of the constructed models to assign a sequence of labels to the sequence of instances, i.e., perform extraction. For Perceptron and ME, we assign labels locally and combine the results globally later using heuristics. Specifically, we first identify the most likely title_begin. Then we find the most likely title_end within three units after the title_begin. Finally, we extract as a title the units between the title_begin and the title_end. For MEMM, VP and CRF, we employ the Viterbi algorithm to find the globally optimal label sequence. Features There are two types of features: format features and linguistic features. We mainly use the former. The features are used for both the title-begin and the title-end classifiers. Format Features Font Size: There are four binary features that represent the normalized font size of the unit (recall that a unit has only one type of font). If the font size of the unit is the largest in the document, then the first feature will be 1, otherwise 0. If the font size is the smallest in the document, then the fourth feature will be 1, otherwise 0. If the font size is above the average font size and not the largest in the document, then the second feature will be 1, otherwise 0. If the font size is below the average font size and not the smallest, the third feature will be 1, otherwise 0. It is necessary to conduct normalization on font sizes. For example, in one document the largest font size might be 12pt, while in another the smallest one might be 18pt. Boldface: This binary feature represents whether or not the current unit is in boldface. Alignment: There are four binary features that respectively represent the location of the current unit: left, center, right, and unknown alignment. The following format features with respect to context play an important role in title extraction. Empty Neighboring Unit: There are two binary features that represent, respectively, whether or not the previous unit and the current unit are blank lines. Font Size Change: There are two binary features that represent, respectively, whether or not the font size of the previous unit and the font size of the next unit differ from that of the current unit. Alignment Change: There are two binary features that represent, respectively, whether or not the alignment of the previous unit and the alignment of the next unit differ from that of the current one. Same Paragraph: There are two binary features that represent, respectively, whether or not the previous unit and the next unit are in the same paragraph as the current unit. Linguistic Features The linguistic features are based on key words. Positive Word: This binary feature represents whether or not the current unit begins with one of the positive words. The positive words include title:, subject:, subject line: For example, in some documents the lines of titles and authors have the same formats. However, if lines begin with one of the positive words, then it is likely that they are title lines. Negative Word: This binary feature represents whether or not the current unit begins with one of the negative words. The negative words include To, By, created by, updated by, etc. This is because titles usually do not start with such words. There are more negative words than positive words. The above linguistic features are language dependent. Word Count: A title should not be too long. We heuristically create four intervals: [1, 2], [3, 6], [7, 9] and [9, EMBED Equation.3 ) and define one feature for each interval. If the number of words in a title falls into an interval, then the corresponding feature will be 1; otherwise 0. Ending Character: This feature represents whether the unit ends with :, -, or other special characters. A title usually does not end with such a character. DOCUMENT RETRIEVAL METHOD We describe our method of document retrieval using extracted titles. Typically, in information retrieval a document is split into a number of fields including body, title, and anchor text. A ranking function in search can use different weights for different fields of the document. Also, titles are typically assigned high weights, indicating that they are important for document retrieval. As explained previously, our experiment has shown that a significant number of documents actually have incorrect titles in the file properties, and thus in addition of using them we use the extracted titles as one more field of the document. By doing this, we attempt to improve the overall precision. In this paper, we employ a modification of BM25 that allows field weighting (Robertson, Zaragoza, & Taylor, 2004). As fields, we make use of body, title, extracted title and anchor. First, for each term in the query we count the term frequency in each field of the document; each field frequency is then weighted according to the corresponding weight parameter:  EMBED Equation.3  Similarly, we compute the document length as a weighted sum of lengths of each field. Average document length in the corpus becomes the average of all weighted document lengths.  EMBED Equation.3  In our experiments we used EMBED Equation.3 . Weight for content was 1.0, title was 10.0, anchor was 10.0, and extracted title was 5.0. EXPERIMENTAL RESULTS Data Sets and Evaluation Measures We used three data sets in our experiments. First, we downloaded and randomly selected 5,000 Word documents and 5,000 PowerPoint documents from an intranet of Microsoft. We call it MS hereafter. Second, we downloaded and randomly selected 500 Word and 500 PowerPoint documents from the DotGov and DotCom domains on the internet, respectively. Third, we downloaded and randomly selected 4,000 Word and 4,000 PowerPoint documents written in three other languages, including 4,000 documents in Chinese, 2,000 documents in Japanese and 2,000 documents in German. We named them as Chinese, Japanese and German respectively in the following. Fig. 7 shows the distributions of the genres of the documents. We see that the documents are indeed general documents as we define them.  Fig. 7. Distributions of document genres. We manually labeled the titles of all the documents, on the basis of our specification. Not all the documents have titles. Table 1 shows the percentages of the documents in the first two data sets having titles. We see that DotCom and DotGov have more PowerPoint documents with titles than MS. This might be because PowerPoint documents published on the internet are more formal than those on the intranet. Table 1 The portion of documents with titles Domain TypeMSDotComDotGovWord75.7%77.8%75.6%PowerPoint82.1%93.4%96.4% In our experiments, we conducted evaluations on title extraction in terms of precision, recall, and F-measure. The evaluation measures are defined as follows: Precision: P = A / ( A + B ) Recall: R = A / ( A + C ) F-measure: F1 = 2PR / ( P + R ) Here, A, B, C, and D are numbers of documents as those defined in Table 2. Table 2 Contingence table with regard to title extraction Is titleIs not titleExtractedABNot extractedCDBaselines We test the accuracies of the two baselines described in section 3. They are denoted as largest font size and first line respectively. Accuracy of Titles in File Properties We investigate how many titles in the file properties of the documents are reliable. We view the titles annotated by humans as true titles and test how many titles in the file properties can approximately match with the true titles. We use Edit Distance to conduct the approximate match. (Approximate match is only used in this evaluation). This is because sometimes human annotated titles can be slightly different from the titles in file properties on the surface, e.g., contain extra spaces). Given string A and string B: if ( (D == 0) or ( D / ( La + Lb ) <  ) ) then string A = string B D: Edit Distance between string A and string B La: length of string A Lb: length of string B : 0.1 Table 3 Accuracies of titles in file properties File TypeDomainPrecisionRecallF1WordMS0.2990.3110.305DotCom0.2100.2140.212DotGov0.1820.1770.180PowerPointMS0.2290.2450.237DotCom0.1850.1860.186DotGov0.1800.1820.181 Comparison with Baselines We conducted title extraction from the first data set (Word and PowerPoint in MS). As the model, we used Perceptron. We conduct 4-fold cross validation. Thus, all the results reported here are those averaged over 4 trials. Tables 4 and 5 show the results. We see that Perceptron significantly outperforms the baselines. In the evaluation, we use exact matching between the true titles annotated by humans and the extracted titles. Table 4 Accuracies of title extraction with Word PrecisionRecallF1ModelPerceptron0.8100.8370.823BaselinesLargest font size0.7000.7580.727First line0.7070.7670.736 Table 5 Accuracies of title extraction with PowerPoint PrecisionRecallF1ModelPerceptron0.8750. 8950.885BaselinesLargest font size0.8440.8870.865First line0.6390.6710.655 We see that the machine learning approach can achieve good performance in title extraction. For Word documents both precision and recall of the approach are 8 percent higher than those of the baselines. For PowerPoint both precision and recall of the approach are 2 percent higher than those of the baselines. We conduct significance tests. The results are shown in Table 6. Here, Largest denotes the baseline of using the largest font size, First denotes the baseline of using the first line. The results indicate that the improvements of machine learning over baselines are statistically significant (in the sense p-value < 0.05) Table 6 Sign test results Documents TypeSign test betweenp-valueWord Perceptron vs. Largest3.59e-26Perceptron vs. First7.12e-10PowerPoint Perceptron vs. Largest0.010Perceptron vs. First5.13e-40 We see, from the results, that the two baselines can work well for title extraction, suggesting that font size and position information are most useful features for title extraction. However, it is also obvious that using only these two features is not enough. There are cases in which all the lines have the same font size (i.e., the largest font size), or cases in which the lines with the largest font size only contain general descriptions like Confidential, White paper, etc. For those cases, the largest font size method cannot work well. For similar reasons, the first line method alone cannot work well, either. With the combination of different features (evidence in title judgment), Perceptron can outperform Largest and First. We investigate the performance of solely using linguistic features. We found that it does not work well. It seems that the format features play important roles and the linguistic features are supplements. Fig. 8 shows the weights of some randomly selected features in the Perceptron models. Features 1 to 10 are format features and features 11 to 20 are linguistic features. Although linguistic features tend to have larger weights, usually titles do not contain such features.  Fig. 8. Weights of some features in Perceptron models for Word. We conducted an error analysis on the results of Perceptron. We found that the errors fell into three categories. (1) About one third of the errors were related to hard cases. In these documents, the layouts of the first pages were difficult to understand, even for humans. Fig. 9 and 10 shows examples. (2) Nearly one fourth of the errors were from the documents which do not have true titles but only contain bullets. Since we conduct extraction from the top regions, it is difficult to get rid of these errors with the current approach. (3). Confusions between main titles and subtitles were another type of error. Since we only labeled the main titles as titles, the extractions of both titles were considered incorrect. This type of error does little harm to document processing like search, however.  Fig. 9. An example Word document.  Fig. 10. An example PowerPoint document. Comparison between Models To compare the performance of different machine learning models, we conducted another experiment. Again, we perform 4-fold cross validation on the first data set (MS). Table 7, 8 shows the results of all the five models. It turns out that CRF performs the best, followed by Perceptron and VP, then MEMM, and ME performs the worst. In general, the Markovian models perform better than or as well as their classifier counterparts. This seems to be because the Markovian models are based on local histories and thus can more effectively make use of context information than the classifier models. The Perceptron based models perform better than the ME based counterparts. This seems to be because the Perceptron based models are created to make better classifications, while ME models are constructed for better prediction. It is not surprising to see that CRF performs the best, because it employs a globally (sequence level) trained model. Because our main focus in this paper is to conduct feasibility study of title extraction from general documents, we utilized the simplest model Perceptron in our experiments. Table 7 Comparison between different learning models for title extraction with Word ModelPrecisionRecallF1Perceptron0.8100.8370.823ME0.8010.6210.699MEMM0.7970.8240.810VP0.8270.8230.825CRF0.8340.8430.838Table 8 Comparison between different learning models for title extraction with PowerPoint ModelPrecisionRecallF1Perceptron0.8750. 8950. 885ME0.7530.7660.759MEMM0.8410.8610.851VP0.8730.8960.885CRF0.8800.8910.886Domain Adaptation We apply the model trained with the first data set (MS) to the second data set (DotCom and DotGov). Tables 9-12 show the results. Table 9 Accuracies of title extraction with Word in DotGov PrecisionRecallF1ModelPerceptron0.7160.7590.737Baselines Largest font size0.5490.6190.582First line0.4620.5210.490 Table 10 Accuracies of title extraction with PowerPoint in DotGov PrecisionRecallF1ModelPerceptron0.9000.9060.903Baselines Largest font size0.8710.8880.879First line0.5540.5640.559 Table 11 Accuracies of title extraction with Word in DotCom PrecisionRecallF1ModelPerceptron0.8320.8800.855Baselines Largest font size0.6760.7530.712First line0.5770.6430.608 Table 12 Performance of PowerPoint document title extraction in DotCom PrecisionRecallF1ModelPerceptron0.9100.9030.907Baselines Largest font size0.8640.8860.875First line0.5700.5850.577From the results, we see that the models can be adapted to different domains well. There is almost no drop in accuracy. The results indicate that the patterns of title formats exist across different domains, and it is possible to construct a domain independent model by mainly using formatting information. Language Adaptation We apply the model trained with the data in English (MS) to the data set in third data set (Chinese, Japanese and German). Tables 13-18 show the results. Table 13 Accuracies of title extraction with Word in Chinese PrecisionRecallF1ModelPerceptron0.8610.8430.852Baselines Largest font size0.4360.9080.589First line0.6890.6040.644 Table 14 Accuracies of title extraction with PowerPoint in Chinese PrecisionRecallF1ModelPerceptron0.8580.8780.868Baselines Largest font size0.8230.9500.882First line0.8450.6690.747 Table 15 Accuracies of title extraction with Word in Japanese PrecisionRecallF1ModelPerceptron0.7440.7410.743Baselines Largest font size0.4360.9080.589First line0.6890.6040.644 Table 16 Accuracies of title extraction with PowerPoint in Japanese PrecisionRecallF1ModelPerceptron0.8750.9100.892Baselines Largest font size0.8280.9650.891First line0.8030.6160.697 Table 17 Accuracies of title extraction with Word in German PrecisionRecallF1ModelPerceptron0.7230.7290.726Baselines Largest font size0.4360.9080.589First line0.6890.6040.644 Table 18 Accuracies of title extraction with PowerPoint in German PrecisionRecallF1ModelPerceptron0.8200.8410.830Baselines Largest font size0.7290.8980.805First line0.7970.5860.675 We see that the models can be adapted to a different language. There are only small drops in accuracy. Obviously, the linguistic features do not work for other languages, but the effect of not using them is negligible. The results indicate that the patterns of title formats exist across different languages. From the domain adaptation and language adaptation results, we conclude that the use of formatting information is the key to a successful extraction from general documents. Search with Extracted Titles We performed experiments on using title extraction for document retrieval. As a baseline, we employed BM25 without using extracted titles. The ranking mechanism was as described in Section 5. The weights were determined manually. We did not conduct optimization on the weights. The evaluation was conducted on a corpus of 1.3 M documents crawled from the intranet of Microsoft using 100 evaluation queries obtained from this intranets search engine query logs. 50 queries were from the most popular set, while 50 queries other were chosen randomly. Users were asked to provide judgments of the degree of document relevance from a scale of 1 to 5 (1 means detrimental, 2 bad, 3 fair, 4 good and 5 excellent). Fig. 11 shows the results. In the chart two sets of precision results were obtained by either considering good or excellent documents as relevant (left 3 bars with relevance threshold 0.5), or by considering only excellent documents as relevant (right 3 bars with relevance threshold 1.0) Fig. 11 shows different document retrieval results with different ranking functions in terms of precision @10, precision @5 and reciprocal rank: Blue bar BM25 including the fields body, title (file property), and anchor text. Purple bar BM25 including the fields body, title (file property), anchor text, and extracted title. With the additional field of extracted title included in BM25 the precision @10 increased from 0.132 to 0.145, or by ~10%. Thus, it is safe to say that the use of extracted title can indeed improve the precision of document retrieval.  Fig. 11. Search ranking results. CONCLUSION In this paper, we have investigated the problem of automatically extracting titles from general documents. We have tried using a machine learning approach to address the problem. Previous work showed that the machine learning approach can work well for metadata extraction from research papers. In this paper, we showed that the approach can work for extraction from general documents as well. Our experimental results indicated that the machine learning approach can work significantly better than the baselines in title extraction from Office documents. Previous work on metadata extraction mainly used linguistic features in documents, while we mainly used formatting information. It appeared that using formatting information is a key for successfully conducting title extraction from general documents. We tried different machine learning models including Perceptron, Maximum Entropy, Maximum Entropy Markov Model, Voted Perceptron and Conditional Random Fields. We found that the performance of the Conditional Random Fields models were the best. We applied models constructed in one domain to another domain and applied models trained in one language to another language. We found that the accuracies did not drop substantially across different domains and across different languages, indicating that the models were generic. We also attempted to use the extracted titles in document retrieval. We observed a significant improvement in document ranking performance for search when using extracted title information. All the above investigations were not conducted in previous work, and through our investigations we verified the generality and the significance of the title extraction approach. ACKNOWLEDGEMENTS We thank Chunyu Wei and Bojuan Zhao for their work on data annotation. We acknowledge Jinzhu Li for his assistance in conducting the experiments. We thank Ming Zhou, John Chen, and Jun Xu for their valuable comments on early versions of this paper. We also thank the anonymous reviewers of this paper for their making many valuable comments. REFERENCES Berger, A. L., Della Pietra, S. A. & Della Pietra, V. J. (1996). A maximum entropy approach to natural language processing. Computational Linguistics, 22:39-71. Crystal, A. & Land, P. (2003). Metadata and Search Global Corporate Circle DCMI 2003 Workshop. http://dublincore.org/groups/corporate/Seattle/ Collins, M. (2002). Discriminative training methods for hidden markov models: theory and experiments with perceptron algorithms. In Proceedings of Conference on Empirical Methods in Natural Language Processing, (pp. 1-8).. Cortes, C. & Vapnik, (1995). V. Support-vector networks. Machine Learning, 20:273-297.. Chieu, H. L. & Ng, H. T. (2002). A maximum entropy approach to information extraction from semi-structured and free text. In Proceedings of the Eighteenth National Conference on Artificial Intelligence, (pp. 768-791). Evans, D. K., Klavans, J. L. & McKeown, K. R. (2004). Columbia newsblaster: multilingual news summarization on the Web. In Proceedings of Human Language Technology conference / North American chapter of the Association for Computational Linguistics annual meeting, (pp. 1-4). Ghahramani, Z. & Jordan, M. I. (1997). Factorial hidden markov models. Machine Learning, 29:245-273. Gheel, J. & Anderson, T. (1999). Data and metadata for finding and reminding, In Proceedings of the 1999 International Conference on Information Visualization, (pp. 446-451). Giles, C. L., Petinot, Y., Teregowda, P. B., Han, H., Lawrence, S., Rangaswamy, A. & Pal, N. (2003). eBizSearch: a niche search engine for e-Business. In Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, (pp. 413-414). Giuffrida, G., Shek, E. C. & Yang, J. (2000). Knowledge-based metadata extraction from PostScript files. In Proceedings of the Fifth ACM Conference on Digital Libraries, (pp. 77-84). Han, H., Giles, C. L., Manavoglu, E., Zha, H., Zhang, Z. & Fox, E. A. (2003). Automatic document metadata extraction using support vector machines. In Proceedings of the Third ACM/IEEE-CS Joint Conference on Digital Libraries, (pp. 37-48). Kobayashi, M. & Takeda, K. (2000). Information retrieval on the Web. ACM Computing Surveys, 32:144-173. Lafferty, J., McCallum, A. & Pereira, F. (2001). Conditional random fields: probabilistic models for segmenting and labeling sequence data. In Proceedings of the Eighteenth International Conference on Machine Learning, (pp. 282-289). Li, Y., Zaragoza, H., Herbrich, R., Shawe-Taylor, J., & Kandola, J. S. (2002) The perceptron algorithm with uneven margins. In Proceedings of the Nineteenth International Conference on Machine Learning, (pp. 379-386). Liddy, E. D., Sutton, S., Allen, E., Harwell, S., Corieri, S., Yilmazel, O., Ozgencil, N. E., Diekema, A., McCracken, N. & Silverstein, J. (2002). Automatic Metadata generation & evaluation. In Proceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, (pp. 401-402). Littlefield, A. (2002). Effective enterprise information retrieval across new content formats. In Proceedings of the Seventh Search Engine Conference, http://www.infonortics.com/searchengines/sh02/02prog.html. Mao, S., Kim, J. W. & Thoma, G. R. (2004). A dynamic feature generation system for automated metadata extraction in preservation of digital materials. In Proceedings of the First International Workshop on Document Image Analysis for Libraries, (pp. 225-232). McCallum, A., Freitag, D. & Pereira, F. (2000). Maximum entropy markov models for information extraction and segmentation. In Proceedings of the Seventeenth International Conference on Machine Learning, (pp. 591-598). Murphy, L. D. (1998). Digital document metadata in organizations: roles, analytical approaches, and future research directions. In Proceedings of the Thirty-First Annual Hawaii International Conference on System Sciences, (pp. 267-276) Peng, F. & McCallum, A. (2004) Accurate Information Extraction from Research Papers using Conditional Random Fields. In Proceedings of the Human Language Technology conference / North American chapter of the Association for Computational Linguistics annual meeting, (pp.329-336) Pinto, D., McCallum, A., Wei, X. & Croft, W. B. (2003) Table extraction using conditional random fields. In Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, (pp. 235-242) Ratnaparkhi, A. (1998). Unsupervised statistical models for prepositional phrase attachment. In Proceedings of the Seventeenth International Conference on Computational Linguistics, (pp.1079-1085) Robertson, S., Zaragoza, H. & Taylor, M. (2004). Simple BM25 extension to multiple weighted fields, In Proceedings of ACM Thirteenth Conference on Information and Knowledge Management, (pp. 42-49). Yi, J. & Sundaresan, N. (2000). Metadata based Web mining for relevance, In Proceedings of the 2000 International Symposium on Database Engineering & Applications, (pp. 113-121). Yilmazel, O., Finneran, C. M. & Liddy, E. D. (2004). MetaExtract: An NLP system to automatically assign metadata. In Proceedings of the 2004 Joint ACM/IEEE Conference on Digital Libraries, (pp. 241-242). Zhang, J. & Dimitroff, A. (2004). Internet search engines' response to metadata Dublin Core implementation. Journal of Information Science, 30:310-320 Zhang, L., Pan, Y. & Zhang, T. (2004). Recognising and using named entities: focused named entity recognition using machine learning. In Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, (pp. 281-288).  The work was conducted when the first author was visiting Microsoft Research Asia.     PAGE   EMBED Equation.3    Y[   c d e n o p y Ƚ}pf_fXf_ hf@o( h]@y(|hih]@y(|h]@nHo(tHy(|&haha@CJH*aJnHo(tHh]@o(y(| hCy(| hCo(y(| hiy(| h+65CJ h+65 h+6CJh+6CJnHo(tHh+6h+65CJ h+6h+6hFh+6OJQJnHtHh+65CJ nHo(tH h+65CJ h+6     YZ[ $ $If^gd8i$ p]p^a$gd+6$ $ p]p^a$gd+6ھ0Hd  d e p gd]<$ & F $ pVD]p^a$gd+6$ $ p]p^a$gd+67kd$$Ifl g4 la    7 8 B C Y Z [ \ ] d e o ۹xqg[T h]@y(|hh]@o(y(|h^vh^v@o( hC@y(|h^vh^v@o(hd)htl@y(| htl@o( htl@y(|hkhtl@y(|hhtl@y(|h]@nHo(tHhh]@ hm@hQ@o(y(|hih]@y(|h]@o(y(| hf@o(hh@y(|hh]@y(|p  C Z p c | } gdd)gdm+OgdTgdTgd]gd'gd'gd^vgdtlgdtlgdtlgd]gd^vgd]     3 4 D I M N T [ \ b c d j Ǽp^#h~h'@mHnHo(sHtH h~hm@mHnHsHtHh'@o(y(|hh<*@y(|hh'@y(| h'@y(|hih'@y(|h'@nHo(tHy(|h'@nHo(tHh]@nHo(tHh^v@o(y(|h]h]@o(h]@o(y(| hf@o(hZh]@y(|j { | ~    3 ; ? E L O e f h ¸xof]U]M]M]M]hf-nHtHhq0nHtHhq0nHo(tHh3nHo(tHhC5CJy(|hsnHo(tHy(|hsnHo(tH hCy(|hm+Oh]@y(|hFhT@jhFhT@U hT@o(hFhT@o(h~hT@mHo(sHh~hC@mHsHy(|#h~h]@mHnHo(sHtHh~h'@mHo(sHy(|  $tuU Exf3 "!#$$x` x`gdKnx $xa$gdLgd3$a$h $&013:DE]ACU8G]{14Ҿʲ楚hSnHtHh7h7nHtHh?(nHo(tHhh+h hfIhKnhfIy(|hhKnhxf9y(|hsh\^hZ!Ehd6^hxf9y(|hxf9h"nHo(tHh"hKnhf-y(|645:?D 78=ELMSWXehkp#$ȯȧ}hNnHtHh_G hdy(|hInHo(tHhNhKnhdy(|hdhdnHtHh_GnHtHh7nHo(tHhBxcnHtHhxf9nHtHhs;nHo(tHhSnHtHh7h7nHtHhwnHo(tHh7hsnHtH-$,-./:;DFOY[cmnstuϷwof[fWhChjHhjHnHtHhjHnHo(tHhkTnHtHhYnHo(tHhKnhho(y(|hKnhkTo(y(| hKnhhkThKnhhy(| hCy(|hLnHo(tHy(|hCnHo(tH hXo( hU>o( ho(hp=hSy(|hp=hp=y(| hSo(hLhLo(hLhL6o(hLhC6y(|"z   U^cemrvvq h_+}y(|h*nHtHy(|h+nHtHy(|hnnHo(tHy(|h_+}nHtHy(|h,nHo(tHy(|h nHo(tHy(|hd)nHtHy(|h{h"jhaha0JB*UphhsC1hKnhho(y(|hahanHtHhanHo(tHhKnhhy(|*:>?"')   OS  Od U Z _ d r s ݎ~h)$hKnhky(|hh:hwnHo(tH hJy(|hKnh/ny(|hqR hWhqR hWhWh/nnHo(tHh+ h{nfy(|hKnhz=ry(|hz=rnHo(tH hky(|hKnhy(|hhKnhhy(|hxCnHo(tH- !!!!""""""## #!###5$>$?$$$$$$$$$$ʼʼʳxqlelel hTno(y(| hhy(| hCo(y(| hho(y(|hw'ph-hnHo(tHy(|hhnHtHhnHtHhnHtHhh-hnHtHhw'pnHo(tH h*y(|h?+ hy(|hKnhhy(|hKnhhnHo(tHy(|h hky(| h,y(|hKnhky(|h{nfhk6y(|&$$l%%(h)++;./2/i/R08222M34I6K6 $xa$gdrbxgdd6^xxgdBx` x`gdBxgd/T x`gd9 x`gdLgdh$$$$$$$$*%,%%%%%%%%%%% & &&(&-&.&3&:&K&S&V&W&&&&&'''''''''''''''''''''(()(øhTIh(nOh(nOnHtHh(nOnHo(tHh<nHo(tHhhnHtHhnHo(tHhahanHtHhanHo(tHhChLh9o(y(|hLh9y(|hhhho(y(|;)(,(.(7(8(9(:(R(S(b(c(l(m(u(v(((((((((((())q)r)s)t))))****\*`*n*o**********++2+3+O+U+W+\+c+Ƚ h?hZh h?h?h?ho9nHo(tHh\vUht!ht!nHtHht!nHo(tHh+h6oh6onHtHh6onHo(tHhChLh9o(y(|hLh9y(|nHo(tHh>h>nHo(tHhChaxnHo(tHhaxhaxnHtHh=h=nHtHhghF)unHo(tHy(|h=nHo(tH h=h=hgh9y(| hshs h h h6+h96y(|hJW.hBh9y(|hA h9y(| h9y(|hm$nHo(tHy(| hm$y(|!1112222222222233333z3333334*494<4C4D4E4444444444595һvrnhz&h9hnHo(tHy(|h/Nh~anHo(tHh{hnHo(tHhq6hd6^hFgy(|hFghJW.h*jhd6^h~ay(| hCy(| hv&>o(y(|hv&> h]Xo(y(| h9o(y(| h9o(y(|h-C'hBh9y(|h>nHo(tHh,0th>nHtH)95:5M5{555555555555506@6H6I6J6K6O6P6Q6R6`6p6q6t6|66666666̸|mc_h!uhx[hx[o(y(|hn3h~a5nHo(tHy(|hn3h{5nHo(tHhn3h{5o(hn3h~a5y(|hn3h}5nHo(tHy(|hn3h~a5o(y(|hn3h/N5o(jNh RUo(hd6^h~ao(y(|hd6^hJW.y(| h,y(|h{hJW.hd6^h~ay(| h>|h>| h0D1y(|%K6669;<=l>?:B8V8l8p8ѰɩљѰљ} hkE2y(|hp!nHo(tHhd6^h~ao(y(|hwhd6^h&]y(|hXh&]o( hXh&]h!unHtHhH 7nHo(tHhH 7nHtHhwnHtHh&]nHo(tH h,y(|hd6^hH 7y(|hH 7h!uhd6^h~ay(|hd6^h!uy(|/p8x8|8}88888888888999$969_9b9f9i9j9q99999999999W:\:e:i:q::::::::::::; ;;;;/;4;8;X;Z;j;k;l;m;n;;;===hfIhD#qnHo(tHy(|hP mnHo(tHy(|hd6^h+y(|hd6^h~ao(y(|hH 7 h%y(| h1y(|h*jhd6^h~ay(| hkE2y(|h{F==u>>>>>>>????)?2?@@@@@(@)@B@Q@n@x@@@AAAAAAAA$B7B8B9B:BŽzqhfInHo(tHhh8nHtHhyPYnHtHh*jnHtHh nHo(tHhyPYhfI h^y(|h, nHo(tHhohd6^hoy(|hd6^h]G(y(|h]G(nHo(tHhd6^h~anHo(tHy(| h%y(|h@h]G(nHtH hd6^h]G(hd6^h~ay(|hH 7(:B;B]]]]]]]]]]]]]^^^ ^^^^^^^^'^(^)^^^^^^^^^__(_*_._/_3_4_;_<_h_w___________hsnHo(tH hhh\hlBH5o(y(|h2vh\hlBH5y(|hBhf y(|h3PhlBHy(|h3PhlBHo(y(|hLnHtHF*^^*__`ia}aacddeIfcffijjLkekfkgkhkk$a$gdrb$a$gd[Hgd&vgd/ xgdxnYxxgd3P x`gd3P____``1`@`e`s````````````aa,aTaVaWa[aga|a}aaa>b?bEbFbGbHbIbPbQbRbSbTbUbabbbcbbbbǿ˵˭˥ӭ˝˝hsnHtHhm}nHo(tHh3PhLo(y(|h3PhLy(|hLnHtHhBhf4my(|hhlBH5y(|h3Ph2vy(|h2vh3PhlBHy(|h3PhlBHo(y(|h\hlBH5y(|h\hlBH5o(y(|h3PhlBHnHo(tHy(|3bbbbbbbbbcccc-cccccccccccccccccccccccccd ddddduh*hlBH5o(y(|hBhK*h h^nHtHhBhf4my(|h3PhLy(|h3PhLo(y(|h3PhlBHo(y(|hLnHtHh3PhlBHy(|hhlBH5y(|h3PhlBHnHtHy(|hj ynHtHhUrnHo(tHhm}nHo(tHhsnHtH+dddddddddddddddee e e e eeeeeeeeeeeeeeeeeeeдЈȀrjbjhLnHtHhBhf4my(|h3PhlBHnHo(tHy(|hsnHtHh mnHtH!jhN1hN1EHUnHtH$jJE hN1CJUVnHo(tHjhN1UnHtHhN1nHtHhg~nHtHhN1nHo(tHh3PhlBHy(|h3PhlBHo(y(|h*hlBH5o(y(|h*hlBH5y(|&eeeefHfIf[fbfcf}ffffffffffffgggg g!gEgOggggggggggggghhhh!h#hLhPhahxh~hhhøhbDQh ` h jy(|hU2 htXy(| h5y(| hp~y(| h|0y(|h m h/ y(|hCnHo(tHy(| hxty(|hlBHhHJy(|hHJhlBHhlBHy(|hr` hCy(|h3PhlBHo(y(|hf4mnHtHh3PhlBHy(|6hhhhhhhhhhhhiii&i*i0icidiiiiiiiiiiiijj)j*jCjHjbjpjjjjj½챽ңۣxjҶE h~sCJPJUVh~sjh~sUhyh mnHtHh mhh hp~y(| h y(| hZ+<y(| h7Gh7G h&vy(| h/ y(|h/ nHo(tHy(|hH+1nHo(tHh[HhhnHo(tHhbDQ h-6y(| h jy(| hbDQy(|*jjjjKkLkMkNkakbkckdkgkhkkkkkkkkkkkkkkkkkkkꠗޏxkgcgcgcgcg_Wh&vhhy(|h(h[Hh ;jh-C'h}2UEHU$j@E h}2UCJUVnHo(tHh-C'jh-C'Uh RnHo(tHhhj<h~sh~sEHUjE h~sCJPJUVh~sjhrbUmHnHuho{QhYNnHo(tHho{QhhmHsHjh~sUjӎh~sh~sEHUk l lllll#l$l,l4l9l:lXl]lblllllllllllllll mmmmFmJmcmgmkmռxpffbhEQhxnYho(y(|ho^nHtHhZ+nHtHhnHo(tHhxnYhy(|hxnYhPnHtHy(|hPnHtHhPnHo(tHhPhxnYhPo(y(|hxnYhPy(|hxnYhy(|hnHo(tHhxnYhCy(|h-;nHo(tHy(|hyFZhCo(y(| hCy(|hO %hO %y(|&k l,lXllmn3o5o_o`oopp$q,q$$If`a$gd8l" $$a$gdlgdMQ(gd $xa$gdrbxgd9 xgdxgdPxgdxnYxgdyFZxkmsmtmmmmmmmmmmmmmSnYnZnnnnnnnnnno o,o1o2o3o4o5o9o:oؽؽصجyoehn3h5o(hn3h/N5o(jhPUo(hChEQhWQhxf9h_|nHo(tHhxnYhCy(|h/Nh9 hCUnHo(tHhK*nHtHhxnYh9 o(y(|h9nHo(tHh9 nHtHh9 nHo(tHhnHo(tHhxnYho(y(|hxnYhy(|h{$:o;o=oWo\o]o^o_o`ooooooooooooooo ppp%p2p4p@pRp\pqp}ppppppp´wmiih mhxnYh mo(y(|h4^)nHo(tHhxnYh4^)y(|hxnYhCo(y(|hxnYhCy(|hC hCy(| hxiy(|hMQ(hMQ(nHo(tHhn3h5nHo(tHhn3hv)5nHo(tHhn3hxf95nHtHhn3h5nHtHhn3h5o(hn3h_|5nHo(tH'pppppppppppp$q%q,q0q1q3q4qBqKqLqyqzqqqqqqqqrr6rʼʮ{qh`hXvnHtHhSnHo(tHh{hC\y(|h{ho)Z\o(h{hC\nHo(tHy(|h{h\nHo(tHy(|h{hC\o(y(|h{h8\nHo(tHhlhl5nHo(tHhlhC5o(y(|hxnYhnHtHy(|h mhxnYhCy(|hxnYhCo(y(| h{o(hEQ!,q1q4q;qBqCq4kd$$IfTl\ l' t0644 l / aT$$If`a$gd/l"CqHqNqTqZq[qfqlq@kd$$IfTl\ l' t0644 laT$$If`a$gd/l"lqrqxqyqzqr6r@999xgdxnYkdP$$IfTl\ l' t0644 laT$$If`a$gd/l"6rQrqrrrrrs s$$If`a$gd/l" $$a$gd-MxgdxnY6rQr\r^rqrwr~rrrrrrrrrrr0s:sfs{s|sssssssssssDt^tht{ttttttu u uuɿ觟~~uqh2h2nHo(tHh mh hxiy(|h5=nHo(tHy(| h14(y(|hhyFZhCy(|hyFZhCo(y(|h-Mhe5nHo(tHh-MhC5y(|h-MhC5o(y(|hxnYh2nHtHy(|hxnYhCy(|hxnYhCo(y(|hohCmHo(sHy(|, sssssmSSS$$If`a$gd/l"kd$$IfTlFR'PRR t06    44 laTss+s-s/smSSS$$If`a$gd/l"kd$$IfTlFR'PRR t06    44 laT/s0s:sssuuzvmf[f[PP x`gd8$ x`gdyFZxgdyFZkdI$$IfTlFR'PRR t06    44 laTu!u"u,u?uHuuuuuuuuuuuuuuuuvvv@vFvHvxvzv~vvvvvvvww www2w4w6w:w{N{ƻضζζήخΙukhdfhaT5y(|hdfhdf5nHo(tHhdfh-5o(y(|hdfhC5o(y(|h nHo(tHy(|hMhyo(y(|hMhyy(| hyy(|hynHo(tHy(|h hyy(|h{hyo(y(|h{hyy(|h{h o(y(|h{h \y(|h{h \o(y(|(zzzzzzI////$$If`a$gd l"kdR$$IfTlrL3$"' t044 laTzzzzzz/kd;$$IfTlrL3$"' t044 laT$$If`a$gd l"z{{{-kd2$$IfTl4rL3$"' t044 laT$$If`a$gd l"{{{{{%{$$If`a$gdV{l"%{&{'{/{^{G<33 $$a$gddf x`gd2kd#$$IfTl4rL3$"' t044 laTN{R{]{^{_{`{q{t{u{{{{{{{{{{{{{{{{{{{{{{{{{{{{{{{{{{E|ĺĬĬumh{hCy(|h/nHo(tHy(|hClnHo(tHy(| hCly(|hMhClo(y(|hMhCly(|h{hClo(y(|h{hClnHo(tHy(|h{hClo(y(|h{hCly(|h{hCo(y(|h{hC\o(y(|h{hC\y(|hdfhC5o(y(|hdfhC5y(|*^{_{`{j{q{t{$$If`a$gdil"t{u{{{{{{I////$$If`a$gdil"kd$$IfTlrL3$"' t044 laT{{{{{{/kd$$IfTlrL3$"' t044 laT$$If`a$gdil"{{{{{{-kd$$IfTl4rL3$"' t044 laT$$If`a$gdil"{{{{{-kd$$IfTl4rL3$"' t044 laT$$If`a$gdil"{{ }f~n~~~~~$x$If`a$gd!U l" $$a$gd_t x`gdb= x`gd{E|X|||||}} }<}=}?}I}J}K}N}O}S}T}U}`}g}h}p}q}x}y}}}}}}}}}}}}}G~e~f~l~m~ҾɮҔҮҔҔҔᔋqh_thb=5nHo(tHy(|h_thb=5o(y(|hb=nHo(tHhmnHtHy(|hid1nHo(tHh`InHtHh@enHtHhnHtHhnHo(tHy(|hnHtHy(|hb=nHo(tHy(|hh{hb=y(| h y(|h{hCy(|h{hCo(y(|+m~n~w~~~~~~~~~~~~~~~~~  &'*+019;<BCDTU_Ǽxsjxsxh@enHo(tH hxiy(|h{hCy(|hAnHo(tHhS nHtHhB1nHtHy(|h&hb=y(|hb=nHtHy(|hid1nHtHhnHtHy(|hb=nHo(tHy(|h_thb=5nHtHy(|h_th5nHtHy(|h_thb=5nHo(tHy(|h_th_t5nHo(tH)~~~~~mQQQ$x$If`a$gd!U l"kd$$IfTlF 'NV t06    44 laT~~~~~lPPP$x$If`a$gd!U l"kdz$$IfTl4F 'NV t06    44 laT~~~lPPP$x$If`a$gd!U l"kdB$$IfTl4F 'NV t06    44 laT1:lPPP$x$If`a$gd!U l"kd $$IfTl4F 'NV t06    44 laT:;<&FGolaaVVQVVgdb x`gd x`gd{kd$$IfTl4F 'NV t06    44 laT_m"$259AKPY^bր '+0PQWb$%&HUh hGnHtHh}?nHtHh-[nHtHhnHtHhyB:nHtHhCnHo(tHhLnHtHhBhf4my(|hh hxiy(|h{hCy(|h{hCo(y(|>&.26FGPQZ^r{̃̓ԃՃփ׃ &*ظ|w|r|m h\5 hPb5 h!F5 hb5hn3hb5nHo(tHhn3hb5y(|hn3hb5hhjjhhjhhjUhyB:nHtHhp(nHtHhPGnHtHhiEgnHtHh?nHtHhGnHtHh<nHtHh-[nHtHh nHtHh#nHtH**@DEFGJSTW„ЄԄۄ+,01YZ[_`abfhixyǷh}?nHtHy(|h}?nHo(tHh}?nHtHh RnHo(tHh RnHtHy(|h RnHtHh RnHo(tHy(| h Ry(|h{h Ry(|hbhbnHo(tHhn3hb5nHtHhn3hb5nHo(tH h\5 h!F53…Džօ@BuY^nopquvwyǾytlhn3hP5 hy(|j hYNUo(hn3hv)5nHo(tHhn3h#85hn3h-C'5o(h}?5nHtHhn3h5y(|hn3h/N5hnHtHy(|jh\LUnHo(tHhyB:h RnHo(tHy(|h RnHtHh RnHo(tHy(|h RnHtHy(|'oqه4<$$If`a$gdl" $$a$gduq x`gdExgdw x`gd{gdgdMQ( $x`a$gdrbه܇$&;dez{|~ÈɈ͈ԈՈވʻʤҏʊʆʂ}uplh hoy(|h{hoy(| h'my(|hho h'yy(|hCnHo(tHy(|h{h'o(y(|h'h{hdy(|hi:nHo(tHy(|hdnHo(tHy(|ho)Zh{hCy(|h{hCo(y(|hwhCy(|h{hCnHtHy(|hn3h5nHo(tHy(|*#%/156>N\hkmnvx~#)*,0\`it|Պؿؿ鶮ضضضئئh 6nHtHhlnHtHh nHtHh]nHtHh`nHtHhx7nHtHh'ynHtHy(|hnHtHhB7nHtHhsUnHtHy(|hnHtHhdnHtHy(|h'mnHtHy(|h{hCy(| h'my(|2Պ֊׊ފev '34:;<Fbch}ϿqggYNh{haH\o(y(|huqhaH5nHtHy(|huqhaH5y(|huqhuq5nHo(tHhuqhaH5nHo(tHy(|huqhaH5o(y(|h{haHnHo(tHy(|hnHtHhtb5nHtHhEnHtHhnHtHh$oDnHtHhu*nHtHhnHtHh`nHtHh'ynHtHy(|h 6nHtHhjnHtHZ@@@@$$If`a$gdl"kdMc$$IfTl\ d'     t0644 laTČɌʌόЌՌ֌׌܌ ŽŽŽͪ͝tmih hlLhlLh{h7wnHtHh7wnHo(tHh;*nHo(tHy(|h]Eh;*o(y(|h h;*y(| h;*o(h?{h;*y(|h{h;*5o(y(|h{h;*y(|h]Eh;*y(|h{h;*o(y(|haHnHo(tHy(|h haHy(|h{haHo(y(|h{haH5o(y(|)ČʌЌ֌Z@@@@$$If`a$gdl"kd#d$$IfTl\ d'     t0644 laT֌׌܌Z@@@@$$If`a$gdl"kdd$$IfTl\ d'    t0644 laTZ@@@@$$If`a$gdl"kde$$IfTl\ d'    t0644 laT Z@@@@$$If`a$gdl"kdf$$IfTl\ d'    t0644 laT"#$/JKPekuv˼ˤ˼vndn]dn]dSh{h,o(y(| haHo(y(|hhaHo(y(|h haHy(|h{haHo(y(|h{haH5o(y(|h{haH\o(y(|huqhaH5nHtHy(|huqhaH5y(|huqhuq5nHo(tHhuqhaH5nHo(tHy(|huqhaH5o(y(|h{h7w5o( hh7w hhh hlLhlL hlLh7w$v|ZQQ999$$Ifa$gdl" $$a$gduqkdQg$$IfTl\ d'    t0644 laTB,$$Ifa$gdl"kdh$$IfTl\ d'     t0644 laT$$Ifa$gdl"D,$$Ifa$gdl"kdh$$IfTl\ d'     t0644 laT$$Ifa$gdl"ōƍǍ̍эҍ׍؍ݍލߍ ĹԬzzvzle h%:o(y(|h{h7w5o(h. h.h7w h.h. h{h7wh7whhaHo(y(| haHo(y(|hhaHy(|h haHy(| ho(h{haH5o(y(|h{haHy(|h^haHy(|h{haHo(y(|h{h,5o( h^h,h]Eh,y(|h{h,o(&ƍǍ̍ҍ؍Bkdi$$IfTl\ d'    t0644 laT$$Ifa$gdl"؍ލߍB,$$Ifa$gdl"kdj$$IfTl\ d'    t0644 laT$$Ifa$gdl"D,$$Ifa$gdl"kdUk$$IfTl\ d'    t0644 laT$$Ifa$gdl" B='gd2=kdl$$IfTl\ d'    t0644 laT$$Ifa$gdl"!$&'RTnǎ̎Ўюێ㾶㪟xmxbThuh;5nHo(tHhuhH5o(y(|huh Y5o(y(|huhH5y(|huhu5nHo(tHhuhd5nHo(tHy(|huhC5o(y(|h{hpnHtHy(|h{hdy(|hdnHo(tHy(|h{hCo(y(|ho)Zhjh{hjy(|h{hCy(|hwhCo(y(|h%:nHo(tHy(|ێ܎ݎ$$If`a$gd1l" $$a$gdu x`gd{ێ 3EFGQRdeflnoy}ЏՏ֏ڏ˸{pf^hO(h`y(|h{hp\y(|h~dhH5o(y(|h~dh Y5o(y(|h~dhH5y(|h~dh~d5nHo(tHh~dhd5nHo(tHy(|h~dhC5o(y(|h1nHo(tHy(|hMh`y(|h{h`o(y(|h`h`o(y(| h`y(|h'ph`y(|h{hpo(y(|h{h1\y(|#I//$$If`a$gd1l"kdl$$IfTlrL3$"' t044 laT 3kdm$$IfTlrL3$"' t044 laT$$Ifa$gd)cl" !39?E$$Ifa$gd)cl"$$If`a$gd1l"EFGRG--$$If`a$gd1l"kdn$$IfTl4rL3$"' t044 laTRX^de1kdo$$IfTl4rL3$"' t044 laT$$Ifa$gd)cl"efo$$If`a$gd J]l" $$a$gd~d x`gdZŏЏI//$$If`a$gd J]l"kdp$$IfTlrL3$"' t044 laTЏ֏܏3kdq$$IfTlrL3$"' t044 laT$$Ifa$gd)cl"%0ڏۏ܏1239;<FZ`egopؿwlbXPh9Zh`y(|h{hpo(y(|h{hp\y(|hphp5\y(|hWthH5o(y(|hWth Y5o(y(|hWthH5y(|hWthWt5nHo(tHhWthd5nHo(tHy(|hWthC5o(y(|hinHo(tHhMh`y(|hO(h`y(|h{h`o(y(|h`h`y(| h`y(|h`h`o(y(| h`o(y(| $$Ifa$gd)cl"%0$$If`a$gd J]l"G--$$If`a$gd J]l"kdyr$$IfTl4rL3$"' t044 laT%+121kdjs$$IfTl4rL3$"' t044 laT$$Ifa$gd)cl"%023<opq{$$If`a$gd J]l" $$a$gdWt x`gdZI//$$If`a$gd J]l"kdMt$$IfTlrK$"' t044 laT3kd6u$$IfTlrK$"' t044 laT$$Ifa$gd)cl"ǐِڐې%&9ACWXilnotøvkaWOH h`o(y(|hBh`y(|h{hpo(y(|h{hp\y(|hphp5\y(|hmh Y5nHo(tHy(|hmh Y5o(y(|hmhm5nHo(tHhmhd5nHo(tHy(|hmhd5o(y(|hmhC5o(y(|hinHo(tHy(|hMh`y(|h`h`o(y(|h9Zh`y(|h{h`o(y(|h`h`y(| h`y(|ǐ͐Ӑِ$$Ifa$gd)cl"$$If`a$gd J]l"ِڐېG--$$If`a$gd J]l"kd-v$$IfTl4rK$"' t044 laT1kdw$$IfTl4rK$"' t044 laT$$Ifa$gd)cl"ABCMTW$$If`a$gd J]l" $$a$gdm x`gdZWX^iI//$$If`a$gd J]l"kdx$$IfTlrK$"' t044 laTiou{|3kdx$$IfTlrK$"' t044 laT$$Ifa$gd)cl"tuxz{|‘Ñđɑʑˑԑۑ'.9@AB֒ؾhwhCo(y(|hwhCy(|hCCJaJy(|h+h'h h ay(| hMy(|h{hCy(|h{hCo(y(|hMh`y(|hBh`y(|h{h`o(y(|h`h`y(| h`o(y(| h`y(|h`h`o(y(|1|$$Ifa$gd)cl"$$If`a$gd J]l"G--$$If`a$gd J]l"kdy$$IfTl4rK$"' t044 laTđʑˑ1kdz$$IfTl4rK$"' t044 laT$$Ifa$gd)cl"ˑ$$If`a$gdtll" $$a$gd#{R x`gd&Zxgdr-TxgdO x`gdg,T $GINQ^nuͼͳސteWh#{Rh#{R5nHo(tHh#{Rhd5nHo(tHy(|h#{RhV{5o(y(|h{hnpy(|hnpnHo(tH hnpo(y(|hr-TnHtHy(|h{h&Zo(y(|h0GNnHo(tHhhnHo(tHhWnHo(tHh&jho)Z h ay(|hh{hy(|h{h&Zy(| h&Zy(|hwhCo(y(|h%:nHo(tHy(|Óēԓ#$%BGHMNSTUV`afglmrstu·zzzrrhhh]hV{nHo(tHy(|h5{h o(y(|hMh y(|h5{h y(|h{h o(y(|h`h y(| h~nFh h9Zh y(|h{hV{o(y(|h{hV{\y(|hphV{5\y(|h#{RhV{5nHo(tHy(|h#{Rh5o(y(|h#{Rh5nHo(tHy(|h#{RhV{5o(y(|h#{RhHp5y(|$I////$$If`a$gdtll"kd{$$IfTlrL3$"' t044 laT$%/0B/kd|$$IfTlrL3$"' t044 laT$$If`a$gdtll"BHNTUV-kd}$$IfTl4rL3$"' t044 laT$$If`a$gdtll"Vagmst-kd~$$IfTl4rL3$"' t044 laT$$If`a$gdtll"tu~Ĕ˔Δ$$If`a$gdtll" $$a$gdg $x`a$gdtlu{|}~Δϔ!"#$./45:;@A¸xpppxhh^^^h5{h$*o(y(|hMh$*y(|h5{h$*y(|h{h$*o(y(|h`h$*y(| hj0h$*hBh$*y(|h{hV{o(y(|h{hV{\y(|hphV{5\y(|hghV{5y(|hghHp5y(|hghg5nHo(tHhghd5nHo(tHy(|hghd5o(y(|hghV{5o(y(|$ΔϔՔI////$$If`a$gdtll"kdi$$IfTlrL3$"' t044 laT/kdR$$IfTlrL3$"' t044 laT$$If`a$gdtll""#$-kdI$$IfTl4rL3$"' t044 laT$$If`a$gdtll"$/5;AB-kd:$$IfTl4rL3$"' t044 laT$$If`a$gdtll"ABCIJKLVYZjxٕͿܧ͜ym_yWymWMh{h3o(y(|h{h3y(|h{h3nHo(tHy(|h{h3nHo(tH h{h3h9Zh3y(|h{h5yo(y(|h{h5y\y(|hph5y5\y(|hKh5y5y(|hKhK5nHo(tHhKh5y5nHo(tHhKh5y5nHo(tHy(|hKh5y5o(y(|hf2ECJmHnHo(tHuh{h$*o(y(|BCL$$If`a$gdtll" $$a$gdK $x`a$gdtlI////$$If`a$gdtll"kd$$IfTlrL3$"' t044 laTƕǕٕ/kd$$IfTlrL3$"' t044 laT$$If`a$gdtll"ٕޕߕ 3GOPRfgЫzzzoe[h{h5yo(y(|h{h5y\y(|hph5y5\y(|ho+h5y5y(|ho+ho+5nHo(tHho+h5y5nHo(tHho+h5y5o(y(|h5ynHo(tHy(|h5{h3o(y(|h9Zh3y(|hMh3y(|h{h3o(y(|h5{h3nHo(tHy(|h3nHo(tHh5{h3y(| hngh3 ٕߕ-kd$$IfTl4rL3$"' t044 laT$$If`a$gdtll" -kd$$IfTl4rL3$"' t044 laT$$If`a$gdtll" PQR\cf$$If`a$gdtll" $$a$gdo+ $x`a$gdtlfgmx~I////$$If`a$gdtll"kdц$$IfTlrL3$"' t044 laTgx}~Ɩǖ̖͖ҖӖٖؖږۖ׽׽׽ޮxnnch0GN5nHo(tHhmIh5y5y(|hmIhmI5nHo(tHhmIh5y5nHo(tHhmIh5y5nHo(tHy(|hmIh5y5o(y(|h5yCJmHnHo(tHuh5{h9Xo(y(|hMh9Xy(|h5{h9Xy(| hDh9Xh{h9Xo(y(|h{h9Xy(| h{h9XhBh9Xy(|&/kd$$IfTlrL3$"' t044 laT$$If`a$gdtll"-kd$$IfTl4rL3$"' t044 laT$$If`a$gdtll"ǖ͖Ӗٖږ-kd$$IfTl4rL3$"' t044 laT$$If`a$gdtll"ږۖ#*-$$If`a$gdtll" $$a$gdmI $x`a$gdtl-.?DEJKPQRotuz{ɗݗɿѷѷѷدѥѥѥsiiho|h5y5y(|ho|ho|5nHo(tHho|h5y5nHo(tHho|h5y5o(y(|h5ynHo(tHy(|h5{h9Xo(y(|hMh9Xy(|h5{h9Xy(|h{h9Xo(y(|h`h9Xy(| hj-h9Xh9Zh9Xy(|h{h5yo(y(|h{h5y\y(|hph5y5\y(|$-.4?EKI////$$If`a$gdtll"kd$$IfTlrL3$"' t044 laTKQR\]o/kdn$$IfTlrL3$"' t044 laT$$If`a$gdtll"ou{-kde$$IfTl4rL3$"' t044 laT$$If`a$gdtll"-kdV$$IfTl4rL3$"' t044 laT$$If`a$gdtll"$$If`a$gdtll" $$a$gdo| $x`a$gdtlݗ <ABGHMNOPZ[`afglmnopru{ĽĽě|rjrj|rjh{h ay(|h{h ao(y(| h ay(|hnpCJmHnHo(tHuy(|h5{h?So(y(|hMh?Sy(|h5{h?Sy(|h{h?So(y(|h`h?Sy(| hU h?ShBh?Sy(|h{h5yo(y(|h{h5y\y(|hph5y5\y(|ho|h5y5y(|h0GN5nHo(tH( I////$$If`a$gdtll"kd9$$IfTlrL3$"' t044 laT)*</kd"$$IfTlrL3$"' t044 laT$$If`a$gdtll"<BHNOP-kd$$IfTl4rL3$"' t044 laT$$If`a$gdtll"P[agmn-kd $$IfTl4rL3$"' t044 laT$$If`a$gdtll"noQn>_CĠw x`gd[xgdCgdMQ(gdWf $ a$gdrb & Fgd Rgd Rgd_Zgdy x`gd/ xgdw x`gdw˜ØǘȘ̘ӘԘ՘ (,6:dhPQ]^ghnorĮzrhnHtH hJ y(| hX y(|hyhB hNy(| h_Zy(|hwhCy(|h%:nHo(tHy(| h%:y(|"hahaCJaJnHo(tHy(|h+hah anHo(tHhh^X/nHtHhnHo(tHh{h ay(| h ay(| hy(|hh'*@GT}ϛ$iozɜʜ>BCDEF[\^_cdefgȿݻݲݝݔ h}?y(| h Ry(|h Rh_ZnHo(tHh <nHo(tHh}?h}?nHtHh}?nHo(tHh/Nh [nHo(tHhRhBhhnHo(tHhyhX h Ihkh hJ y(|hXJnHo(tH4Ui"*7<àĠoź|mh`UMh-hCy(|hCnHo(tHy(|hChCy(| hy(|hn3h_Z5nHo(tHy(|hn3hv)5nHo(tHh}?5nHtHhn3h}?5nHo(tHhn3hWf5y(|hn3h/N5 h ;h_Zjh ;hYNUh Rh RnHo(tHy(|h_Zh Ry(|h RnHtHh RnHo(tHhh R6y(|h R h Ry(|opVWXns|ãȣԣأ lʤˤФѤҤפߤۥdfgijz{ !+/03568AISh 9hDhDnHtHhALnHtHh9]nHo(tH h~[h~[h~[nHo(tHy(|h+Hh+HnHtHhCnHo(tHh-h`y(|hT hShShShh+h-hCy(|h`8j{Ѩܨ} C1Ef u_9TW15'gdNfxgdCxgdv8xgd~[ x`gd[SUghjlsШѨۨܨݨXq{|} ˽qfqb[bRbK h]vohIwho/&nHo(tH hhhNfhNfhoh!mHsHhoh!mHnHo(sHtHhohNfmHsHhohUfmHsHhUfhUfmHsHhUfhNfmHsHhUfhNfmHnHo(sHtHhChCnHo(tHy(| hCy(|h~[h~[nHo(tHy(|h$5hv8h 9 h~[h~[ h~[h`hDh`  ݪުߪ$4@ACOR\]ad 9<ÿwwlahohUfmHsHhohNfmHsHhsnHo(tH hshNfhoh0mH sH hoh0mH nHo(sH tHhohUfmH sH hohNfmH sH  ho/&hNfho/&ho/&nHo(tHhUfh*MnHo(tHhNfnHo(tH hyhNfhhhhnHo(tHhNf hIwo(&<KLPS$&+.0?BPQUXx46;BDij߮UW»֛֐֛֛և~z~s h hNfh9ssh9ssnHo(tHh`nHo(tH hQKhNfhQKhQKnHo(tH h`hNfh1h1h1nHo(tH h1hNfhUfh(nHo(tH h(hNfhNfhoh(mHsHhoh(mHnHo(sHtHhohNfmHsH)W\ceүUXcdhk ()-0Rgڻ򢞢򅁅zqhl'nHo(tH ha}_hNfha}_ha}_nHo(tHhmnHo(tH hmhNfhGbhGbnHo(tHh5nHo(tH h5h h5h nHo(tH h5hC?򦢦򛏈vrvvh80h80nHo(tHhnHo(tH hh hh nHo(tH hhNfh),^h),^nHo(tHhUf h,mhUf h,mhNfh9N h9Nh9NhnHo(tH hhNfh,mnHo(tHh h nHo(tHhNfh4*nHo(tH+ۺܺݺ Sʻ˻ϻһTV[bdeǾ赱薋uehoh<:mHnHo(sHtHhohNfmHsHhohUfmHsHhUfhUfmHsHhUfhNfmHsHh-^|nHo(tH h-^|hNfh({h({nHo(tHh/nHo(tHh'nHo(tH h'hNfh2h2nHo(tHhUfhNfh}h}nHo(tH h80hNf%'e1Ƚؾپھ/0235689;<EFGHIabgdh&`#$$a$gd`gdNf׼ "'.0:=KLPST[ȽڽݽQɾξվ׾ؾپļᦛpi``Xh.nHtHh;nHo(tH h;hNfhoh,mmHnHo(sHtHhoh;mHsHhoh;mHnHo(sHtHhohUfmHsHhohNfmHsH h~W hNfh{h-hmXhmXnHo(tHhUfh<:nHo(tH h<:hNfhNfhoh<:mHnHo(sHtHhoh<:mHsH"پھ۾ܾ/0123456789:;<=CDEGHIJ]^_`abcdeؽر皍h`njYhhrbEHUjkE hCJPJUVjhU hy(| h0Jy(|jh0JUy(| h.jy(|jh.jUy(|h.jhahnHtHhjh0JUhCnHo(tHy(|"bcde$a$gd`2&P:pNd/ =!8"8#$% 3 0&P/ =!8"8#$% P 5 0&P:pNd/ =!8"8#$% / 0&P/ =!8"8#$% `!z/lP^^-R4Q p/~ Hxڥ?hQ]sFiJ CI )\--ELGՊSh8ġ I38:MP'(6s`=vP|blTYi: uLʽTKS'/5 P >^2iJjPktזuIݟ,Zģ5KD<֟[hM0V^i!-Z~N C_OE0 6HϪNl8-;wvl9oy.P^㜹C tx޸xkgxk7vC< _[Eu`4+!u:U'Υ-oY|\h[ԩ:hx]e2Ub`oҘwW=<姥qw TZ*骹mzW"37/j-䭌IQQ*~ }L> Shv.Uw\Q&3N 񂉕˪ 8˹g!}Dn~'yu؂Myi@ )GFkA8ws0:ecvcL$$If !vh5#v:V l54a ODd'%#L  C A"`b6O|I&>OASO n O|I&>OASPNG  IHDR-OsRGB pHYs.>NIDATx^}WGziJ$D$"W{m+.XEw_H BR."D6m?" UW/[%ZHKVRޤM\s̜9on|=gy3<3g_D"@!5m%D"@F9 D ຜ &D"@ D"08Hr6"@c"@ `"@ "@ C?.g D9 D ຜ &D"@ D"08Ҥ'NV'K.'|rt_cڤapΝs[ LӉ:Yw.=lfGjp^kmszHҿ Q յPv1fN]0a:Ɉj!.8ŮM%KeKN X='''u7n:r'!v)`gܴiYqLj~=r`X0i>"sa2Ӡ^8 [fWu5fѣq-SWT ̐j?+fvA^U+2mu[s)y|TL~_կ~c%)k_~3UoPW駟BsNWC֭[3Hvwͼ)"čO2`FOLӠ>f4߸qOO>Aͫ~С|"3x~,7?E:ҁ2fTf$]FF<I(WHq·j4<^Jbܺ?mJy@}*2RY6V'9(";&QL*"WiPQiP͐l):Kx=(e[#3Ӡ_)ڤՑd6W#>CvQ '|aLLϟ?ң. sQ]kj[ЧD mLa `f+s(^'?.=$eE̲ALȂk~nu)YelߊQ ǽ->7d~loF϶U:qO@Dy*&2`uvͱRȑ#{?.8lٲŋOOue-[֮][ݣ]loAz翚s*A9zhbjWmʦ2۷oʕp]t5C[f6[[P>%KʦhLzn.|+z~޽R c'&D̙Ӻg-+.~tF5|_-~fъY1mݺyϞ=gΜ:1O^6YCʮXQI-_{6{&ҿIp6*`jd lÄY6FXihap2eIhpsocɘj/ #udΊ)Y9&loٺLEª;YսƮ29E$ /hIhBm D7j|cW07>S֭wcU͓y=&e^YpfV*Yw0A $K&YOŻqŋUjT_oe@7dLBԘ1D.>j+©;۷9y !855y近̬ZOх؇8;ކ8Ijƍҋ7op8~zad*"gU/h5kTMe._\' [[6zMK"@|"iu`_uifTTE -5ᨅ;_ s&+V,Z'YLOk1 9Ű m1Esԕ/w{?:r)E4Pm&&9("5WLw_1(cH=&YLy̤FKkjĂVSdUd Pm&6m/p@k&"1v@&;ʶ1 ~)e2KSkHXm޼Ye: (E=pH$wkvrF ڿ_:P?A2L S4S,gmՒ,{'"c ^]?x dGջį4dfRH=DB .Iq*Џ%ib )ѿaC_ݵx%}qy_|f}駱DGi􏊺Vj}:h_Y ZƮ]$_>|8Ss=V Yq$3*AE(잟%<(/;'V01Sqʴړߙcծ9+[cGas1@mVe2!dR$>[DPgcSE|Pq TԞ<\V{Y[%'FTlyU #s摔`S o͢Մvxmm G!Dv<]]/M"hPODS?ƍk7(U҇PQ;)ն̳$a*Vyɓh3 l0K#ZrMWƩ.2KM^F k[F&V"pwkȢc6SOї2+$3Ns~-qχ~'<8@? ˚̖28*Xjdzd-@1Uv$pE#viigl"@ ~H6k!D"#{6"@|,L D FH1m&D" 'X"@@c5L DN0 D k"@@'Hca"@ 1"@h3 DN;D"@bDgk"Ny4 Sanl< s"@@CfN\}WNaˎ~;Wudspj$DE|Qs2́,.7+' XqWg_]ax_7Y"`(_}YN T6gf%)Mo2toP(|^޴. bY"@=4a?R ^ !Ge~GsOwj (_^*C7𻆶H_?>zu0EڸLʀ?蟹/1HA@I~ãsm10ǡ '/4g:q8ӿWOFt{@>|@dwm س,NG 4$ #!NWqXEs-c8@Ai&9hS>R>fΡ!CBbЯ&0_ND~'/2un\ R4@z["r0肢8?.lٰaQzt#>6B%3uj?{+G܊ y+ÏZC踿zH8g>P`å]SS#EL-?0P%ooѿ Xz4HD &5e#|"T_(?_y w@%D.r?p2  5?ėn*"~A $3#`P?=*U@ o5pEshn9fѿZp)m@.vp_٭ÌZ,>r?: P6 6mh׎޹g=dW`?=cF @ ڇQZae>ߖ2XeԱ%{u ~ày7}t|8Xu`Xņ_u4:c 1hhQmj ^迗vwSIF:C N@,o/|(}ӿA䁏CU@,o/[Ӳ2?F;QaKu [J0o4B=OsM!F +Fbs2D>n+2̨$>S60o1bMIa_ѭ}C"g߮SL>OKڡj9T<HV,d=5 29k$ wN0g^`_;,9fj!@$6PfEדCwmzX]5L(CbD^s::N8nu: mh@vlo}x"2ǎͬ[21Ȉgecp'0wh p@[1[n-XﵶAX$^&<=6]-x_]*BI@+^FpyvB^7WUGGH|t# 2`?j_2`}Ȝu7~p19{ FVư>;2 (Z8j!@ 3sN?%^򍗭@}.&C0k1Q߹ 00EU>FNy"sr bG@oZA˱4mfs@sM.a}2LK׭=}wo7Zq??u䅛Cx͂5χ˫̷%p<;7Vd̵CD,q2@!rOM0:Czˍ-.B6Z1)Dmkw2IZFMoZ<ׯ߼ynYeZbp^%(j_A_˪ڿޅwlK+dN_زN-[uQOXhw~lW\.Zyj_1&`}^}:G[ZZ;o 0*/?bߢkcţ- H2h7W0z^.A_v `oCe.ȶ&C;omkc߮m vW +utvпZFKUG"-p߸ʁU+g=6|mCܵ<) :0Uģ]@V6aM]W{GVU2Sw|[[SBzÒWu[k(P+k7my[h|||v|lZCY7DCf_3`<8=[wIǞx$_p;/\]Vo2:m޽ p|Vu3م亸-i hMYi=`!t ;[Y 0Oo؊GM]ON>- 1;;S񞹮T173Fޔe?sΏuRL&.-鿋y!ԕ?~ mڊGUyvNYt^}}A/smp=afocLZXh|YǙ~~}'`wLD[Hi=I@y?F tq ֣}^}Ώ+V̙3a0g|o7Ј ?|뇳e'`K0ħP 袓e#І[MC+8cg߅ {#m({K.q%Àѿ1n޺^kExXp~_Mx毜 w|֏.V0_5 пA+Czcz|˖-SQ~>FH){qx\anֆZY_ԑ!@E ]hL.&Wѿ> g4Z3GvwøokV  []ɰ1a;goǐXˢ'''Պ2>CCY_f^!77g^cS暓ͿUi솩0|YA`n0RcF s= ,[ :h{蟷0"V8G'si44 3IO!4whZuId-,[{i4!Ta"[FzPjD3.BԜqõ-1tU:ɚN$PÌ,)2ޣ+Kl26[IhǺ+,/ʜMߏѿ6.jO/u->vpѿoK;jj8z,I5]=ɣ``oWP͐ߛz3M-9\o:{'{_Ǘѿ# }^'Noh{P`z;*XO=NSm?F>O]GY[юuW֗wv4Z#`Bߺz,Czc[<t-Z4LITh,bҿE0MU17E\k}ƣ8v55ϝ;"?.2@'_B;1fhm(N@ѿa?ZIV@n$ٹ'lţ*Q'wzˠ HV`͚5@ H-jmSzv}E [U.gFrUElw{3}fxcƐKيGUi,yӣ7=>M%={v…!c6[hY=;EVغ,Enݼ?|P}[PJ' \;y?{H&?lݢc,ԭCBuEb*CIwCѿ;lK53tUڊGˢQlq{]_8/^Ç_̮]z-Y-{n{H=G²])R鴊TAW6< $5"Q[hY-[löںum`믿 v([׃3s_Ʋe׫^!H==yVc" ȧ2;9%z$ Qh{@YD 4=:7Oülɿ)j.U@>S.DD?Qm Fr^֠v֖7G5Q-Hƣ>;hDۚo/4݋Tgv/dom{sXGhhdS%m+ocw .LlR'b!xgmhuuu !7İ]?(/k)p PT!x >. %g;No7X~I8Y`3h~ں8wޢy;h76Ԭh7n{sмwM6kה|3u^w۷oG]QX{n5_R GѿPKՁ{6"fA'X10Q;w | ~aΝ}8 2_l<JDFa q߆ٳg8r*Wj-c0kN}Ä%V+Qǡرc[n^dɲew灛!ҥKEF/^X?xoY dSN )HNLLٳG<+*Wjck)v@%L~>+՞G^|^fM#Z B,e"+oЮ KxT船e]:[cg6ܩ2z?|8333;;UJW>ʝv-Ilذz'g `+{24,yԲb AMؽh>Љ{c k.D-[kTB&x{yӇʕ+.]‡uɾ<-_h݅ E}O<LEb[inNjz\v{ V?Yz 4mݺәu ~;j{3)mؙ/G1DU(qs=2 jwm^{54R{y .RUƹτMCBb<{sw'O׭G.G-K?oơC*Cx_~nkY9Lc]w !hF3 ]G9τ8Sg@do޼P6L{5eOEG0U$}D"VA=*G>hDFa?Q<'vMfN)T>i~ҿ;ly@0D'ڎ}0Ϩ%$QL}B]ҿ Tsp0D'ځEKG}] eѝۓ}y8ybl/>86;*N?Y K-7ނѳ :4߽%ӥaϖuY\V'y"P;My;Y*4ѿsB3{ܡϟgE 33cxr!Mv_*[*_ Ť.Kk.Pt DO|.@s gB&D W}.~:|o5~Z ~0 E?|~.]˒m!I=F>EZȈ"g=F?sZeFF[ fgFk xIwĎo݇F <:q$wǐE[8qmm_tIl"0clm\MAX7 S%Y};Z_u,(yOwuޓ'o!y# Ҕ"Z//zø,Zptq-M^ncFn:Lׯ_W=)^)\(!s[n`jjĐd~7B.`#7oo_oCT]\jbm+lP!o葲89ovF@zGV<"{EO>/Yd hիJY{Gƅ+ ㆲz/JZGPѿZ(ꏽ5(@Nah,BZVX-[ż\}B 7n&O<wCxG.Hk5dv^ :!G/q]ɨڮfzM$7Eq 9e+UlY ;vѿ|GvE8:J[h=z=x@}n Po;xKNSw (:AFp#k_#DO ,[x4?{bG_kwEyO`ߟx9Ɍ۶Nl/kݽ#'dn%ȞA~䟺?_AD 3Z*RLyfB5SFՆ3@+l9/8?ӿi&_Xπdd'z Cpy?0l5wW% F;y ~F{N x+=$1Eɍ a=:un;+Qك-[; ɧz&? NkS# _otvBWGyw&Fkw D^;praCc;H()rUJԄ}HOLlhdpEs/\ku}_ɀq ޱ=_ |F$ux+Xwϯ.{C|$AlmGKBs;6s=#^wKKڑ4p[hźBX,]~Ǐl_pἳe a0hu~9u_Y4 qTBU?}Upq=Ⱦ%WT}s֮]iV`ztޡ}*p0,B|o0Z*j?]K0S!VP<%SZʠ :=OttUwQ)PIoS#DWBt{_ṁy;}oc(!pDok5=o+Raٻwcu?rrnhm^ m}J3+8ID=jȣɵyL$a6 c] FcO0^X'q g܃'| xq|z`C^e^n9eah@m TCC)lNuE6 שSq3z(\wTƷ@J~7@ļy՘073U2DW[P/5x4AmW9y9ξƒ3~@E.љ=SΪ?uX?O@t50(dIY NC kƨQXu.\?u"<WA{zoWV?zg8:0haI T_F遉|'( +n.;qo2[=Ewav:h/ўU+q?,_Ӿd&$Ϗ2@-WvM~e\BoD;% N_aLQa%3E gH-%~VjY{6GR]0~hDG_J\+2)krd@rșOX:VR{k@z&gD& "s兎Bv~Lۯx 񉶻> @2 z -~E|lQƠR״,`ӭ2P&c+ mmX#@.64>z[ 쉝;]x^aqpNxtv|^y/Ν|׬Y<^2QqlnOU>@P[^,oOg3o > V# j'M7̊{a_{hO4jIdj#F#Qlţ /_|ywgf#4GB0d (o^vM %YCm?{A+ mm^#X% `?yF FNCcSZx!z]Z!{ɒ%˖-mOa=ztҕ+W={Va F~ӦMȢ{m#o>_ {E3gN$+W`yeBy RPՋPo&G+H BYSNe|[hadh5K.[o[񨊼3f?z<-Gݻwom',Ν;qEXSzϞ=p #G,lXb!R(;9tv,I,_\|YXh^s>od.x֭[[oB`-"fHu=u .xTE: =裐?Flmҳ8x-1|(H{-Fl+-G4 F_xwr0Z>d/WPb>aQ6#(?@FAAD/~ x; VD?6xǰuEZ-{xMjC̹M[EQ=]`xYzde?1N.\m׮C# կw'~Ѓ]1R0sw"((饗@N)+e=.]v;X:[X<Ca:|pr|Ç-_uu/b @h`К[c0FY{__]?Eh|؅#KeE__`X$03<= 3tWpcE3j:=mz]V  }'_Yl >*g/(azkOͿ_V*C_t>;@dZhn•vGz&.L8X+^%,;> +_]Xo g~ OЃ$rإc1br>|ax ?ϡ%"o}G?Ԯ?C!7$( ͵h 2-b.[(NkaFe;Q/VQM^&ꓔ)@{;gיB <2v<`^]_5ʑ.ݝX~7wYG Id2?=a} Lyatd[k,LIshFTxەۡRDG|zkKq5"Mbƫ*`@ȍ0@h*KX] |ײ `e;jK օ6mle{Krw!cG}me7dh-  pѠ`qdLPh3ਗcT5b Bw.Mg^kdh-x<"A x4!wq t%Y%lZ<ꨙ m&@o2"9G}0Ѧh :-s׵_Jޑ-j>Td9/jSf@@>[7h$@Dc)x4!!" wni`^x]{AxTYY0X >:4QlݺBffL`p1ڧ"vC@Qy闾ЮE Z =7* ߱c腥I3ggͬ2Ċ'o %@F#SB{ l<''_pA4@wԆ*#yD-^(kU9(ӫ@l+@ jZ(O>6y`{ڵ??شi޽{sмo߾u֭_>R>(>$CEC(b /T;t9]3E_2MC+#8r+ɻޡw܉ w/^|ϝ;+(@ϋʕ@AG_Ukj̷]Bhp::z]~,g"mo֭~ñcyo&+K?it$gggqk.ƛʙYe@Y lh=xofתEV_-[=mn8~XV dp=n+?s 駟M?W\Z p ӫw\Wy?8zs=&A0׿#G@ݻwK-",2 {쁇LSOna]mѿJH a#=H(ݻ?hTV +Ww&Zhga)ohň@΢_<|͛7g\\z 'ۿ pe"Hv,vۭ zG$Ī iC3:5ljz^lvZ+W Ǖ۷/^<="o%ǷַO'}E^!G7b/^a;[Rxb_-+w+s~&CuTO% wwpيwߪ~(v.Bd{^W~+|8KccGS\܍5WP{k{G]wz&COO ]zi+GffL˺+6n܈}x;,j+`T?<p6njኔuWx>'COo *|ww؊WXn;}zrŲ_/<âO>>v `Ed?:;퍍NSj0=`al|k!+p>?3|mqc|y7oƊS =z+: zr[oy̒/D-(aNB CF]*ǀۮ<ϱo7ZJ+}Oxzo/_|ҥvիWjR6O`?t FsNUa2r<pA{Π|:j .ZŒ=/n~pDM+{ӟ>Y~8y$R|ľ-[HVQdR@X#Xd8>weҴA=dѱN?5Ďx'*:is<W"PvӋg~:eoh@WO,O/@%F]zhe>bG9`GP3d>@+Nص2Ld}Zq7/SSO=Xg]qOnt}܏N=k=o+~A?=c, T_<1k|@ѿ*2;*G?sh7B 0&ad/=߽!b@/+C׸ď?>}w`׌.\6@p/@ʝ5E+-,=s딯4w0?8ĻMQr8V@@{w5сVe3~y0lǹǏh~HI0/- +W( p?}@2H= o+c8:`O HGU{p&|ʼn@ڛZUVA?=$Ĉ={P?tt/QgԐr& 1JCh4:j- zq=ֆMbj4@?gw͋nb_i&zi he4&_po$AB2cLzGHX%Dz4PgF2}~$4`0 \qӿ 1mh."Pw16Z#q?8=0w /vIJZ?ڈ_fпN@I:TH"')$3"f-N+"TE!wޣ9l6e>#S\_` 6"=#l ?EL-O'C  S$~|; Hu{"A9Q?,WZ O`$&uJ N&Ao2Rč@~@4? 49@֏{D}9X.%o$E~@a"6pם@O{g@9YѿO&^Wo"D6P蟬 7+@@ ?' k{e rz~e/ ~f"N2Y'P'tq?IEY"@hW-H[X"@ PFdzoN*#D2+Jv;rH~fmD"@@&"@ ^ {"@@ DxEnVF D ?NBIENDB`GDd)f "R  S A"`bFIDATx^?vI]0~ 6 L `!Ј6f Fbvڥ0 P$&RRk[Ĩ$$J~_w~Ϝ9u|9sf>3533?~ @@W뚛 @G@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@x:W2:?O/}{^}}gO[Ggx;1>[s\0ꟿ뿞Z|_%??L3?Q>,ɦ͟NK#0%9;4*ɗ!%>w3<sc;N&4 d "784]sy2_??*?1>kD!mo{ /pJ)!\Ka~R}S1h5*''e(&ĞF-kE|gl_O, @V~~ncBb J\[[zf9g>CPQ(0 *\"qx*wDW^yeQ~_#agt!d{_Id50R=Οԧb7zyh^|ťr/r&G]FC\iڛ^#(pG'U4$aqވTbhZ=nP`n*up\&P\1KG;Bq]{8QL>"Nou0ml"ߴl/| ?C{!Q!b{7;6; @PFL>ߢƄ5B2bESĿc޾ Om#u@0Ocs}SLĥm/\Iy#d=o(e!ĤH]oۑU9uI@^bAC \E/ Y"‘ЫK$qS`ѿ#ے2)O?K6cWǡ> CX]%Q  S91ȕ6ގb}=MJE<>Q'vcDCo}iU\N,f]iȼ"N$Y$x{0bF K&0OL%[6 @- %aDz[mPz$ ύmQ:;.G/ "iNbsuRSngyx!#:iSD#.5d#-$gn铣. HS?Т}j,s1Pe6~;k̋e v Tr{Y @8"zNbiLVbȩ8o [NEق"i!Ej}]m&  @@-!ˆQ \E]YYȬe$[nI٘HV'k@,rSToF;9bBɧO=T#hJ6=B! @ #!ˆꅏϾ/"Yz&Dz417-Yr.aW46'u9A~'2/P<󅑒mFd0澋=w^ňqiLf,dKcۈUx Wkʇ"4F.x^ڌyv] xAC[fxooדɍmM+DV*8~siL񌵀cػxo߻Fw9RqlD xE2Nױ~|ww֞$@6Gzޢl˨E(Kx%FI!H|Iy-nCbx%J,cgRw󝗗S.1dBGaDe,{<#H>!R_GxQ!k~g~&v3{C}J wy >YRF}c>D)ը)!T<{6(dC @DGqX1=oR-ԃb9WCL(o&6 &2R ކd5LilKӶP$swD.èĘøϿ* @DQz)$6$ 1.XؠPr YӳDS9b<aDöcZI 3ű1o*$,mtSSRA㍆atղ"@m 0 ^-R_͜KSk5NZV|"zUDc~j:1Cly:gSbshӍޯ6-”xrh װǛF, 0Zv:]E$e{.lDmb4QNpJbXG>2H;kJѝuHs0i٠%VԳ3I y_~uMx{yNx/w h KobP_5cl+ Q5˵4縍"V[!38*bk=Ɯ42b"rVKʁPjƦUuA6,2a,I$w\ܑ5#NyƯN4q%TG)I @tW0Bjؼ:Ց]<7-(0'pcؼ:`>MqҾFr85#McjQ۬_ A"M_9]P` [FlL>+3ފ-\[1VC(Uܧށ @azRxHlߢTȬĩer]$Md) ≥FM뫽*b*(^:6q5WʕE֨ZxoK&.՜}EVZIm[ qhx }#rѾo7=]f'H@>W[ݥרk k女AeR[6*`Ϥ4QFQ8W1:jjrm=ܐaqr2nQYRQFVgFod2DY-RxH@o 7 /v8K؊1-pX|Jtr?FTsG/:S::M`#ˆMY͖6:aC,udK9w-69:D޸7olJOKa$4&]K8c|f zb0(CO~U<\1m0d#٦pg6x7Wg2X{KlҐ ^.}#iT! F#kd+)  @ )pͻUȸ~oߛyyp 7=.l09MKqLl'gw>3b_bO@rRq}(B>ޡ`ǓGE18Ym(d챈2KҮ:g?Ք @ȝ&t7t1=^f2zcR_m-]p4+잒Y-I܌6ݴP3 =<9fvD{ $o6V`vN?2'@@F|9>eꃳVsޑa /lHWiOկ~g /$ =WS=%ZqoQ$Kƴ<K)ƜFi~/ jku묊I@8Dd^X*?{3OVG2mފvQu vۺa0Feof8>M;¸G3;37=d8pMG=}dЎx~LV~и)Bi]ɂ-ݨ2<>Il|9']1xr`TգS#ed2Q#{V˩{pq(6}dHa\ # 3?-gٴŧ~ꓲUfm//ra|Iؾ=3bRSS"pU+lDnoˆو|3afOA+sLaR)+97-cbvsܦSGxKA܆':LbTh|DRoر>dvcGc?V"G0Q-}=gDGR* @ a YIW믿8$.%{2d]b!Y)a/SH'fKiak_Z,Fͥ XɟG?//$[*NQȦ$ЏXv\E3E0h.;R_s;qG?QMcM[3Ɛ>2J kVnT2Cセ1t/Ϗ'U|iKT7D񖻸6n<ss"(^ci @NaİˍU !HaÉ6m[n떋SodS+kDe̦{y_5߯m ɰF:41gd~"Ԉ'>x Ud9yd)  p0)oTᨒ[rjd5!Mc[aXN5V'fPDnW% +;6a k+5a!#;b=O=T<d$ @VEQ&bmx%#ܶ6D[B˿K~ۉ#|T| ;vl <&m K$JtŎ!ڈwl (gcE5FoF E~+ B\[oxsͦQ]؇lf&_d]eKsb4a$p~ksH,&olء+ˆoޞx/[m|ԥmg5H1C>/~qv#:gf!,,/r""~MwQ 8V%ΏKЇ2i c`wZ]T' ]"t(]ብwa3Q0V#M7G K$QKJJݑo "^-P#^44 GNAMhs6 1ˆ 1­ǕM14UuzWWodyغ-/WoǩFO^oL#1G$C<+p#jx4<%Q lI^xaSGxx/Kc$9Ϫ~=쾅-$;:I\?aD<;^2⌷ÿg7Dg^~6=B)%W^f1<[ܗg qp-e!rTopy#qLԏٚIT9~Gyu֛CƜ(/XgWdH'-9D33IW %#@@[hl!+FLضI%ehW Dn_7M`L>oLlΉj5XfG:'Èi_@+y7ŠhDsQ"0mvnݝbeaӖKM)N'-%|EX]*0J&1sVqK @[aԌF6}#axi٥hkQJen%?wuM %Qiԑ%!"$j{@8AWIc[{c>((Cs=w-! :. lZ(}:˦4=]܏ce7E1M +ca%"K0K_%: F)[%u˾@'_櫤]'Qb:.x% 7{I2' c+SG%s\DdBm 73!ȴo>QbŸag#bX4EzXˆqT[K&ߑZ9=#xZ45boEyxM<06EG.V\B.|jaWȴuI]8<*njU,dnKN.4J\hh2$J轡1;n*Hk`#R 8t}#D6E-]՜i;bh|#kb2 ؍ő9P-qj"2k]84-M5S򓟌1j$ @@^Ʀ'4'onlJdߴ$cP5K-1n**>.㖟|&J1n\{GIb#Za}&cHD0b-W_] u6T]ˏ 9tޭ:kcR,طb=S!߿ݪK-#JTcë)灎N:оwzoEԵaQ?xc6HhR1Qh @a7#  @"r$@D)G>"#2 @[aZ<EO`?Te)7Ŭ2 @I><  @`d$(հJEI pb01q  @aZ8  @D֢C 8B+(5ƭ @#0 @nM@qk-> @"#2 @[aZ8  @D֢C 8B+(5ƭ @#0 @nM@qk-> @"#2 @[xuU(F}B+;n߉ի!@#cEp; )2vI-֠JC6 hd&kf4GN<[Ë#WP @8F*|ko\OTO,M ERŘz,*8cTktTlyPq%p9$Ma eS=U%L'F. eИ}MОyJ0*mҼQݽ=zԫɇCD2 5l0[CI6[l4J͡}F]rioSHLjGK2 G%GīdhU\݉l[J@mGZhucX%NQl:(quNvv2ce!ZuXB^\=K<9^ޝHt=BN3ͿMN6lG5.`azÇwr&@[$cјogf>0rF)=fG1[rOC!ϼz @~J2;_&롽YS4ss|!Θi8.lN,ui:-,ma!^/CIct%z2=ߣ=s*$H`uJ[G irŲ@Dլ'a~}vד!ܤWIn}mQ- @@^rM?#ߖR @`Y@;H-ǩ @M'06Gb @X%92 @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @#+% a䭤$@"t @@^@ 0R @yFJJ @ + JIGDy+)  @ǏgN};N\Gx[85#fR @Wx& p#YD @ ؇q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'aN@q&S` p I  @Dk2&@'8Vѣ)" &pcNCys$@[a/4 @4?~|NË^0Ktf E8WIG@ \y{}dHAsG P @;  @0t @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ x`HXa .%_^ْy}˳́IEN#DaDX0j;ͼW!CSF/#t=TO*d}: 2S۹ @.hEc4NQRJ}x)(OLX,- 9,QԫG4c  @Vkal-(͛ Çw  @KDGo)Xʮ!d]"LJi C`qñ pWIF;1K_Rцz#:l=:h.dw[Iz.p9df ye_n}Rkr3dH.8DqINzMTC  @Jr{jD0 @Ur$@a @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/ o*G @@ @DMH0 @  @>@a7#  @"r$@D _@T @C @ 0ʑa @@FS9 @" }/sMK/꫉E/x([j?ϱq'w# ( @ #p#S8i @ ؇q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl phơG @ID'm8&@ayT@q҆Sl ph?b{v,H 1˜Ոk˝/%N:P}9z|-]88Ұ={Q\2goFdwtP)8J1Oa:b=>{T~fS'eXVgٮcq!2izeh|/t_nW#Q[Qϯ2K/~p2MRGHJ*pczQ|4HO!K_sobJK[ٚ~鼙!<=J[ 3}gk.I?bɴNIԍ*;>v,oIexticGpҽ]7rw&g ~^%kOO/{azQl=QȴzΒ<(X7ShluI_u`1u왽owr44íKPYMv[0l9$f}Nu3>UL9~"hÇ{}[ʇ21pWH پ2=K9@e%44xY &`WyLk b#?z4w7Kߥi5Xڨ/$2C$!-@Lfei7Ѹ6YӹðjQ3Œ^UZ` 8!"GաtKyN!'3,v6I|`ԎobE?BѝFZm5ֱ].o.E&\]hnGo^_RBɔַ hP2v2ULv-VG$`&涻&KhvQva_d׬4 (gfƛCtЛor<ҥt,d] d0‹ 4N!p坞2{vIḍJ wWL]*u.u`Q4K4lO۱qi1Ŵ-f[gګ 2%og.Φ_D5Z*ZG~Ç/+X! ܰ]ڮ.oKag_ K0ik\hFí0DJ1mYM_s)4i`1|0nxXR*I}^lB/ђAK|HO;(l3iF_K%jS:-l2Ϝtu3mVx5j;flW? Y-4A,K0bUX%p9$t7s4 PCN2!ٴ_.]u,[ ̎uf9\=miʧ$)Z*R&Ѵ;6VFn7Y*~ Qݪ0\38~;$:]uR[e^0.JE`ftd8wDEeof༣4?"ɓ]\s]ԧoo't']T#@ VIAnJM5 @0AnJ@qSͩ2 @ "4b @aTs 8  @D7՜*C" 8HC()M5 @0AnJ@qSͩ2 @ "4b @aTs 8  @D7՜*C" 8HC()M5 @0AnJ@qSͩ2 @ "4b @aTs 8  @D7՜*C" 8HC()M5 @0AnJ@qSͩ2 @ "4b @aTs 8  @D7՜*C" 8HC()M5 @0AnJ@qSͩ2 @ "4b @aTs 8  @D7՜*C" H9 IENDB`g/Dd/cL  C A"`b.OYQ7mҸ).J n.OYQ7mҸ)PNG  IHDRfsRGB pHYs.>IDATx^_KZ5{嘲:٠H8p6ss/TUcjv}=[5W_yꭧzꍫWWϮ I|ʡy/1.XF^w+IY`x(Ohx]ݿ;ܔ޳^^Gi1SdYrl'>|T[Rz3̥c8ϮXuSǑr]_29* / a^MEo|SeVF٥Wɘg K2ne\[;̹'8]2n(=ķ!g)yCӇ^.ݱ62ś%=o]INXR~~Weq`;{O*Ux ޗbʼnzaWW]"I Ga^)|CI2jE}Q2-ۊxd.;GRVί$0=c. s kcs|.}s.Ǫb(ѣy~PRm/} 6Yxxs[#؆?X])#vvՖ?n \Pf>=~͐V)*ġ,/g*2yF~Iȝ*&Ӻ3ur0EԲ#42fWGXv~lܿ\ph8"M`7=/ڥԎOw]n! meGn.!i`+w颌n'g({j9Cxmjh#mي)}B ( C;pi6Ssq18w'隢}ZH$g|x3?!흻 [}a;Fՙ_Ž G Wk_GWGC/F>yn@̅'(Ej3 b2]23ߙCJ˘w8_R9)L dOr&b^6\xpH :PiW -!/D|ȧ%]4{Kl[幑^}6cRo/=+P nJ=$J ]3y/[[->_ ]0ؼg &)5+,쁝;=5U!ݣ~J2a{M .rz'^&6;鎠;[R^)O,CWoZCѾo1_'>A~0_zj6YU;Yd_1j3w< Py v^[e~٣.mxD+{d{}R9Y:泫t㱦X):,V[m>:y/pT寢=yQvtg_Pߑ vVLf\sh9ҍ,g#|hgA3I׬ Y&D~m ꜬO3~縪3Ny52_˫+f}}?Ғ}Ŵ "|_g 3C7ԅ|{"(ԝ 7E.mA{hWb_3Ls<0Hv7,v2WL]no f1\3t~YPaTt@FUNyCgA[BPB~yH1+m,h!ۆzcŰ,oڷQo8ݔ\vzAe(?TxQ""^ѐUO`T%Ϯ)κ>_+O;)LZ7+%j5i "t#;6Lel,Q_̴>RVطXeXSIփ[,,pC]q)N`g='#Ċso^"R˦7cfH!fw߾'J庤lނ{Jv&< |dg:߹EȾÀ(8݅hW}DᷲQ[ť{cئgyx/̴dw޹o黛gPn:3|7WZ2|؟%R~f8hE9 0:>"Kx1h;^Dg")p`?d8V(Mu%e3]Q\Բ vˇ3|ufſ/̓ѳ2V[ wxAoG^Ki[b莵㴎Nk䫞 ohz/ kd'`1'29LĠ6'<ލ_ 9B~U%VB]~t_\y%}u-901}#<\.;So {2/zܠnMZ_hHe;Kkᬑ)X"R>e!bgrDUƽ>mbv¯#qRDu Y4Tm2E|\~xD% /_^5ܱ%w syu,%y"zOcK F˺Gn;̳$xZernŴ#w^ H!̾7kp9 1gl+ \s/8.Zud2]BKO "O!GYBe;K2$3!>L?9˷n}JaE.-c>Zd}q?^,RLrFG{Ss@/#?9\#͚H9 'w^'s0c 2̼BX5UDd? n4]VхFŸ2j^%xg_5/xc.7ҸOGi:R>mRT]:}c{w.;_cnh)us+5^8KEoM,YH"ý>cDGnњb}GSӇ垜} Y2m8>WT#(}n늣ȖDaqu|NxjG|4'}^^58w=w犩~o94qOG8_AH cڄ_䊩(\ӜJ|0,)0p 2dḨߩm#5'"DXa*[ 8$^}G?䝻y} zS.dy_0G^fqNu|;wpV ?e+uHKY.yl֥;dFTMܩU*ζmq.ut{vrCVzaNo|jD)-vnv#OԶMN?xHn=)A) }h*fw< ۇukZ9˯VAzZbx*ͮ}mCjn_/y6z jmoJ ujlcټ3IAWn<2}DwxwnM<79$=]mPoQG|dQSF8cGݵq,""`YTM쮼AT%R0(9 ʫOdt:Y8\ w M!*5ޕ"c{-1a@mXc&٨>C9%=c<,'J:U/mxY*ÉC.֭O)樤_XR[CJktu##=p4x7ÇҙG[_,Ǻ!'3mXGjk- YKM#RS9«b:ht*!(rW<2="ek:;lms0_Lac7"~/}y_;q:m+篁ۄ3̲JfdPWidjy%`m}#M澋69wiC͵r ekZAț?Kw?f \͵bքs<äp.S"&[=%NV4jV I%2UVwY,$u)3WRSa*$du3R9S,p<_F7_'#c#ީlq9 t7yF󕻈ߨP-kl!sz݃>8܊)*֐iHݤmw 0!m$y}q]s Tޏ?6z sDDA;z+3u犳R=t -eَZ[.Ev'+jkǧ!6LR#N|uEJrW26솉Ñb02dLeq2p϶Eg| ?P3Rdπ膔J3b{k[ۉ6&mnbBeRH-HUj!Gؐl"X,:W ;S|^9E;S|Hllo?o1<ɾqTLETFeժ&{R%6Hj5$jѻSSR8Ke-Ps4Xu Z˖>~,i`[0,u¦gm3wI!ADA#;TD,!1Tm)g/SMg#o3O@MRw V>DА+fGʻ_}ן9;̩<ގ+/ΆNQJ Ϟbö pdGy}rIh 2昮00rD,TNqZԊ0eX Q Iȯ69־6xs2">,y9mrW], }GͲp$fhfەwmQե%#*]gڱe̯24Q3TY[%:vM|Olqc^ArLA?A^Ŋxgno)gji %X%uP~tPuZP2n2 kOYx.g (ztiۀDu?]VͬE}/(c~6*:"G7L q>-Ļ M ,EF':.qRC6"V-G6ʱ=QsLHu%RgmU\3-:h1K郘Oj*0vs34\'uGͪΎ9GWCfK PٽΓJ Ix6Th(c% ƃg穡;6*}eKp;-g0 3LS%]{3C<T<6-Dl?)Euqݞ P*Vj_y3VUAfYV5|՗|tYʟ&XiZ:Ge8,:<&)Wz0_k[e l#G3 O7!$۪Ub8z~rw.҆V͆c1*d]Tm#=[yAL53WFx[S1D r~.jX`6foهҢwlix9P rFxiA!eP†ͱd *nA {Ǭ FX[6[4#'a638GIhx9ɐkMf-KPӲ0ImlxZn|I\3D4nC/r}?l1>Ȝ6A<&wmfݍ|Unnfȿ+sJq No [s{]_Ϗ62Db.:;e/*QY.T@qPm2i\onP6vǟo`~=NnsdsgN2b l{h4-J_M~_N!ȭ(z3W h䍼h HNGUȺ)Z)9<|+)\\s*=gp"͂^W.#ڱ ^l0h(5\7UF]8}̒R[1{a,wN*/$ <2:r^>Kz\DF-""8Y|vRYA2hS`*:>w- 3*z_ a QG/_Ok}Rt,+N9Gi:/8O*|s- qx5w"XRFd/.tQa Q^RL7pX3h -Z5ӽJQ͕鈎S癵N Ωk ʨ4Ȩdq6h4FU˲ke;\yc#-C/3*"2SYS\9r6˲րʛi'rO_ ZbrN)L5M:|Q4o)B?6HeSbcvuλ6$̑O/f'QS8 ^^I"/T%kQKlǫYb`XZO#=%/mvf~R%,c`cˡ.'4ZgxmƐrM+bOG5_[cRv=XRŁ/ų7Q˓íi 6p V(ނq`gګ!muJD֍7ˍRJO**Wi^̯77.&ͽv>u C˭f$kOO/v]*Ƃ\JjCy<(>i;OMm Sʮfʿ#}r9j1Y)_,{AM( #[d;:0Ո!%*h\f ǘq/eYQ$( o 9 ~j=,^BZK%(ؖ"K +vF:RA9e}Rs+ES,W冦ǽ[v:x-qnٽdjIs]lKgumFWBVadX>VR.Lגv߭e*2_SёC&2!22 e)p,-r|7YLV/ S8S2VQWao1,wi僨ReT>[Z;f/酌Ϡxl dđ5œ˞qݕG;ȓfK>+\Ak則kE $8f:g##3|qЯ5C8;g/>f,s؛]Q 6Vz^- K6ߑJe 2OP!ZWݖ}N,㷞&9[{ԛ}Fl v/AЧv͜=pI0j1*;gE 6΅,F&_UQI*ʊ'iu[|p[[P,`玳Ǝsٺ[9\<ζjm suyeTrh=s˜ p#vma6HR)7,j:K5:<͡37X|uo*bR{7_.֚g$T7wYqMvܖb8xxPrCE.7LDYd,›{͜ضdܶ 81Ӗiܨ6.AZnuyb2ӧN w%=+bdfAӔsl`oFE6_^k^X颐#ۍ^*[liTi[tq.(}fcHF7k9!QRAeA)SK+-U"^h *K'9+%+#&}ovOess+mae~I/4 `f/94jn{:5{w_X{7gk4]--w!1R>Wm[qfFJcȨ.Mxnt,2ջ%"  PЮr~3V[lx_: x Z_ޅT[7TREC_ޣw{ZbXk^w㼤+t33iugH`ZEkOyƖ)&^FyOTsdBRJX5 t ;jrs`gG^/W sY?, ASSK hMl<!1 c@ms1kHeY #Fi?gkJmyÚ(F503 Ծ2O3 "t,=*|80Ximu0=7?0]n?R3.MHJ)*WP9KHR[E[i1k' ַ_fNQ*tؐ<t/ؤ| {f"<.#9e-vP۬ Gf + ư;\f;v*~aռutN׏*Dn*W#R)-Y3C&ύזPQ=*,6r?h\SZ`Eke~S>ԧh[E*l 㸔7w;>i[;OVCWCԜM1m};aSgH)Y!R;qZ˿^S8cԑ+S`GD%Y!$.l }ʨBⶳ2z}lGy4ܒMǐUA`Nݺ3~9Մh;vMKv6>{$'R+v~O)`3;2-w[KNVv-[zCj'_ETmj/yVЊ3Jd"%%N@k*US5C)05NUx,ʔB/f]{nT:S-JVfdk uFLX3ƿcP eJ6k'q[@ Ⱥ Ջ8.cm2+mQ_k&7GI" տ>ǣʎ[+O+MFS+l8"Fϫ3n\mwa k̯Ds>U<>w:^kXjt_dzS鼙>Y9J;̶ܗbYBͅ6Q[% R)slsRk{ OF52^Aٗ#F4ُ[(ϧmkU E>.V{3CC;ΔZ4a_FPa?,TdhRƒP[B )/a( Y].ėFgc^Pw?5t]˪ع+L6S?gY'\ZX#sF=_0C~VlmOc8-!F1uhf x^lt܇YJU岹BٙկZlF( rʜ|񲨨Dnzo6K|eޙ{.|n,_ 0]ҒPVMry%>C3Z| )85,,.s[3=󀠻 ]:`yHY"*f< /JjZanSxpr,WV\ֳG2KԘĊ-~)@|I69S @\0Pn &KUjʹ}AqCy%e,sƀmٙv9{nNTS5v N9{mYJR,OQGJ=Sg!Ha$CFЉ:kR@́f*D_CT,Y,%RFz1pGlyU Ṽ( 5Ua;S޹޸撴kB);hؓO}1mV>*`ayO쑝X{㈗=TsI#SI_u+mW>[M:oxj;Te{ŐeyW|>Tv+LձBm3Lb^]TQ!oMq@pEJFkdgvбӟlܥV޾{˥L!Xi+v`4)QIh&ӄv7.sX=o䭭&/Ir+|ɉ#]_.k%%4.֘ȑq1ֶ &Yˌdf0=XgjirusT͍ qws9K[]Ji MBˆ2{q)Z6PQEH!D:1c rB6y#%tp'"3 "Jľ2ҠmF=R­U.(>B<|;gN\yg@3W]ΫFs[.l[p<Ǩ15X| OWAmLnI *TΙ!@Jbstg O[S&{κ Pa+[ǕLFVj GAϙ68M)W۾]@ӱݞzaUmcMB搤j\q j nX1'e݃m%ˌh^ol*׳JFwyԊ .v2HO#8fKcU)n睝({pecY)%]^o᠍4856g2ܬ5'sԨnX1s+Ra`ex,W6Jp))m9qOK5%zv+iJ1!fc摛mM]b6g t5$ d;0ȯVy/+ގyL]*t =#lUuDhN.bi 栫Nb6L~h w1B8eXelfT;M<[GET-gp}+ y9, ّșMmԜ=`7A%ΈNpYѝC"^-1NPn* ,oxӣb< n=3c[qHUq %6g!:J%eqvt9IWl-9'?}p;3-G1K{P&% JqGbޯ-a=H4 T>l=O \X2\e ŝs#Ϣ}x_ Ǥ]-FSX,;NJ9T_y]My;wHMR;>}>h!A<𺤦8w+?\oRec-_jthF)0~J)gxXSW@Mt|BLJ:gy]oW~JYPeT.Ϯ8CP]zoi(^ ]C9[]e[2/јwܾd_~;ߎ2O|w奜]Rqۏ#rq2IqlY2hE/͹Tu"RڋX`:UԑM b1f i ~BE,KSv{]XS -:,@5[KƟR5 >;w`54"?m9aN\Шe^S|J"R[byyj~ F lw7MV)I%4TdTn9897nN'M-\7WͱugeVtS93kU"־0¥S&w1?}Fڕ}'Uw>:ǰ3(Hw-&!3T\IhVHTr5zY%:t/3j,1xki_Gc>y_(BuV+%Oi\ɳUFꄮ)p͋.+ ǖYtuS>QͅU}Įv?KGsָ|(ZJ6ts3chs){hcUH97䰵7ĝ5/oОx)Yb@ς*Vy.iʡiz~ӊsn;-L˰O8Yo-1 /dj?UL ַU˜T>hN9%bYcfbgg:V$mi%[fghyضZy%TgdC`ݒ]SHKz@EyˎUW[Tǭ]e:X5}6e@,?˘c պTBYq,F){kL̐FCpfGCp:qnAt% hD#Cjw/qL]]a1K~mg1,ScnͲ3efYډG$[x%"ZTJw՘4L.6˼`3bP5BCpqʋR$f3'Q5JԳ{#匇!lv2'0UX=`_9=ڟE<+c[ogIu@hdv|"<[viS/k2)n$>*r5;X X|4/8~TS mη %tɫm#7+Ps]JmQsUK߀S3+R; /wwVޙ"b8ϿHT}=|;VČwtC2 "SDQ'[X>WoFߏ"͊x3o\^y>rU7W]gqj uJMyɺ y=yJ!ET:-H& -wWo"?x"'!7}(Ɣb)Q@~rb FS>!M>KQgx6rĿ1xfmouTژ֝6zs?ZSTlQq@#CR9ݱ)*FvW6=7/iפ QA NS\/{vs&P额HX3n3QCkpf.i[q !EVdU$jQX ?&FΔ=]D!:tN39Su`q!,HjnIY—ۚU#btK/VwD;T0 ٩n{*kxF|nA}DԐL1?ҝ(h><.,#Go|}1UnKVLGf12j^rX_MϬ/9RU$CxlSJSo^GSnqB<ҕ # rL7T[q0Ћ?zywXS?kU3pgjW/F-"!_MUeDq\<ڛ~*R3RTyhy295NV-ȭ3x‡GDeeu*n)ՌbM됅FrC%-3躭d^gW/6I#l,QU83}D3дs==}R(8_gXare@x[ 7),*&AoUVs 1ζ^e%zܺYz//-2K3#9sOy/cg\)3V+̭pD#;0.J]FWIC27Yud q_(?5x`P j&Z\1x iMzj:jݬP\s Yq^PAAk; R(y6=L.'eS-c7JXjeƀ!x 4Ҡ^w,F2_⑏QL08l6+D]rm4oC|bݨurLVpï 2+*9BD1d0%e粎sǮ6:`J;K7w;sM"㫐Rr|8:doN~؋-Z gAIś'[2o!c &P)lHN9JoFί3oGvCm+;/ߛ#C$S82< ͧo}5T6#(Q7smx"ί#Le]}v,/H99kAe'S単#op:BEA\/쨎^}|tR|~%^SMQ5Eӷ:NCNlY&R{JMVzp!?Tvm,?BUW4xV$øR+-.A|.g"OdyF3㜎+q^N|.h1tK”tp1c0E˭<_LHQߋTy-ɖ%U{c;)qڍ/ME_*,2ItnUY̜F\u 2§/yb@VMZC^'LAi\5k+6 د;P w!.vBՠrCQiXX/˙"^ Jf 9k/IH!mAKM_FLluZUcYjpYtkJ:VA (ܽx>[{+hSK:ɏC1D?jGzWWI=%^xb Ghf1zLwi[ԕiT;b`Tb )pi8:n̓;@hCBC7H EiY5k[0q]FWbrZ͗rR)38". ӫ\Bax]L?I),HLcfU')?ƤFW99qUsa+_@EI`K!Mi6CReޔ6|(q#a"ެRT=rRP@]∃ ÚM9'YT$|C#r$jT `7RE9ދ/(1"cox #区h i*8m/ 29OC)N##ܻa@v"S0T1x0I'oZ C(PaS{MiVL e;2b:~"RqBj4ꏦjK u${O_}!JRʯJ&W H'I@0yI,#4K ŝ#dŷm`xd(cZ}#5GgOO[jIVLI/-<*nOQq@\0J/E-. ;-ſ~El-Uz (5yzCY#DM+_d?./9?i<}Gԑb7 頜DÞA>OM/x^粡OXqBV-RoY_!/{b0(R*I"{yp"j-S\h% ",4XHujYC𙥫aru ,giF/fu'ROQ,Cf>DRWG?wV"j KQs4´a)߇ܚ/j`v[&wOH`mHs> /?znL|ls9rɁ)e8z@("|,e!7ulVZ:聓AD Ixިdm_ {ny*k|vC--"6uCڻ2]% z: ;$Qqئ%F0h@ J!^CU&>0eƥaqF7U;vfx IG:ˣUjd'|싑z!+\' Psƒ#AM̫dS[gKX0E&^Z_\nJrpEVo?%NOpn^޾} V柡؆l>fN28hmΜ#]:]mFbC%vr7nznk2;x]e`f70b흹_KEVmIa E%%GIc1Y+|r?7.pEW=ӫgOþ<4/L~d#&u͵'֔7%GIf 9ɥJYؚo"Lѹ&Yf3rN~7%ʞ|mԛ H0iP:YƜv6L H+?i<_a ;ϰ?prNC:prKi;j'H1䗑qeR|=ilNY|V<*n9ʋG^eՁ%3Κɠx.VsP@A';rno'|MCf0I󍁴8`gEGIn7@(w'\~ unN% 7` )0h :"lԎ:}w)+[N7{z_CZ7߆5=#ܸ0ZVtvޙVSO=^s^V4{Lm}>T b9h}zm0Sx'T8Y gf7%^ 7N3]zCn kB2dhֵfY)="Ζ_vXw$ll{:s%>:D~Su)O;Jkj3#OKH/^嵿5' w><;q'[=Bh!1/ZA$X:vqWZdKɏ+A֐C:]T:FM[<#~ -ps˴~_-(9ená`>!sq~w e=@i4c%)䮹EUvV! ΃E:-͠vw`j|3<]^vT4G^3PYtM݄F"-m٥\sjOov$*A*Bʍ= ȍ+/6w.rpmᄼbUR]k nŴR)֍SHf()d4$1MFnlI? #78h ]jۀD1[˦a 'z4N]xgLFun˼R.Dófs?ʠP\YPe@2D-)qR0[dPa-΢.w$K)oe,]wVVKg3c xGqb[W;gꚶ;_qrBٶ aaؔQQ%Ս}ԙ VZ 7^pl㳳́b}s'BV/D\3 ,.18D%o-[X[Xt8:SJR?6J)Ƕ)q<)%F -kf2vcoBc@ؒxxax.Jgߙ*75%E5'-.Ye)y5Jjl/+ST;* դk;WxT9w: mt#eFm'_ܜd]2]rt ."2v {?>aI4)@F_Bwx/HYòظUJn*}蠃ˎ qVOȭ}h :4L[ |@"4)Ι+YY<Hg'j9ѺPٖLGvR%S!et,\4H)$z% 3P=$jV^vK9w-yH[遊;e " msVYa63]+7qvjUxЛ\R։\57XǤc=410IA (uE3H,e4%fRղџ?5*\H|z~2WbۤH 1' M]¬[>ᇷ_;ALKt!L9ȿ,./SMKYj6*b7jeQQim585 * 2b=<º9 nLd*hǗ/4#ɛ? eNWFT}R4sMJ+Rq%.TE@yKsnQU]Ab1?OTeިIc>.7tu ag6T`}QܨrS5CI\`{FRT \ic+*Bf66<  c ,6s_ AG &\E*=tQz$(xI-M/M=h=׭%4+wEm^#k>}Paj-L?7rk\X &ޖSzcww &4H)h89+ѺAf;lҿ[B.OhB-wvD 9_a3Ί˅ $⬸DF CTsǯt[7TGj;? 7 I 9&D IyW5N-y]KB4ߺ#CVirDc);G0y\;8npK $4sXbՑmR=2z.yG)p#ͷ<#7yl!zPI}z}R]_?ZIr&a14>e;|ގ[BZMt?AF<źӤ7J>ple: J*OūMsʚN5gEK%o4LYFl [?9vH+oFѻ:"Nޮ{RjDنP8J!%+՜eFZ.<-aMwu{Nԛ"$wrgs}Lo2s)% Z6Iz~]m9KnG" űJ狅 I72bPdUh/IZr7uiȕ2Ny*CD6N|5Sf]eXx6Jio 1fCV{#ef_1rIj pȗ4vbRag(m~@@"4tkL壗TW*ۃOhȎ:5t-"(l޳Zb[ 6P嶎n /ˮ>B U!EF04)k-+P=~R҆:/NоA2ų-ͲV78ɆYIsTj;w❗o!z\wz64Hm +"E$+44Za*0Y8!H/Ǥ͸9rڕJ qnLvɈo;ޗ:81U羵<.Zb%u#݃0ȥhqETy}lq;w٩0?~R%As򭈇W}\w/3gyϐ!v~YqܱN1K5_fp, ՜RK4@}aDhF~W=?3Q5"h,^H4j uK/0.ˍ?訸{  MDMݖSG9y7r,L)j#&p-Xr!?*If#/?ݔ1 "#Xlgh'Ӟ/4Ll[3&#Au(Шڭ=kxIfl3^ bK&'{n:d(Yp?I[8$d}Kc0«HpukʃHZG2ֻ-LMfn0OoEfq .Pe9t3Dۺax4!5w3̆s޿(&P*ߍ0bhǐfO.U7 "<7ڡ6yQ\Fx&=UǺ3|QՑm+H3UD*_չ"Ə?WpT` 3HE,CJ!zq1xNC?XucȤ)*uMKҏs kfҡ(B*zgiOrhy] XX޹Dq=kP@Ln0:)tXU=-ʈv]3'-5;faZt) xb^teEi%ұAY"-9dmwѮR93>YlO7̓q/b~B u[GUbYFYh@ŅnIj2ɍګ>9f㈗/SDu>:;Ѣ&qY|GO#:֩ΐi[? +?lUt3.`roRQ(_⅌;ݟӥyPn2vֲ_L[c_YuD,kxnˈT~*e|ݨyrE1y~cT.'eiM2f[OРKZju( N(WTyC\XK6f-_!= M1M,(է4>bˡ(6=Y= .P?/Le#{J &9dNT/])L=`; {-RI{ ]#<կnK¤z-|Ku >0o1\fm$~AfhZVJ KZ$bS 2Je=vѷ9$B凹R+Ϩ SC9kVvo'dYTVvtPQ)NSxVe~tEs)Y:?UljC6" ceH5{ mVSFa2JW*)&:CbnUU.?iZN!]4*,j!8j:p\.;HyQC%.ȒVe!lu܏ag~_3TJ 欍ؚEkt R~~A}H悎>+ qgql3|.34AovuWk6hv oM⦚@jʘօY`*ӧRr4B.1t݌vuFuzR|*=8Y=¹ 0px[޾-:B (!Yܙ͢X. ~S/7 (H\.E\-KXαa_BLA<'`d3YXf-ez,ȮkA|+^uk}pwk+Kgt#a:w+׬}GRj3pq8ȶbj;ƹR7R)RXmr.hhaW9"P^(V AB8)7_Qsm%P|?D7d%{oUe|⑫|9e=gsdEGNywpBgiVEL#>xUa E*J7bZhm6 uZQb'/óPE::Z6rCuħ>=+e;cQ$veA2[":i$~ BcWU#CfWEY2NJ$"ʳ7*e39Ejycu4WbA}_7줂^R^=A5wYNf(?.Y*R9>۰׶+߭\X^аHs agVj ?jKWD^nflDF\KDÞ0d Jd0};ZUSƆ&V֊pN50:=GI${\|ҝ#ev(Y%\}%R4<E=rD5TcUsΑE5GKRs"Q#βaicyNy,/:Ln߾{Y0ى)S:=p朣[a(:8V"ummQ %~IwEgP@RIwVޚE]xRy"]܍Wd*e$^cFA/A"es~K-wχ`lӃS1P \rnl u|{IA_ѣ7w}Wh^Cz/2@vVc s@N]ӥ\r~GNcqpۇAc+DڔL-%[؀J5 27 Zɡfj&KZyE\'.Klk<-qqۊgE?_FNO5ض׷j>3~%@ 6^v!P )XP=~=KE vLu\VJ&euR#qb@Ѵ@/c?BM~xV+΂G#>[K؆W .9e΃AN\VgzdfC܅{4Z&hˊc䂀Pj~N䊧1ˎ*oĉQԕWOCbčA+/O8;Ys~OfI-B~}sKٟex.6WHZA;'mް0YWZ3 o[S"Om;wո I\N[a,d{:8eN*^{}5ފ@O8 t^LR^m)Oۨ@౴ٌ ^ʏ3Bf9qGH]<s;EBӶ7$KI]is@#(JѴK X]m#Bg *Ձ#A^* zzaB.*:DP,^,E90̮E\Z8.}z##Fq)Qg[αkTEr)_\ b]1(U] [&S|`;=|Qê=pDagONyF_Ӊ;G6,念"(ڹfm/\;쬩3w+VS3]*y;D,P#M_!=L چ=`@\sS}|c| _ֻJWAQRB4Vaϖ@97aI[A)#v6qlmiWyk RNʅq?R{9 g;AUs9:[&3rn$\T]Ye^sǡg O(fҶff5wf2]{28$CLv ,HH*7jb!a̹ݷ&pukj]KwpB;v&Xւ]j/`4Z kҤ;ϣYOĽGNR,wl~zwu V=ld/ǏE+I"ذ\0DȌU+[hjH)ˏ!޼bw !.RV+ƿ]^wiౖl{>WD|B>?)NQn,!DEͥpP$Tjz3f*.|);b1՟hfLی޴;m<2IVA[N[JlmUL>,([A|*YR&y2L{Ò|%OȶL KΰٸEع±M&fI&,{[& )^*ӐVыLMh!RfWDEYѯzfX|j dXbZssX/uUym1AO [3d-pdm-90]t|G*F)/Q ͯ{Ĥ|.y*Y&H1{4*dcg𝤏>l32(2rNpoϩ3}c gn}vņL]*oDWRS+cnxr/J￉yF'[ZX%ũmK?=8'U[4]5BSbVw;^w͈ȹѓZK-ZٕBdc#[O.b^-ffH9C[Iu8 .Qz.ݪcQ$)&Fn)>7Zvj+ip$P\m_vNY.iكK>3ab{ i3~eShrEe[.)h.8{K]Zﯚиtx^3覂 rb pO>^;}_qS>?6kQNŰX*ư`oT泙f.cF,Yv]˳2F`n\CyG~yWS?v=5˹{GW䫭2[a췙7%rՐs/͟B/o=4uT[s8VkAS=1 @m~gȝTF'{Jq"hёU3,Tfy0_#zZY1Q{3\&;_?6 -?T̛2Zejr5TH7+Tշya-켲 F2 FNFXe:YY3xe\#s? Ș~+p0KmɏO43l(T?Gɳe>P9}^oR^.at'݄3ؤ-*][W/xcj0GP,1p rLX#el+N~ }Poݼ`y4!yT BK։;u(&yYڲvuӧU}_BOwyR dj*#1jd.]YsǕײ'L3gțn |*%O\ \~J7;2+pqi;=+g;9/,B)67Lroci\[ "WWaw~?4bMtN"$G}Fum8C"n_ſaԺb%w c_̊v-X5{i!ˍFWVG&<˼t\/1v}>ſe&oow^RWrcE(ðnKy.[VμIY71m`f$hLKJs6zsrg?}FE\dUܧrUآ|'hvnH m*D"eb^js6F~;DC \c?qO *?F@?[~_0 7~I$-?7sI?W? 1rX䷾9E}'C 5g35)jy-BHZ)8W[]58?  pYKUBTLMo*yn0K.gx~[R'xxF}cfʁM}c?K)w201 [!eW%Oz:ki/'^ו7g|W!x?tna8W͖y4ބon? /[}g$F k3մVFTb,^N#'>œ2V>Tݗn/ ,+7՟S=Ts>1hOJJ^HMg,~m( zUs_}::`2M^7m 48qtIUsǐӎ8Oɚ\3/irLx>g.$j|oCH9""+j7~>[Cۊu)7S'U|*wv8O}gݣjh距Ee?f{~;1Q5gw!#YYlZ9_~R/ac S?R^yRsXmh̆k(j!*u9lgNⅠq=ge{9C{[;z2c.ǬcŲh@E&*W5*Dt|)ōNRk΃qL9? A]:5eto7-z0>Nk01smaC w % ah8y%_^^v]o䡈;i˟H-v.Y~1댉sa65F4o: ,[bBygguTxC߬h|ti*w$DM$gDkvmcwZ?BlL,||y5GyҒY~6)Nd49 s9]u6̶ͤn5<­kM}F!1odP08 ooշ&X}[ы`/OD6%Q[w NZ-|R8Bߓ??O|Iz*5-o*S ?^Ѿ,D0OãOb$'\т%q~OkVDu9'cPς1n@ϋ*|ٿi^2-qaIlŧˆ"1O|MûII<8gB?sH d$r|gϠTZ~?N1o7ݟEbhawE5JP%9rY/"/ +UsweFY4tKkW|X/"}1m%S5§߄i \K#o6F.evcg %FLU,[X{z޹Կ_m]'B|?V}8k%r׳)WIqhPqւI9 _&$,g얗y=$!S-gf<܅&@X5˧sT,d^_z% X‚2MzV1b0-,O,/Ahf htP#1NPJbd4,[rrփB/r08ZkN?q\ƈ8y#mwr 멫b8}!nhF3/$yæ)⮖RG[w"rQ%nâ00r yF>q< @Y &r_iT"ƹf/xm1XxgwƷe>HY3|Zsj.cKWf)x14n*SB( 5C|4p)c <9GL@ &Ω-\ƽyarbl򶲟*0c+ 2h8Ϫ6̵+_ZUU.&"]X.3x;,^9}EzBqNi7WۭM}Z{}Ů9eSj]Ixqn! qr;e.+>t R.Ⱥ<6͆=]`:A}x9t?#L&WLt)l&kT|"+KJ7O_/ Y'23)I#TcF sRͥ,nt(R2 2I瑯,cfs-uDĆމRޭx#eO~eB 樣(ΏF(e rxOi7Nk=?[gi;1[ya2aeQ8~"@PQJD\كR&kAAy\q#v3g,謿g4r>jo`]Ew6([ խ-ca8s>3odq(ܟXj0 \ O$s8M18G;6k(ޕ+19\ڛ6?frX1>z#'aG,_'^j&vL'ԌI9 LTF-_e~>-Dc$;#FBB0vY F5㡲t<}¨ W*ըLq1?,Vں> 'sN!Eb PЧ![ȕ\Ɂe9sX2w68L&}Ϯ~'GJU*r?v7U|֪:oO:qQFr6 fs3'Ts疤߿b/-W~Su+<5c0~)~b0R,dտX3qR?mYx8ˬʦ?ML(DU%w}Ԃ/>;m7/dPmusXh n/:1#Ţ^sE)rU 6m7a _WOH_Mނ}IOf3X p eK>&R5S$ ]&t+MߖZki"4TG2Smit%Z.^Vs\mAd՘#[E"Ӥx3&M pEc-J5t>r?7`BDP71yt~InaؽROb`+J!Sԇvġ̫O|$l/)˂Y^./mMw+[JiiPb[Y(G˨\ET4Yۥ4.Iŧ6oCA>ܑHr:|)b5{V(IUxQ_[vi=)6uVgA@^7dEgbl-/fe,<7Rp^ЫaJco=김Ӹxʿ2Xfrx@{]i MѸjjy dn:"U3TJ9m-缙"Yf*Ku\6)yF>ss*-u0RFfXf߶W1kxH]L $I'w)%ќ; 7Fn4ork)Kld;Lh9a#nH*[;Wj!ުVyZaA{Ѷ3q"ŗE啲($^6KwA.X=Js0 }fbCQ%Zb~ +g' Vkp#6-[C)'l\Uce1aK!TS:ɥqrs (zbFfs> s5V!Xfijxka6Y;CD/> VND?l<7Rs_|ϱWnU<#.;ZewdxB)MPlVR^~ ݦOk[D<LR9i5qKzaqP/KCv`MOc1zAk[_ܞu|Sb7&n.*n-ŝ \>JҠ߳D!" +PGZ[9Dv] PH ++6G]&/[eE,ܙӮP* 55J&N3 l*-+ږ5:3Gdpʐ֮L{`kW)?Fc#f?%B \lSu]Ҍ%rke>{CKE]%q c=r3ɴ. 7%pSzEz.FVs5DRRUf_uDC^OimF賵aŘ8Nk &QO֥ݦ)&em }Qd oM>dvT.$8>қ[[ML&geN q"R.Su#nNx?LWJvK;XL2t"W=fNexR%ʧTID,d1BTfl(KL:"E_kr,8X^[pKy|c GkȥnO56QB#ykk$-52./=tXi%EzJ$]`Ջ=|悌s7̌*z%ijoju^3uN$KJ. 9ϙ ED6jΐdBn lK-Q5Mx~mUi~w}&LMh8.&k=`jyY"<#9D.)"og¾{eiejvR͘O⣐Gӏrd"y_IUlgZWF6Q?9_5g=ˀ[V 9㪿 1*32$G\rl( ~t2KZj*+GW3N/TU ku)Ti!JgW)ktdH1lg]R3ݙ $՜%HTrޭmQ}tcѨX}g䦶z5ovUm{&f.sˉf${8ٸXXU֚XT0= Jqz Qַ5Pi*8;GŐi)/q[[$ڗXOJvc;l(iN>g^/w.=PCUB"BcuP YQ3gguYhe_q2ů=`LٶD òsm7ħHh9#'5aɆ7t[OTsd5{ʢ(޸l߱iVQ+SCG4hyr/": x#ۉoǜ5ł=)RЙŠXE+V yv90˙b-Kg`$7,&TuGf?P*lO+Ͼ-H "/ lP dގ,lfċ!zSv5X,ycU r Is@X_Wy\05.-fR]yF3ӜHY':7r`w8)uQ|ڟTI¾"nES5t X0:ZvN1 4V`~n])b l 2 TôZGQ.`:']BeMPt׭䐕&.<%fp"%:瑘[[b0]ч6rYV5dʵs~ɯV"hurܨ*˪e/\d~2rz 3oCo7jWK֫%,Qnpm[9 h[>p x"z\]œ,X\,dgY0~yXdՋuMd4'j<?"L/VX/$ag1U'?gQ1 88,1hX2\@)j^[ݕ;,5h|G"Xjh@ VKƐFW{yg?GW)ek2r`<}sJuC{7 ͙բx:Uf-5; URhUKT *ԮaEtxW=C L*[rCݽ7}dS5;)̃@ #;&R<NOՑa*蠫7TUTj:q2SϦD B}?5Jo8TÃn 8UNGX*%r;VpW~{jJy=HSIJxԖø'PYץDyWe?Ă#8n(_k%ŶhQl;,">X&+ʊ 2w J]! o=GCCL5<'a~ʂ6C{N\ά԰&[/6<1 ƏY3?`:$2Ep"elE5p:cv\b Fณ M^ eHj&!9QKW PD#|:-t|Tj5ՎMڥCD+ʼn?G.G O7\x•fr̺X.l?q`x9~=5AnмT`W NDr7q\ʓ"Ӯ#'X޼Ѓjq_ˎ (p_M@;>3f![`ȫvt+ݬp+DmqD1Æ> W)κT`Q pK~XR~MeUaoʵ?[kȐ3k_N%_!C Ayjj{HFu1AL ÷_-ڷ7awǢvᬝʈPigf>D;a"xcgކl]h8) ~M"<"̿+` fc8^#\`(J<Ȅ/ؾ'c'C&!-NS0MU'qYT㮿ei퀞ZL&y5$W= `;.5%Rh=\: i66< Aj!0Pb1kf->&΋p߽H~˧繴C,~>uIVsq͔Ua%oؠ|g 8(Qsf{4ѝ,Ʊo}EcT 4%~.wE$i2V"/ԛ:8ק瓹Ūw?73ņl Tʐt4% "3e #gnsۖX53`K_5k\ka2mLc>OH<&Yf*}$py[f!nBI.PTwN=SoӇr+\wD1IhbG3fvNK+Q6HQ''D /;jWG!FG./ͭS%ÆؒAYZbOL[a|!SP[YK5-9͊33i3JAZ313= oN'_wFO,-]gs\%jPW-卻к\V깓PdGMA}r]4- GvO bv0>z!33EYeD91YBNLAB0^!5,D݀؆MF4n[tU^;"y(nq\Sğ EL!'EGc{b[9mځ 6Z(!&2<^],rQ+Y.#C*{)d~ )QKT2c2J@[t ˼`vY?Φ #c~;_)=8jԐ6 ;'x3wF ܑt^.gBLu0&;\>jn9\ۮX& VuwE?N /?h`+J}*݄aHNJ 0'BW "I[m,rŒF2v0d6]Y2g:2_{r;KdNqpbl2j~ߧ x38OJlg(2F 񘵅W-՝Pn] AȒz51"+'6)[5 v9?@YjJp/uynĽ،dX[tñb5^w0Z}nי8EL|LdLj\"?(lZMODL0E>trՏWQE0J9Hwtv% m$SFT@TnnT:' vʅds@UA7:؈|_潣5Wu/Th4Qp5#Lո{Q͍9a0h0{N[lve'^a83Lq_cqѝE(}/)Oq ͅmBN)Z3Db- %o'Gذ~7b5oDǷu~,snXQ:c*;\u.Z'ٰ"KD3\mz!9<XKܹIQDTa<<<ˈwaI?br)葏0D$q1s[ #,3e:P;/,9#ߕnKon Y=*)Z=@>/ o^m)f`!Z=gM˲PVZ3JV޺ɺY37jό$' j=5:HH|$|h|w6kBӅ#':Ӡ9e*Myi+A$ xǠr~kD8FŘ@~X):poXW[*4)ou^$oqdG:oÍ[oG+;Mdӑ>&w8KM 1Q>>貋NA[έǧ ru^~Uѭ|GYU ֯؁&ՓhG&zN,#܈VfQd$Kc?^8PI?QFڭg3!X J,-5-"vN9 VD̛Ukjan>4!pbZâ !j6#ɸ[zwVV%Ep nk|ⶬ0jBsgv?sۜ- tm%+cD =rm&fػ@7FIHcG)2wTRT= #YOs͍9A=# ڙr3". դ="s;b?۠-i~ꤷZdb;ۢ.g xyϙJ#ϼ%Knzf-pCy׶h R$*Ua\U-W)Qgso)puFm[wCI[N1R]!aDžSG Qj݀)" bkl6gVVML§.>^ Xܜg3;zs7LIL .Ks {v0 @W7 )Ŵg5 Qö9{h'0b[[ͰTRAJeEG\'ٰ43%Թ>%?9:긭Sx T{P>uɜB A?炝Q#q ,jNeUh1Ow4HQ7FdC ýrcƻUs]sA6>/\ .6n֠JAxci5(t\!1+4Te\>1MĖ;+Fu\;.BePlPp\%M=%V|xώtߚxgVz-JT3%L6;Uۇ`Ty6r6Δd)UQCDxT Bߖ-c h}wHetVU/fpNېmFօa3 ,ncy *X܈pHbO۷Of\Kk#& Er@ N5u'9\2[QT/WQ%Oz9S2>ߋjc.U!m,Pm/BO=Õ#Mب䌿Z}՞W +kmJwO"S){[AX6+zz^<jcB } T A5̰_׷Dm5G9)zmaշRC(0Z-zF[deU[ȋ9wQ(Sw'jWƾZhZfR3`Qp+D&R k뺩ᯥ^jhLk?Ÿ9e6w(06;}lMrEs#3ڇFI쩨,8egU~V> A^Dlrk_at+8޴M#<#8uـr)ahV\գIF7ǦJ*CRmq48҂BcxXQ| bõt zGvds6rK`t!(mr.C1VUGKeߝ[b|<1ݖSfMUC!Uv5td {BԐYTH R/1)wUݕazW%c1Udt]jf.70}` aȶ ҇\e)5ryOsr-[-Aqʮ֨f~[55,7q̃ Z_ش/K'ͬbUf=Mrj35s_6\j7Hi\@6FQ6F~=|][ǯx?f-'ÇoߺP|P8ɳhO0W~?y.cCz& EiĹ 7>VntNL>`<33%0/cR(w1% Ě~3bܦL.1Lm,4"d*kPzӼwQu-wsm 0Aӿvf",)ϚVOT~A x t˓6s%lO>`ﬞ;wBRvľ,~FS4}16;.LћdHt\׳V͊Efg {l _ hu#sxS5?JL:[)A6xqG%\ϭ*iwʼn?;|bmIDۑ^Bu[M9 *~cư,5 gs$]TR$\_;ά22Li\@ 'Q}l0S[ؾ 踵9ݕ86Gb^ N#) huǚ:.9Jd˿SSR^G.eF9K.# Ҹ;5%"~F][6dlZ3}L flv~ r(rƐ&myTpB$)&גzӮk%}Ze7iFz'޲xdog}n/79Yts2rur¶ӎA$ ϰcV#f- \CnqGISoݙq6>"jxDMⴼTR\g꽂/a]>8 pG<3=^2zJM %g 4=3M x*zge3z>Gy1] '_fdj~S:N {*"y1C γm{\,˓ e&.0P[kUSTݖTm 턪I J@Pmsշ\ `mhIլLً$jkp S[!Bn2ķa@~Esu|۰DICud;FH_q !E6L-lr`Hyw?d, ŽpC~ l_אq%Un.ZCq2xtURiwV4MH150Qvf۵suS;PQ-A]~gֹbF{5nwΝMe"/;s_?QZ$g=cΎ֙ߍKmU̝+H`;6tTٰ';£Qd1Ag5?!b9}?Hu6'nI@wEO!`dC 6j%s_N]L;wtk5>yw!J#3SAˌm^MI_C;< soH ) q"g3Nr>V<?Kܷv3}.'mv&A0àt;g}7gr;W$o7"}' ?'#"1n@VN_~s^aީh6g/eVz@dS6N_ LmcYdZro;=̪C3hIfZ 68)bM9eJge0+g}b+mviQW(`+/ʞ_f _S;T..]w VcMJå#`TY̹К*>t+ n3ywqH^PNw>og ͩ\<{Blf݈5ۘRR!g\NwY ㌳bt$z3dxnONYê>Ĥa)U kڿwjFuժ9CTsF , sF2qQP3ɜAʅ{VGxa3fͷogr #O+'ϗm 5'-,3q1o <9~"t|7%S39SLWNYӠa7>iJˆ1#s$Bh5U1hp [N"Bq9\iQJM9D\='M U"Ύ/@٤b(&:;t&Ib6c2ٶM1M~Kr\΍W00?fWcG$x^> 8908pw}_q |W/YG?Gq\0-e 퓿S|móߓ~ !j/s?PTИQbL7$4?wߒa~Q5Vj0x|b3Þp ҝkyTt[Vrh޿d ,n6aȉk_Ke[ZoKbn)2(ӡ,6wM 9.֔ЅB~zu8. xm&8yj㞄:+xljE \B2ur15P-صY}(R\+G.3o3G\ߤdrpj 8,fa#IuVj1ih8Wh#<6Q|W~TnmQmƙgMDZOKϒ ?e>Vvq?K SeX-{Z'>J6\}里\6BH}*AF|'PvǙZ Z O9|'=ߪ~%L2`+vrA֍LO]ܱ@ yY)9eB\ 3]*WJz mq qHu71Z'|cT׼cqκcRBP8)EFUokhZl`n\ (~pdڂ)C"NXy8/g<9f:̐iwOa5Pq]t3N1F#bh#u(k'=99ζ-(\5r6ַ;pFNIM^oZWw(PUoN UpRJ˖-cy2$A ]`S t_*?wᗜg_ǒ!tij>ЕT#lo›2^4{PmÀcm"~2s?p席XJJܜPj4j[j~x}ĉ@upno5'G>덱 &t\tj<"St>쎵WxyQVUqg?je GǟL=MLBFxμek_r-;l!98'oƲ^Y7u,#vα%Y_I%`,NVPYjF{G|.T,X C%uJA| rmY-OAg*VkdBó\󊲗ήa/fPp*QUZ:lY;w܆,}6 2[Bݯt GGƀۜsv͗NϸW$ۂt_vm\j!%sr:nN d+Ǵ9Q󾮀zw6ObJ;[ȫ3"D|ulTokۙAA(n+bhCg9='돰kR)Ps!$ֈM?T Df'U^;mP*%ݚ IuՖ|XޞbO7\ZG6ZL7G8`WRmp^#N8 bA?tHh/-ƌ҉J9΍Ckcz9*u6rBgN>ڙJX) P9E?d{"V"-dB[ \7M-+laBvH87qlH&3PyƂo?8䝃< \-9~:̎"'<ЫM "dC s=<&M5q~ :_>[( ]]-LɃYOj#YѺs퟼|ޙo9G9(x87"VMCa lE!qnt78kbuU@n<ҙw{t2ZYVчٯwf|İ M{C%׵{, ):"`v u8B{1TFvڰho3Y"7.F>n#/Cc}qq-Dl-N]g9?sunؤr; _\r_fXndkQ3˵ }{7}x܏NNA c`>Fkk2=#1h`ReYlft4H!-Tv?/ 5ʻjb,H6l& `GrfN#9x U#ϵw+߹@ffq =N^O:w'7gthEtW*TH`:ŒNb݂b5NEփKPsu~NgɍxyZ֯Z~WH 0Tɢ+mvWFs`yFż J/1?mGk儤Hwy蘿ʢ]ΪWDV{K)$4`)clq&a ~:Y,3m=E"~+_[m˂!A~[ej*UᶸC?746ruK #YɌ˭wn:wm#OZG[ɳ'V,I\dP. _] &W8OgxⳎg<,mȚa̴[\]mPRԖL\hmP:>u;r-ޒLt%~b`]xueGlx$KBg3EG)h/COY>˺)+y!|\ [ێ١,D,߼pČlϛ'4wf/vr xY(L4 '-wI=WBAa11.T[cftiõ^HeEp'~AHT wWb"T!M*k*DY@heg㍸؎d']V]}掾m8;k knj4H>eY/29%AxyA԰J﴾@->x$(>4鹆 y:c8KunB3?;bQE^j&^|O'xrY7e<#klk(Ql~m;k)дȱO*# } W/c IznR1EAtxhQJ6B?TPB53E@~ou@a_,f*]KޜrN5NpQwxƦGK2MJ4eզT+X[N00Z7;Maܾ"ք rZ6>)Q7/ [ W>XSf)nf J|; Y}iqDʲ[@=z$mb`c4wz.KݮM*yͮZn(,*|  kٿ*ܱkɓ~.[g߭7J)]ʢNEebG᪀ߺ}S磚.T!#}BPɴjCKg=r, C[52H5d*>*Y@ʚr`{MC(|XZ.ÂK.d%?T[W=bU%k-dx: ( QqN34*Ł_euv wiLd>Ux]$PcӻH6 U}A *2?Lh%|ZU3=J j!,sg2J/Ԑdakgqq3!Jй{D @rsFU,N)sPvuSr!\ TC-NMS3ĘHZň h bVin UaU}AuyN^f@Es cb3#nela1(/zI:\ҭZQagMW4ZQ3't3IP7_e%Ȩk^uUEp[jUe+٧Jp7ɺAep5:NY.2;KFJT[ 8 !F\s<"C-p%&*O>+]T@N)q.51[X7f\s-?s *.TZ374/2?; LMt76䝩])6C0h6h:XUIemI5/XIF AC={U$ƨ/[sY8yEC.N <V WY[+L^ DR5#U3LQ¿LppjH uc萗t#ni2״F>'kڍ6l2|ODbxMQxbdžV$M* !N׹ )aކ)4]Hդ{|!N7FY@k55E~$gckȧd1^E/U&)Ax yq#C)B+ cPݘl ssm$ {#͜A׌5;o<9^MvgBR|ZO- |-}޸!ꎰʘWV*kl41w[ jaڗl 5h,ǯVCxxp!8YmXQz&MEF.%/ CcKdzA&(eafxgafQPvrR/ޑ9=ü]ms'YDfR$!wzij87hծc> >ڄ8c݆)UV֦6ɹF5^lLSnⅤJ5 Zy7Dbo٤!j Q31z_74Z#7vȹ ܾ(/rVpa}ƂsjIv 㘫R!,sJLxʋq7]*TN/~j;눲w Pƀg,RBB1 ՃVgE,( b$ZbmH_\sĉ?-c> ,+[ ˬO][3aL6V4ԙ@nD_E׶~q Cb!GԷTBP!?{fm2Oh? ]S7@t$akw HDv_{yގYϒvK:Zuژf\UvTľ\3xqe.Ԍ:=L ;KS^ͥ_/a#F4+{/1jf>/)ǹ .˫ NMH?) R#Ss]tmi!pii'R1IrUxBƺ0)&v,0ִZq`H.a#䇃 V:{jW$UƦm־1CT!|8w$ȴ@wVn*fo)PLic;A!7:n~j!X$LI=;S(3O;";fW>[2?K+( M:6v<+Vd :[nn76JVl{&;ژĂ j%)Lb2)DfIfU)f^v; Ye*=I@``a Yaal&8Ӎh5#SsP Tp)Fp''?UCO;ŒBE2˰Cx3a3F)TYbl4`7*\TF-6r7YA\M4(rɶFhW5`Qvh0iplmE~BhE\7.[z<0;R6`Jŧe˚SD9bJx.-SsQ8oM<4+]{ ) n|jGS, ^e^RNe4.nR\;V]$%s?{D:N\mFy-;vbiA[le6Ά[>D6+ĭ6$Yf(m &͇f-]w `4kFq+dۮBݜb|tCI֗*(m'>oI F-fn?Ne'Z2 hTDm9ĶyPDݧ)!Fħ9e*YxQS!Aaa. (O*L%Ne|AS 7j6*Q6;f$ڠ _"UeR=0C,'UE~dHCn6Heq)QXZ?8(_)m@ qDt&]RGߕr%c 3QQ9Enb,w!\g "[X! Bb( 6kmiQL2Ƕ"W#iauu6.(ӧBnh:rPsZ<3hzQ~7_ K2^Kl^%a),i#}jyƷ:#J /#puuC`?΍}?/хZg%_scr0dlVkPsS %NE1jvS?Lb/({rsM^V潇@Kej˳ij[3R :<rTP:)]|R>"\)όu oRp 4)m(%oޙR5R[-w2MVG>d[Z_g#VZj| ܍.SD#qI3,nYJeeٸp'k%oNsϝ ,A bG8'5oo=gaά(\__aQkV$c] aőTb}3>5p ZS`ao췵FaV |>V3|UWrg9Y ~&RgfJ&ȃ" 2QL}OwQR湯sj< 8E䉞Ɗ%[Ă'#Q:jLXMpsHch Q"$]e".ۯj>C]Hq-.Jg\.X֯+hb^ƕ c)jIE=|m>WZy:TzJ1l=ϵbZGF]SN9ÕW6pwyWfY!C9WY=$k#t*BcZ,kurET?)4P;G$}WtrogfhCP{ljK` >Ny1|{`}|R0\*W2PKWt?m@Ɛ|܊y4zn}J9T}*qOivܴ8VbKD541 c %"ӄgNs)>T5̚yUOϝkCȦqh"bc?#uyh-ңt0?~ht*Leh 8q>'<9?QD$+PR>e4hšV8uV[J[֨1 .M{@{l`vv{,"FP{XԟKC*Lg]CU{ XL[ 47<&H*Wt0&BQ~qL&ϴB.ZۤKslk ZƦ lA ~fP|p%9K!~.̘3zȃz2ޟl, ߟ,Df7EHpuN\z;ˈ> c4@ͫAn=2E!h# wDǜ~-G 9QCA(NSj=Sd 7^#TF),\HE{jDSvęܱj-bOU,R[([}݋&bmX# k@h\V`aj (@ag\0ˇw3 a#n]ba1Bm}B2![LPڐʬeDZ5S`;Oky9*9ifaxc|yj [mD߆ũP/-]ְQ^D!Iezn̻ LmvLϗJ*1Y4DZNCk,!X*0D=$~M/fO29TXQM`b't)tG+]54(uƿTߕz˥eYEέ3dfђJ~GKeҚ?%֫et UN$`xygڄ`d&fRv yνbdBz.)ӃWd,XWhG)o 3"뫦0f!bDoxъ{5D3~e{1Kg?tMnT-f#<8ۭqtUdpP52-e6v ̯"|!v{{lط>vR.$X2= rVxoqqsïa2ukXY[a}oFjDst4Q\vͦC֬`~ɚ7d**$#v &b?9ĶC|*UПݨ\‰tΞRXy-U0xk$?DvY"+P"KȗfqƸոahb|r2ZՎ73gUp]^M$ooz%P(ֲh`9$r?[G7)nl=JiղUʑb5"h NVbs:Tc-WBFjaLr&d?JW-BjEbתh//C}]cm+i ܑ '"\*O鑂q͕ bj~_rJ%>]Qϭ0A{CFYو蒯B[ ŘwU։*wM5gzAzH(E7f5^hQyu{%ZޙkKQ䃨1w_9Wٻ$?f]Iy1M_:KBՉjZvd<;Pb~z a? Z=z3S##mekP>D>1Te3M{>]Ԡz(j(@ ~\k`r[澦iI[{N՚n8y7 lnK3␺0fBl5<%}:$g\&,fsŽUnF@ ~m.9_׊65 力sz_+E%>Zl89 !uQ%옓n3utaxr LdܮPYfbkBͮght; !hn{S '-lx 9}!Mb2is-S 6&u?;n1 =[یb]'`FsmP o oFk@RGk5J[H;n26ۃF\s& ti Z0F( OX)AQjYD_C<ɶ6%qct F?L*a<#e*]hsixHfCH\|Рȵh`4Sm6g+$X-[ƕad ;6 t fN՞5k_eD'aԠ.tYybjxbf)-wXưWY0L$6=0eա #qxyDK\c'"s҈6dJY?t vV9֐5wLS 7@Lb\G.<$ޙNP|+J:+)2Miqjr>j֘lonθC"Z31 B ;fmU_={1KXP>+37< juUȑ e>-C_z;X*b͈OD14'1"JiGaUՔ`fO=GU@ղ޲sD}Ih`|Uό^k')2LT @0م!)[3om &=^x銽9tE1(u \N)pj5Dp&jⴺ{HjRpyŦr%qy𰑨Җ ˂J47rJȏz4Ԝ_: 7/ėu_EMhKl|"{ؙCТY=NɬKOO,X)VO`6SL\}d.t^_18{3<@^6N]M6展*㚉C|Wނax!!5,Rkh̿!4gc@[@mI@9#k5@W2D&EU?7D $[M ĔHrNPFL+&³e^o$<)7c7H?1B*XOǿ24-om/\%˶5r0YNFR.ԜO[ קٽfɷc gcՒzAy3[Q-:$'g7%vvGo+ :iap.[͸5O4R ~m'GfgK)]L/9A3[[" ]uΝpZE^X(0?zw:m5:VCd}e%7-{TYNHkQ?sڳʈGHۚ8+]-ֶ7GL/GHQNG꬘R5pRKSO\[L0f;tr{ИCxz\1 3aĆbŧ)т)[xN03THغ=Xrlm m̰:8Ů4t o vK>gCZyjeȭ:V+wśCgcO cUV57`:BOKEl )OAz%hA$ +ph@%"=AM qq`h5OΈNk@mW!3YFFmIhwM9.\ߑ"-zC7j"WW q]4W`R Hű1o>SL5CZ z"%.%jj"VR{i-BoKtz᲎(Y 7mf:n67=؎p"@UzJ כ޳Ћv9SoRyW<# <99Kݛ Y/键ןİv8C60)]Zցnh>4]+ kѱL=}zB!tkk5+婥02CK?jyb<["lYJǙvچnjYn+APi4oYCCƖ z\J8凌v) ;M`d>ưz-!SzG"u%gJAnen ʳIlc}f7<*\b m 5hND-¬:}a_ITo6 65 {aTo s/)}1)D)[תNr47F+sjmt\Yb’$tsՏΨɔy$jĩzcK M܊&S'T u^;)0ržaѽMYG%%JKgn-7]"#V?\v5gm}My#LkSx:fc:9U!~M~F-e3ـn3$^+~0mڻɷ /+"qwgWW ŏR}EƊhFcl6ޙa2,HΡUxӞŞ+gl*@[[X9N3⎡qʷ,Z_nud' c =49hf= Z#(1zcP4ea`K#)(C.scq1xv1a#zNꁞ҂A8|@9/\x+26Io}8c(/1Sx|F8]1IVU[ gH6sC!l5{qEv/杷z.|sh:K9qtњQ?^+Y=$iU?2%1pC#ϼBz;{ !9ieo6qt>0z:Ǝ%ΫF7)#jی]L+PPPb9aGU.,B*LrkFE.YհpŴy#M6+-J<ʇFg 5"MkŰpruԏ;Ot'D hXAI 8(YԆyLO_ukL<ܾSPuxjG3Pf|;y *z`2~m%: [lmg[P[C}`_&'GmOYuy=0^PsE`g:9[)b4C9mwU>%ZsD4RַglU-JĴT@R+>Ѭ}bJ(K͘"4bP.Yjpǜ3tPB%}]u^XsTURpZtIU=UMktMe#졅J$}0CȡVl sč8.ʩܒ"wPӹ {*3^8וV {b tdETsqgUi2FE"*] 9ɻta>&~.d<5 YTZgji ˜*zH\vi qţL6B16qG {\<X iˆCp bv,O,CzzwI'u~xG/+T7b㤌6;ޙ^d5A= {F}>ʹ,hL$!uwbr0:ǒo6!}8[,p[l-|h: 6`pİ1cG0Zo!w,J)C>WNES; {BYN~tNÝې)xSyY^vc~uWK5t A4@QS3kxڜrFc7 vYaϡMh*zhiGޖuf ͣ-x3SlGreoI86KzeMx:0}ŝCב8|eosΉN(XTUW*iAJ>NәM R;ac]>eJg83Lz),L:qGxzքGs>@vnp.3;X;q~ opqR}+kt*ァ5XN<:«Kg_k331O&rUM5sِGChxNmA72ӟ}ΕhgiߩSÓCZ>—G}fwV@YMȍ屟onݹQX;7v`x |qx֤Qy ;b96Sw!ԃRl3U=/1ko5kFɩ(3- JsFaPb,FM{^u -p >Wؐ,[71c bu͓5+[5JYsm*&VksXsDU ;K;e"'H?yȅSMo,ω& W2*}Va@ݏgXo %6br #zK-F2Gh )ΓeSDocU9)Qj"M5Ya*Y&U+c_\ϥd(_nj0yNG$?Kl}Vc6k߆}I]);Ѿx^G"-c`N@LҊt73n׍Lnb[ q5K@oϲi&i!TRxEm`ě<iW/]d|ۦQeg,I gjOSBrQRiN;F88F=,auaQj uY83!5yn$\.Tg霉F٣D;l3i\U`0 idaWqngOguk+aOڼZTn_"tfֳ΍]?@un͙kl;vOf%G4 Bg }4l4}l Mc'^U6i_gP3QxL,̯6N8"}F8DYd< EZmTj聰ywn:E~X3"@wHSρѾBXtrZƒ_xv&Ŗξ3Yf\` [W4d3h qT 1 ͨ82Qr1'Fa,3Ñ2Ԗawv :{|"EcXhVJst2v69ܖ2>y޶!0zwC=Mw6AaIl*6FD'>$7`!̓T`Ј6i[`@%ܷnoݽ~ .<<-Lɮ~Bw,G"@Q:?'1DY,W3G~[m{bPvgK?m CΝh מ cP"wqNyIT Ϩ-.O]uUyG9y%v7#ZJ{ʹݜyc `T2LG^k+[>޽TL(Gv=d>'e5[!Z gm6d9a7uϡuؙoƵ(;އuoys/47B uPFL$ ";i&@u4 IR֑}8{jrûƤ^/VhaYEݎ~M?6W9xY#Oi}eonB122ё%#y @k(|w=䂗8MO>}v̍t:v8l!ac,8ӏt笌.U*Sjs@- z};ڴ׸ѫ9CYʔQWcky .Ukfx=HO5,ч3?q^j\ys:t@ebj_,Naw_< uPt3!yhLD- }pҀ.PwU;Gz&pEB:b?P,Ѽ2DAgPJ(!82Kk1ҩL-EddA9$5^,Y("Q? ?]\RCqo ScP NM,Waa %^%XC̄TxQRQ9j9Cك|(9)h!-YJ%>~ȉEC08Ax+ ^|SͷrVDsO3*| m֢q5`(9JnK=<~ڞ_'+ POR@^noIrwc~-P*pߗ׋؎ݷ7ۈ7{pE嚨GO L&$[̏!&hR 3se *M#ɛ1~)= d7ȁg:o'~~.{8n1 qNV=-7 }LI;)4UXcܮz=y`=j 1(ϮO`@1\#LDsj.3"?~q<5 R5YyWmiaptJFmyR?Z+- HJ sM4sT5buiF]/ӫUgѣ,c5bB!X#6D辊P5(BpNn[ddsݎ>-ʧj8Mx>K]uw<} jآk{F|Ԁ'c,衋aCQ`qCX3ጓzIZŷ} ֎뤡Bc˴M15Rˍлv7:مF-}n2M'YY &|Mj)H7&jV,n5 R!Zu&ʞD5'.\Wg+墾 cH%VQ \[d[啮㤱Se|za24P:Ωo]BGϘ}nG7i?'m5 7x@mGWOpWA#l~Ļ'VΝ#CSxPHY'{[l޼Ibd]ym:7b@w5Bdžwj7wI`Vwq<j(%ucj1>G=riE[ڱf[P*Vse7#&WhnJ(07ы[~f}{]F.ۼ[<{Lhr B&QxࠓXe'j`\T붼'mHW}х? x2xi'=)s#Yy!f‚&!b{d̯6WJV#A҃}D D,,ͩA]fS#fl-C6ȉ<(*w[YT3!MebIxafƂi.UCecVDs`K4L]Y;Թ7Ŭ''_0,z}ϧO2GIq63n LS'x_RmwC$@NXFuwMhSԥ.[U:6։J)rꐎ]mN;/Qv*jxLӯ,9`_њMd7wHҲfS1+ǯ=Htb/bY~?Q(4֏ Emt&Dm{;S`yr>q˂ڪ1S砇#%(k䔵r;\" }gYqQ {|5d Z犊/(JDk<>ػ)ϰӭkOr,HW~(+pBCYH$1bAO/}n}tC#y6R_1+FehN^1+-2@RT^*I)!Q`'O:Y0@b9 ?pًQշѭS|t !$E3f$VЇ @- @KU!K} $E+DKkb}ZmVODvmώA]ppwu-VxlP Vqz?>|L} YSO^U!0t5c Q+Eʏ&BMv7}t]ۊ'.tW'U^I<:F{]xni//o/3B)m\a}1M( # X0E3Ǘ9_VCWJ4]' &YFW*g ugkf JEmd)suS;> Am%yGiu&75[yom-XCăgn2pBF"T6:tNp)(aky6Nn}ѭ9&Am4f[(G SG2?1 b--?wn!yruڠki؊bK(عy5 ;@w:h'aܔ/ڼ՟jOeAp [a\Fei&sݬ1)jk78hѺssXZ0|3U'v:Vۀ7ifT4꿞߇%^a(,F!Z:d)` Ml.^:Swrb3ѫLۭFM[w}XK`aч =%zPfÑSv1RwL?ؙ|tNn<hiq{Vy0}2-E;mۓ] Q=CZR@YM!]!KG'oAqGB\G~MYƾ[74&莳{;q4\NTFB$x `pyǞ J,'r:yf 6+M礳ᑟ+*S4J {N. JĎ$,87NZ|hr~$ BVfG+_-YKO$:)#l s?bm%ƪkDiD^A` 5D&'q3| >v@EAtƏ V1 [#ALZ \%IK '"@3C"}k8J'feSK  l_ ;Ƿ ˿Q!C3y} ?Xɷ(q@/t^4t Mرyfok-C0M&OԘNC)xV*pVH89)ץqAM U_D5F=(-b;8]8iڰ9б♜HƙG >ktΏlg&9ZˍMx|4؆Q5RT3&-c!p/]J_#zYN{\'F ]i?sBNjCџ!!A"oXcqzחxO*l-ɁG7{|ePKgӃ(W~Nj@~ _on8hEq>EZó P4yw,Iƿƪ!e]AW{Rq`Y&18 hqϷ>6T Pm)|,+O]/9 "STbSuJ혥8hDv́-zBZZ|EIYsqjèNP=1㟅o֭ķ8R7DgF`E!v 1DJ.2y/DȻj ]5O@+('!PV+97lAOt:,?JTx8ESj@ L?!.V{vkMP*Gwj嵃=֧7YDK=NJv|~?K&MvNV3}P-5jzߌiDziwn)&1 dNll/"u!MQe%`;Dx>IO 9.E~Pp dze(;e|1xAk遊tc I!ِ"p-.^w=`ds]n ⚇) Z;uF \{luql\O~(`Y` bZ,9KMm+%8H)}bG8Ǟ )QsS9ɥ饙9 fFoSqr%k#) ͻ ʼRk)Zj̿L'*qn2^7)}cdֈ ̍#uXi1D@5Ej_f0sVܴ:G 0&eP: Lr-m\O @s}. 0!$:jVzC9qba#Y S*Chg/ 1!K":(# ]~J#48(a=\JSĂe>ճY]Urj[``!Bg(jPVo"NՀf{o$r.l+y#}lv 1d!"YG9Յwf=m֨c,AxV0Xif:[-Qo2ԣ5gDvM]( ޛזx.xHtʢWZ f}r1{Q I~ ,2e} m̈G xΥ丳pOX熐I| xr !b0Q A u[֍;,{ #'p <`kwrnUb{άanŔ@7+qaL=Yh_ΖOxd8=z X$ً5 Ͻ&I3-"z[`rY09}R~9+gmQDl]wa%^Ri䷜%h[u@lكˈ 0b|4St_3ōB=k.TVy:bcւ.hi4{Uw"߷ACSײ) !zDS%hx3ِQ4EqSWy䳵DóŲVٹ,ma p/ZX60R%DwM}e(o'Y.OVʶq DK閇Ӫ~*oа,GeT:lA j槈9WPKuޭqJmJҮj 5}DT PqwR)$` 6RҸ&c:N-K]"V+T@w,?cYH8dKx&H 0)ƣ_S. _̨pİS{#%0Wx{B ~H@/c$7&Aċ=Sv|$vּQf+)ͼP>`ʀ VZ%jr.ܱ_ y:ڹ% <"eT5NNNht[EA=+H@DžB-H;_\3%;[|P3AxmŒAC0KaN\9ǂ熵С'^FatFNt{}>PtkU,.O4`wwJqBdctf ǖ0jaY%Ѻo]xEaԌ>m%"f"D)p'ȺQ ʖfc)è/):N<{v6I#Z6t[acK+cmХ{Na)n DXrUUfnRh,-f4FP^`!ڂ@н,jhc]af3VtIˡwom`MOں'Fnq'rMQ,+8"P AMlȁ|s[_&9{[EXyxYjU'GޘcsoSG̲$`R-Bs?FeY0ʳѷ2"\Z1B8>s>hC(`F)ZE[bK~kPyct'J<5.-v  ҵ8R,z̬UMԕR<ڷJW)tKl}{d$mcKx* %{;(,+WN GEr)_)Lٚ|E=?ԅc IJi\JF߮_^1y)&ZL+ +}³l;J9Dv %11T21y_"1lv+\Ǐ;4/;^.Đ]y9rR+ii-ADEggX32c8n,Hԅk؈GS UKXݽ2m1 \i.n`4sBVi)E*Zc˸2"1e D{1g~Xixn|qM(f3[ E>uQ7q&cJen*|2N::Z=ڷj^!U3|J@2|(TxY醔4`&Dd#i ( .5^}7[} _CM B[,}*e75k.1fžS‰W9*#&up#pE 1GgU'a]Bs S "Eݖ5L0~ϨۮlstnX "qOjm59,w9i#v4@Ȟ ^r/փֵpo)ji@g0ޢIt#۲^DtpiL* &=ХyY^%BŝD="/qĦ? 8  TO<ܠ惄TG8sGe)&RKD5JỎ=\e2f¥ Z#<+D+Q$UXHa䚪'4"_Y. Ϲi93wlx#g+>ozP}Gt2WS1H}gCP఩ Kcj|,xN/,<[=[aNFљmR ̃S=Wr $4b e&+-SxdآVfY",Ò|h̺ޱFCC "oQA:Nրq\[l H e2!,xn:-ы|eNRej5gM YwLt|hڲ[ۈuаM=@#Zm':dxxzmYGc񰼝~׳{Q=: Z6Lq:w2\#~8i̞WW]kV`:.~p 5Gwl\`P$TtP /D76|ff9'/^z\垏kŮ|]赐o9\FO_* gG#, jDKRW7IC= BCĽG7a5v]+aVcG&HR=DM/ ׄsrHA )ǁRNW^&jBLU/V fF4j#aRYw }ԩAT"Vg4D^dX >GpxLCX3JbHyEVEs0 eE)OuR k@"k7o|YWVKX۞PvlP2Cy }XLt 1ge; p;X ɡP5ₕnkuP6߇ϦvY ڟ~X0,,0ˋ=>ƊPkR&qa;B:6 ls=.q\qX%Q7ncMXr3uʠ&V)?\)~ֻ#җ haRXwo Dg^7AtW==rKVW"U@޷<ޱHkqʍ)i!l7pE_fMwM?空Ev{&1MA%OVʶr)\ߥ׸bi=罶&¨ Z'\2Pt(nM}ʧ]3#mjKC9ZZ;U͐򳌁JAg!;p,kE?}\P|sL+.CcNF+mqBPpc7O: _ge׷Ilw{{;tvo0 8ըlDZS= pC};Ϝ9B<g禟O\߽=gdψ 8ڽ#_ְ͘`|J7>Y|v;N(ʵ:{Mٮi0AȫS#|s`g = -u!W i^Uo+>mf*ON"Ru3?wx9̒dHc9@7݁)'iEp5 1avMHZPEf;S9=Jh-ZHux9;rt0@ٵN&B/k[8xb`U_#R# ݙy&(re^66Z,=}$r,v -BڣhzƤy>6o6!Gkq˶pv\0T#! ;Mk4kDM.W]LJlHq8\oعO`5vv+fDR* U4ڿS)a3a%R"QcQ,k8'%;2HmxQ6'杈*0{)Y1~e7𝼬~bAIcX#/r/߁St OyVh\ɴt0sDZ[ط\cLU{L]Cy+7^<)3n NQʵ- $u} +Bϡ/yd<1D} m8V)6]QTrIu8UnO 8.bbnX-qh1sq7ޡG;9#ENh@\knƩ(:لuuQ-|p *m\w_sSWXs @bs]r➹!SY: bEs_5vL.>4Q@e`G#bkƱF+x#7D+L76G+Ps<։|\2;DJ9 q9rfklj%Z)Pxo :ѱtBOПA\?oٍJڢ6fÉk1 BK׾ݘ֬gF΅(c=3B7ekD7b輳9b-L4B[sk֪X9q`[MG#PfW2ױڋDJKh[')DK=dYGOlZHY$x&AgZި\s!SS4; =.2Yl3nM~Rot_ț`9]cz~lǮW`CՕB /WWv/,`9! ueFF1Εo`i_uK0q[ 1j$(L'w cdh`RGz.jB_pAh|gF5|{:ga\ax9qGɫ:=^iI _[=QubWp58ygå@#]8G6:f( ,Q.z/g ;U_\12.zmOS.+:e0 Pպs 㯝;sK3eB͋ zh#ʃ|rЬ[aOcxIwg=M Ɲ^舚zKrwj/΍;ml@Ymw[^9:pw#{M_ W2Ww1N'Jd7m-V$uV3n=r)?$i}!w6P]Ou\qt)@D[g7NK17-z|bH,?\]3s'⤵ɗQM Ϲ;y0s2Bi&(P3yF%Kvm5"Y:ֳ_2 Yvn4)S'Тozu*"Y-ex f GM,ˑ$4^m "ӃSOO!i_nt\K;{긩Zh5FՉsC+ *iK;#adXܴ [0؂!Ν@`u71%]A ;wn/Cځ?e3bQ{X2/ /@ONf+?t*~sT^EPshhi:^(v[? CN*ggsG @fy'+i_ѩ XSDF6r&KuHh„&ѦˆqaIj}0\8+.uE O-sN"bjș]#O=ʆڅUS"#5d ~K$k17$>%a<7~bgjq+O85-JָZ#\*a1"ovN1x+0:4= BG{`< M+aL IBW{3)Hw<:1:܉ǧoXv&]atM#$hW%X-E /gnOuvɄ]xIR6b ٛ X%SԬwkVp|4yLowkBE,[(,H[|n%MNEʜZ-5SܧHk42E"8ňU7 C`Sߝ*7G੪Uy{#_wS~~XoKբD*9V'-=_ع`CmRҝGa2zЩ`ްԌv1昿twھm̄vWt@T8˛A JО#>穃lے}67Snf6vXϪ}msnJ}gu>fTu䭳KEWk 4Aюkm?ZFLK5lddYu-n!2[mfyXu"<%Q"ZQ9Y!tH'27X en|ʉtKp4 h_T%k22GAXfѱij9K.QqԵZ uOTqhkljXU/3-1'N&$qI׹Bv׼$|+ǡ{ 4mn,$q&*۶C$xU8M!3C[{DL+KX3hM9*%!@lw67gv6jtB:hr'Z[WvM ~AjI:RI]V?s…1Y?jfj -;Z;-xy f"D~m-8މy@Dґ#$wҖ'^ޖᘋ ͣ]89HvkЂ :^bpxM>zdw  }dĠSܿ< +>P3n! J9$6K_KG[lqFmQs!s%ᰉ9S,irrc,I¶uD*2Hs^u^8S#4]87*pb*)Ƴ1tp&ys\;ϝ<9kn#5Ĝ6ՇZƭROx1&b,<<{̊q'\*STb> .Q)s"Hke:g_U}⯪)5wzwCЮ D@`/̼5vx(8戾7W"lm< +<;Grd~j1w9i5Alj7+i{ )s";!nd?fx9i~6.;&6, >#̲˪3t֕I:nBO3g jޖ<&xddB(|R뉫⅍߇U^]@s*U!vc k=VBҩ{#p>ٿ}w"~z v }wJq p~dh:l/n߆ВDC?ht{wO 1QuMkk|PѴq@jC= t9qSF2$PD4Ä6+ZGgD.Җ >mG<5vO 'lQ/B\P^{M2r{g BB3N L?`ZTzԲ_" hCٽSFh`Tq'w|lB(]Ϝ=U{j&dNwC!> L+KJRڧ\n􂢶#4xXc(uxx5ibj!L0ܗB٥Yqt[S8:U gs }][yM8aη;0zCL9px XoыMlx(v[Wf'qLGQ'3Њr<A3R@;%Sr_4>;_mszĆaoAQ uL1Rk1 RXKXě98%}!l)\<[=}9EsR]D[n.jj5ņXGx65kO0 pOB9,NWHLǤA.b$a-G`1"@BLsNo($QS׊H (Tlm+L9viw9j*՘Z\I[3VwAϱsjJagaGBm7c*Ϣhhxg쓎abfi&@pj(=aL򼒱uº4:zL~RG7q>hL2Ry̵ˉ vGO<נ憦0!sOi; O̲SO:+̳|F&Q4b BLj@ LXSS#~9"^?a}lРJ!v$)\lNKʏt`DFӤ>Y"BٲuL7/C" =|F:,(VB4V÷FlїÊh=k_(憙͓Ryj Bn=}my!রiel+DsXS3jU%3I≠sx'd 젉/܄rὺt gFU٨ĹlDdh!_.Ƃ-m] I\$ 'T/ĦϦ;2V!v)4 J@Q}GmLt$NN!)CiInT2| ZMg,-;>rz`y1% X }X%v/bP"$܄ሒHwap"O5&wgTKl=Ps0Bwt)1{֓Q¦ [TD)"9%l/^>Ɯ܈ ,,o0Gq|Ә&!MԎMN&4]},z\L^7X& q{9|B!+G$1UhZ@\Mlpn1W'R|e>8-ω{J-;36\.+@ ]RhEdI$x75P60r";VCC['Z1u8fYK"5gDt'--eiH9yj-)F*%uF612𨺚W:$"_x>ۖccRW6&Diќ[g=K w3!+j`p57aOR=F4yaq@(`\uu[y)SMoשenqq8!Z uY/]n"PBMBgb@G2Perh|ǁ:⬰mWAps1-s'{Lhlx.q  hcaPWܖQQIwj>ScvK4],yŔ4FZMKZjie$OK?ӅUNQ詩)$Z&X@O%tr1G}~OCmp~?< ՟skHsԬq!Lm8"%CZ1rii`.+NB{*8b% >o|e`~9f9ɇDژնdlѼgz:HmE_$A @=ת1詹 Dħm]ս156ٵǓchW}!U{e898gifz5$r3 CQO 7cnL暅2ˆ+2Tnf5ޜ6%'ؼѝ^1MJGLa S@5ضc64q0DF h~CɽF% }תGOaq_κ,xWxūag䑓AY"LnFYWPY) ;WYγa!A-l:#SqI愈wlVIL~3r1&MDAp)RNr@6G#̿'}ڔe ܚ jVQoF ])!H,^yuTX쟶!n|Pn,=t,CN\bՅsB9fSҚnLx,--#P~s0܌i㴓M5wSndl{[=5 bӒ>lj,41]#]Byqp5Cd7O"L\5=uMr5ϨI}1sq`ߘ l>]Ο2'5Kl b/M;t 6^Ab,(*kFA}L(+ḿX+#N3Tؖy[v=oX I/2"4CyJɁ^M(W`-kࣅk2OE]u<}n620x^'c=(QMiA5Ɲh x_*d˼)gnj9r8voGOҙ]B(]଱݊^_9\1 .ݴpA6Lskp+k.o[ZHEg;FG+y (&9iu>Ppq c]VFf4MQ4HN$@I|zϓ:Llu6LXסI" Z,53!J'B`&?ދ93Vʞ` {Bҕf cX=\{$Y1B9TI7d9>% :ЄJ_[wݥO8V7lo8! v\:7xm!zt,gnD:WSP+4BhlP&,c2S_']`yEL?'܌2*;]`UR{oŊd:ϸʱ'ZN*U3?k{9UDakke +nƹ8[BTN0x #oh}R:^։I}K ~R]8kHɄh t$Ac"WA ;'BSldБyh.EjXA)7=Atx6c&W:z WSZnnm}N}]AئCSv!ǯl'hC.\\:#A'vnixg!Zu."s)Rl6ftHً,[cv5*wÂZZe_vy<,#[`Xen9-AD&-ӏ8'609h8>qés2bg*$j~9QmӯT&]aF8j(n<"(27 OEʥo>q1f(Q'_ٿ+UZioų'.|X)1ECX7t;'`%`K;:pk+|>S7˽ɷ]Nf^mL]66!y6d;u׆n^_ (>CD+i҂( 2)1Dm3H^HD'u & 4zvzd=^'c"ln\sRʧ;@Esj3q8E!(P6[ ,-c`=kGkua[p0@(!ilbEALi>PsQO:^&x7a"b !я SJN{J8n-csHU.WA\(/#T8DNJL1i%RJڦWK5$uN&RUNѕ93j^Ua\VK%oĕy('RMB569HLQ :3v3}lD%\͸&XHc z1sV!S9k^v"qU4>y4o'܎l" 熚G?[LEw(fiL\3xbPs*hVojvൕ.C9R'W $7L霂uZxD,0E3 q d P99mCQejw•\ T r6i #'? !ņR?Nn։pm\ :}5@!5x2빈w ]Y-5҅o:(+bm.bRY6}C9"  n9AQEw_qƪЏ4]XNu_)E$ҊP{s+dZh\nJgqbNauK4vQ8 X$:ЙRl,ʮ;EEЌt,<C1*N(!+\;n}k!W}ι- [K'ǔg&1m5{˫5*Zo{Zk6b@wEIYZ&r*r4B<8Ï34 )e-AFM"Ds ؤgY.ùhcoPe5ָG8d-zmSۆ|N#eCd'_ÌQdMڣ4Ƞ8d1Ү P9n]^wSj/Zԉv[3&&eAMVo'D  h+ \bׅ5fQ[`gjX̶R$k#7ggVxM#qh%q>}RP籡@!,]QH9grZ'7y ܩ}t3G߷OFK5^ߞ(N1 `ܛ`'h:xyr_ON0ymܔB!Cq. fV-ntZWhl;H?Yq%';[9}i<=]&HɳZF1p7j=<9|Ap#M{|>9>u!92eO;dzl[^|v]*Ԍdx+ eA csÛR9ᱦ_=6&L 2λ/̻=lWDaa@2*3Hڦ"/_ih5=DdB^^T!vٺ=A8%&D<&gg cˎSo蝙!ApWs)QSa>OfWk,I@MqVсEಓ#)&03[`~|F7_,=۫;s[*ȧ'TJ1Vܢ+ٳGw>G:Q3{γՖPW\/7_^o㚌YAfn-H"eP?Y"cgc٩!G~B|k4:&Dw`xDwuъ_! tqȅGc%b> dga)!ȡtrYyعfT*@kH5j\%X|Ej(\T(4KX|ZөAପhūh~v#Flwe1۔F#Y>CQQ~ ݞspUgԌ!5PؙeQ\G ݳG'@ͻ'0j v8It2P3Q5K =Y+0iKbT/6Q9ܾsP4 Bca!bI, [DӇ*Z/h'mٝl?v <5Ʀ\-0rs''㲨%Bx{ĺdKlsbk_}|xy7a _8srk_v{W`? Ao?ٻgѦh\y;WtVeVM80ָPSLIXgx {T2RY7Sظd&oӮA-S4ٲi+:_'h=W=OFzUHg+C; dh?=;A;P-)#dEiyikZZ9Jpnښc?C#5hZdE)j1HdGE7[SYEGFUS>y {.k8;7gՆuylQIw3]~%R6 RYɣH9ʞk{Z:u'Qػ|~__< ьu opMKVnM /'Yh/7כ{7k5:Gp"$@rc~x8N*8}P,jDH(N}zPbw:6LNQLwP1m4kFzUY}iMY&6[? YjHQyF <'ʖuUt[{ҟLGc}Q1GhZPTEVƊIX2Jd1E=A}"A3@1S\{şZ*ğE׼Cd #G {A>Us 杉=@!.yPuk>H).s3=22o@͡ak-ASkևOh,PD'A8-0⑭YA2k laK8PXp=Hd{} kb(T"de`tAT~_;1Ut3N}_|{pg|??h -_zM /\ҕ#~.Vzʌoa+rNc.mqi7Kwš5DbJwhq@A"B2c7_$.Y2~|m :Rz8n;O N4wKq+z6 r[kkHP{>eKy|;fUz!5u\'~OՉsSc0-'&Z#o#noy6t=-4|n&Ҁ6w mߨN:w7sS'Z=h3DSwz,>>n_k:]@}ZDauPˇ7wp hk\?>%8$AV'jV;:-{[n9;E?9 MۡcMV0YuheKyR;zʈ5k\@#4C -ސ&MpAfe|Bǃ%Ht0 'a ({Z m\\+?3?OS?O~s?y}.epoƷ>o ?p?KK_D/<.|-{GZH\Rm\aik%ym6ʌ7]`طдE/ݱC1j,t9Dq,S<1 q3Noy7SN-UAn,q7"f,y_9J_7Zo.p\ ܎vHYmCc}A.B_g! [f[bmΆ;F݅g9d1 5K٘Tؖnވ~6OWL *a/ $IbILd* '(!zfE׷Q#ǯq| Kҹ) 2-2K=G"D{*Vq|W}}s~⟾޿^g~;> se7]?y/ůڵk5_,Xyv?}6EoM߱MCCPƚ{ןDFX8!hY11 Uò.;Еl , q tB7 TBQxxrqnnwb<C!ɖK,{!Y_GYN,l8כ1z[]J63i-ޟ&uӧlEBhn0?熦9u=iP3B5 ŘQ3%Z*{ t8E( Yl߇5kV|!!wi'J~z疇$7vpS7O6Y{[gq6Kȉú֒ tmi C]fֹ5Aoi]c߀&5rEҞ fnNۥC2?2>ZЂQZyh5u'wHv|nSB<޲ $5O萶@ Rs^ҵg~?ϟyolooOBw}DZw7>7qo{ҽvOO\_ߟ?灿ze+}hSs~&Y!ꉟؿS|3{ߛ̛nC w 3g +v\Uk`Q'֛J(x Dsefϼ[C cرD8 8)gmbc.SP_z:N6]롧o4 rSyGuBoz|toJke);hzRmk遰XG7RZMu\ˢrnA3[(brR錎gwE5S r-񸴃 /׬S5[+pV|4snԜy6#DHjhL085kngWq~4+-}_rUs}։g,vYnltW~2!~w<Ge>>@5mz?xr}r]z%XBso/[g`,6L!OϾ@hxv`I~cR9(&V|o=V3wyE+P~+/,=ш=c_W>Q~Ƙah#E34//'J8;g7aykSo؍^yO= }nJoץ ~XnvכW~Q;GU>NW>+z?M4T+5kϮC9i2\yOd򹶭W۽\S4nqyպ|mS C|Mzgxn#3{T-o1CKXY^>$ӛoύA! EzK\ɴ1?s@u|$ qېnBs { |1Z9a($s&@(j|j98z ?>_z)*5SfH{_}gx=GO}gg?3&O}:p Ƿs @,׮^/d_Iϥxy(!?'~Hkdp3# } Vk~Ӱs|paH'+g3ƛZt7OXYk_rIfUa 1중k_y_9 &`J8qQW߷v \VEyM @A*I[[!k.i`7YP(}y;Y&2k5. S14&WDM(Pg5&WOKWW:*tDd;,mb!h*}3Ak Yz,Wp6'ޅh7YZҲI%@,}~Y1"FP7tn&O'>\Èa P=5k@\eƕ?ܧ}_?WGcNw7^ ?AeA=,3s?yۏ|nS7_z3*ߢiBіTN{_67|#.]z'?+WWc7\I]vqW a2*3)jb0֊sr#[Es"Y&k&h4y$oTzR& srFWqBɦ#(-l8"Fr²I!m-u38"5 vMuyg_߇8V jEYhȄVՕcœhL<(dj8a^WQRSn93!6Y'K܉b-u}!Af;CQ#NQwޏz>Z;Za@C(#2<kcFkgܧ#@͌+C׮s !9_:O|}?(~j";M~?( lQ>v1wB'+wl>_<? ' ݏHsTkȊKoTT^3'~'P{cO=gF9ArDWbgG쳟sg^8_(ra<ǯҚpͭp>+iFi@wD9t5,2IRF+Y=]R"\(}L (&r:ҁUED̪O^KO1)E(Ք0nvoo5صWӔL05f2N=X5ۢՕԾ#CMbV: ½q:<:"q 0}-a2hGjer }&q%=O>Mn塮9h?Q#b+ؽ|ր +tנ5\>1Eỹq>\i=+~{8^L4CCRRh$: Gq@I7lA/c| [?f]tpt}fq퀂B~CM@-퉗kL;-)vCeqh\,w/AtΆ/|}pu*Wq,[ 켷* @GGCGȺzgyE3{jZMga50)0TYY&tE',~-khзCgB^#4{ŭJ)Imp)6k&D.ˮC }#o~(E7o"7?GϼPP޿x'>u@Blyg1#R䘐!-o{LY)R"C whǷ3[^"K#fhÃq3vl-[Ln Qu--w2o9 2B)@ͷf_V=zNVHp\?9'b ›+bvrݩ;&l&ZOɕ "ªrrD5ZI2杉-R#!Ԍ:Ωё90}wj#Es"_YXWO=zۉāu3MT?|KQd5A[!1y&W @l: VZՃ9HR#/=Ѕ6BGMS3Nt@H03Ocށ ;\_,ϐ[/O 8+Q4o6^M4fބrMʫn\u-h z]cBph?'Rӯ&X=?r͓X SPqمU= 5S )3˨7k:05txgb-"DO#P클5x8ƞ2}+5"'9]0p Zam ru>'i/keeٻcR_}*te}E/^26)yMe&)[jiBHA@bC=/_$ 3( OzC`f ` ]A/w<1|?>1+S)( 1Bcng}0e;O9͎e`_yW^| fq7MjMf!Z?Zyc%WQ9w|2z*Xgcr(u3[ȈlvM[nzub73Ɵb\-%@FX>̿7C\DxV<^8BUgj)(ьضg0')=jD K/ N6N/p7>`̧7.@c. 7+SXN$c?Git/6z}Mw>E=gCVNƒFL9R=s} h#!x$Z(-x*jʒ;Wz%nj*|jzַ֭xgMCZC9ӥXdf qЊŬt3bZe\4I, GUFxMkI;bXFYG*O~ۂ8~>Om_^~@bbljeooFް[{ğOaM:;vEP} cdGImrD:6*#}Ox-F ݈ W\&;OO!~s@ӁCsC8 cLPߓ?$iqG{[{GMImʄ8'gCDhu%Op(cEmNlڙo`)YVi;Yώ4Hqm@;(ݟÚ\7knC#"Qm ̎v[,Hsϲ@+gv,?oŚHYD*59Xi ;wpX &B֦VZĕDF) A_&A,SK.$6G&0w5Ck"= Ó:]F/aJC#abu ,5SFN{<` <'ܽwUp Cc3@^ݿ w ) . {_?FǀD?$05q{Eq|CxvG~ϻ?]>w2KPà7gC9}lXE-%uLMzUcp5v5P]+@3^>t^E59YVpZlPYO^e7%lS4Dum5lsX(r2 erBYSFvs٫*¬4OB--! M{pE-ygGbZO/]cv=QKsg>~Ňw>~G{K߸ZdѨ06J[ ;Ӫ Y[m<[m߀Ѣ7~dS;/dփ>9&'tT'z}Uwg5e]gAoA(9И-)R[yvj/bx0ՑJ]HQ~Gvmsinb%ׯ͒0z"wV1;Үwkh0$v&6&Zaxy۲$G,蘘7ÆtS9a CI`$.= ڜ,.gu{ujŰśf=UgQX~=8 m!>>Ntn .M|JAke3|-h?~=nA^y/Z}˥oZ9i֑N,{H媩u)Z|M=h(L 50ɏOo/=X$ 6U=&LDo#+)END7u)"n"\pPDGR/?0K~- |TkU[d78Ա^ '8/ i1n#:E;u 4=&P. dau ec0y{}x%|!>%`4]zM|AzH u cL~ب=P\ 5\E@NOe,Y1h=,|^G7}GEzʐܱv(8Jj&LҡqtÞHG}_o[]8V܇Tdz?ʊ=8J\\o.zx(uSxî4yLQ^qBn񧏈n~ Kq%M'< ߠqJ87:ANjLKN!V0< 5=*2>phC=wiĘּēzHyY:\AJIIJ(]_uVդ-t[8 O';w;, 8W'xywv9Ag?wG[8qIǢٵ S HkGnK:QaNF u)JXK;[;: .څ B:3³PG(+>g? ߿~}z[oo{'N!^7,wp?u1zu>G)j2\` w~;qrx7o/K/>p}_Q /9q`.z/?ykgwEZ'|_946]9څ5?<-<0N9~S,[7n>Byro'X{'^c:pi͍M3݃8pG'^Eo@%gS=u ~pssMk9qy=M"6ڑ8F&s䍣qv8&8'y$$< _9~/zϟ)O6xa2>/?<<&X0>=g؟$vww%g'hK_|Z=wW1-]SPʆ86ۢ?[1EZ4]&3#LZzo#C:sjBF­#A ,1%p[E5;fSX`A/h4ܒbCP~GnUI{稭_j9U,# bfib8snɺI݌;wn|d{w[xTl)ꐫnyh&v{Uw"쫃` cdӶbbN "NKw׊!_~K/< OϿ^ڸs㞻_C2=Y>o=yo[ wzG~7ykpEgad `޵ ͦd?{_|kM:g:M7u)Q{IqΕ& yr+ ! Mm ;KeDGtȺDl"-~":#\8Zt|M8 uQ5zڎzGocχM8Eifիmo ]Nt/dJ(.)Tjwk:t>(1E3p _^}&7 5A@l%KfSFYU} Ԍ['@Cv>C,oyҶ3%POo:غ~RG"o{:b_qt!Mx= 4㎳5 0F<Z'ocظc{܇##ѡg0!lʯ=.,ƫO3;n9?7~0&Yh? :Xsbm-ک[D .5Z+))Ӽ+(fi|ˈTL a%¢ h *,WtB٤DA&HhgH' 4k($!|MJ D3!vAw !Aq)4<kCWS?uPLyg"qR p#1]rreRX=ot0ۊşK֥_9?_̯hcso|xo [[?llcڗ~ڷw[wŁ =svS^:xzxw>sZf/֐5:%WX@׾U-)Αǜh7Rttݳ=/ǩ7F> Ԗu~O7-ycdyG3knfW1-zκ<e}7趨,غ6:" 3 oGK7=Jfboz&S4bר9ȗ|. Z!0WmMJ$ޫ82aYasaB= N0jm1( *}_ jHn _~:=<5^zǻX o!~E1XT'.#(|{5b F32kv7j㕕0^c&;VH_L|8EiY 拣6,J_9ݵedCafcg{Xo5IX|91vomTs(K5}^[Y[\)B/r :ֳFZlq!WUkF>EeԕZVιx NhF(딓8a,3Bg@Q`验yf9~%-E^^v!:O):Ίi%G޴LV߃Gȇ?a mhE8Y)oPPEIv[mGD~Cn'X$>tg$ba{Gpe ۊk~6VagcX 񡚤i:B2Z񳮞V- 5 A"}eM e, T(t ;#Lagu8]^}o][r4f$h]y2JA"կzu!\Lv2Ny.dB3:^cփUuיU[w9g܁RLot.ގdY' 't:q^"\;#MovF3c ֆ?D#NS Dy2N~cag}6Εۮ$r,{+Wy >ΎE'Lv>z_pg!ݺv-[ _ wk@(P^2V0Xy}\k|WWbގS/SjW kDЅ4!ɓpiLe&Rfgzf2lx%M L:;ׁ;^+>)>2 =fBj86\2A빀TsyT9v ;+݋/4A&gԾpߧsK ;N,UCf u\p#g]ފ.T*44hG0z`vk ?m]1O8 wW̕`Y-SxOt0&IV+ubEu U^ڟݻ6ނm4ǿjGݪ-ѕis?'?-c9g=4N٭^a*ԫ'(|L8; #]gwa|kx(BpI|GDr_ȚhOgs"~zUt >=81Rۗhtp,쬲GkmTs\ 6f+W>*_Q ;QN<8M=xADExp Gxۻ: F@apq0y'{!? v?$u霏 m_Tz f{Wh "8gek?"آìl(y08{Zd;㠹-AR9ږw:mk6u9Rl٧t˟F~O׷CmLXag+hj'wlxbHrWl<ŋjLc]|{.Lƿp>9!!^ ;ooq<* µ?(@\ k~Rb-"ę[xrw'A8&84P'~]"x}2LeO7_aq!@LY@VpɿAlհ&u[a>rY+Щzk 肁;bԭlIJֶR|n ^U2/K,2t<4NS2:o"wYӪڮPXtgf!hBQǠOgIL -fo 5;7m J#1AG.+iJWuºb)w(}7sIk!h!UN󫯑wFQW(-'𨉘w9Ofb4UM'l) A֩E8GV ܋fa?{pr0\<>رd !Ӝ$^gg^F[,$- ]^z dxuh."zXӒÒ0u+Ry|&}t%!IC[cg0Z~xJfA(.NZSU:2X[mBsd&S1Dzp+-u&5L!hsuKf%J%7{['R1xUKᙦa7h5]<ߡwޠ|t|{7aD`mm nm\#۹NnZ,{`g^S= OW_W?8@5Liu:JkQ[7P/c-pS㱇< (`!qG+" Օj:l*]*ɊQG>#ak :2 =8sC15h_yq+T@8_+kŊSFIW" S^u(}a6]qFȣcŋ317v-DWϩī? EvJ r 0yX0ĮvEzp)Ջ>1~ 'Oɣ&,K¶S!! kWYiwGLZ~GYm _ڼPv AhT B,$fkxlz4&.Fj3=>+ОsNsVu6ju1gŊgmr߽>}(^NK[>̓? 78޿toa*=O$[J(%N(/^:W/䀿BaLN'?{l-}^LO8;L4zo.ǰoM:xUBܸrn_w_gi4'3Яt5WŹ')CKԻhvڦ蜕bM+x*GC~1C3 [fk}7֦fxR%P |{Ost&_\}SNvNK!^;wvQX"| _Hqe%x3+tX7#s*+E7x|s5 y/+Pv=OB^%S^jy> YƷP?r ŏػg{/=}?zߓ??;j}w T5sk׮A: ʚ~tnCГ~*ធ3Y*-gN9Ik%breݺñYDS>(X?]%D6tf QBO)jwjdl:H&Zcl4Z +^S@(m5BiS0;]O9clԯo{MRe VL=Dڷօs>[i=hin5);k BӢM rCgt8#n| Jd3fYzE::3EEh;GIOx }^9}gγ<:__im7`yOzptNZ`\ү<C?'~'?Ïo=x;ٔAL'?.^A~:u_,dt{)yKp9~F^yj2L̊zwU1 `^ȧ5}.cV҈FG^Re*)WӅN4-AT@6/4Zc!^Yq !#hT3qaӴ AӉ08.-zf6OBI1a)Vbwh ikJPF!/ =Bt3P&tְF<*d( ,\ߴ e}́FNZY10{~`U5 N̥?{)O%+=бUc ϩ7H6#k]w7-'4A\wtE{ߥOC篽/R~gutZj:k/}_s, y?|4t= Y+xgΝSnqjbָ}xyY9>ѺLmfX0_`Y(^yWh?xou)ěxD|&.-hΥм>o&XDP$k’";ք!5:x3NZ86 7ȊF͠5 /L2c5S0E4y[#//l<_m>ycw)moMg<ۜDm:y۪ S46*c0ԌA!R9R 4fQ+݆7^{/vokf#':&vv+j j>Xup|0uqbGxFb޻\ ">|=k[΂8U\ok-IR>r23X։%v^ 5Dzş }v<7ʳ,^CC ' `( e6_)yihDD*514E8.Ps[87 j8Kg!(ۓAgwW') yaRDtdg;7,xw:C9b % l eFuhWYeh9NjM@wXymsΪW-wfe27 ڣ#h#ntlM5҂D;ߐ' U};X{A#zGNA̵BpLѫ6-Cnt0'> M9M] 7#e1$]]ggNFZu5)Yvl0+#hox59]7=:i7?Gμ}!L?ED/?rGWސZҴQ&kzB#UCo50eSi2D[}B(m8thGh\n} jH9]Q#c55r`#;++NH @ʏuY3F͜>gWZRYDz)&v@W'[zBSyCR4K@+3Л9~;6ngag*Gv~9'է;/A|;uM [ ;/&7|rNs},g˰u?S{ꉟ{ָ݇#zyu}Q?}~Y;׌j*L85;[.u'?=6 ׷+NNJ'<|}1mMIqFI ^aysؗ߫]AZru|gx~gO(?{b5f!cLVMZ$:JM<$i Fe T: svVB3Xnl?xϒ}`4j9m:xO7dž@gMێtKÛʐ:nYn/^Y/qf+;r@3ؙ^лw1]eo1F7R4in,%lB0lm(Ab"!4JQBMl\IP`XLۤMƝdtӏsovEfs=sys91w|gNl_#1ixg߹sκ${[UV=9߭`$Y.DYݦ'Xo"W hX^wL bΝ>SZic@$y;sG |O_6vg7~S<}?vgxx)|F!A4hL}aGjNّl~Ӭ~nM&Lbq tֵdt0s$~ϴy(r@64a2SHZO/(U;n 9↫%*V8D ^;>[_?3Ц9l4O>8 B"9hSA@IU&m6ǍI\ w5UOqǘy[*_ǁyY,RqpN`9Kaib$n,ͲU5~@s;ǽSzg0߱h;L's5#}ل2krxM;8=y(DoQ>>VA1β~Ggg2ݵcwןeXby_>I-pp"' \\R4l7k]G(35afeϗ^۴ٽ,dޙF tF*.3l(!6u)Dp!TL\dk zvt!d| w6[q}}qDKǀo}u7j<9k'] ᓕ Qˣ.F@Ω(α7{ϊh7hR!;5izJUNbЧ+7zcMHBmf<(f?a'B 4m WvCwX{T4fܵ`u=G#ag rTSn7jT~n^RUй;0%Teκhx05,0ZM}&`M 6񉸫QC? b) EMt"Mw t|}a 18dÝ==~OKN ~ו3.G%{k٥q1;6iV@j!lqa`d&baRY0R1w$μsƬq}9b5عr.ݒw]Viv(R@3DuR (DDButa_yN]7硍G/p62[c|##&m]:} 'kz/I→_7 {3 BmT:8>CEcc $m(ouz/j pf6*갥˥z/K\N/l}Coc[t49sNQ)WI 6\gpxR\kh9@ hW:sè9 eH'V^1Xo]>Z_1D]3i{Υqo KPxfBl,!0Z#˲&7">Sa wÞB%“ւ_FvvtfG[Vv zĪƼy4FǀݒD尳$}KYq6h,Xu{(J6OuƔCḇt0EӢ |gLfuGnX?#K--muyfi|<>qs=,̞ع MO0?Ձt#ɫC$9F(m:n lmhP81Y=:Ex^T])RĖf+sY %+`jADj9@cIi-\_!7l|%\KC\ f!c vs[&ȷӄfLk}C'5wMsLQĐMf4#6͘iq%E ,y0|wEi50z@*D~|gccʧ8my.DVRC6Θ|<{l^Aӥ^c28}]7r[J]KlИ >'_5KwQ߈T-z2bɱzG\"jhύQHotD5XRwkakǾ?"C&1G\uwg7uRKov<#3(U`́daC#xd#gxf\.AVn8 \-k~uzglSȣ b@F%'yw|ewc}sUrd)Bh$lzȧb+`MHC>}@6~ tSzUqַ~9f ˈ0MDcL#ؠ-E&I_⚖ ;|=L; |zUT Ƽ. WB#`pEDhDZ4H` y 0awʱrTg&a"?je uЈ^)[k +zvdಹšwK0T=;#5@\agwΙNdWy/`).}Hz4ݵ7<ĚJËhyܣg>wht-B~pn>.?ߣ+~ڔitņG 5hPs-:Lm,Le8L d In  cڪhZ4ڢzENdbg#e#YW4!d..{+b s<,\ϕ12}vsg6w_7f/~z7LmFZu15\c Y]q>77+{g[l;xwn9JmAG~ogea"zo8~>zt,$~3~\cOÍ;} ¬QE< 2c#GIz%ۗu \ &ϵ@^Cp=57}FF;'r *]#eCoE1}!҂B-h)UNn!ygRLv ZNG,Wly+T,w6rgŨsf.hBTY@gTǂgvq=qήT5{(t =\x͋}" Eh sxiW8-tު0&߫K靧7ǰ3yq>DG {{=rEKK]ST '9jTêVS#9);6PD^zIH,1%+.+(L,U;RGwV`IӖkWrvܢkAD-c֭a vEN|{k7+mw|~Y(X A& %|8˱2kPqp$7 7Mno$m|341Hr!FY.w6Y: "Ȗ=ih/[>ڣ>5LzSAhϝ?W0*Wj6ћ P45I9B^ })t0?}ݑ h4߭녝Ѵ$Sf.h~5n~ݭLsU#5٤M;u{].68,LRB2͈CR?i 9rO R`(7Wuk=0]Hb3]55U@Lׇ=L#0d,N CMcW- ]zr6f4@..mEFىlCy>vi v&lXNgӁR6;KB}w>6䦛o*?c=f_f?hAOSŕ @ȎX|%}ѷoq;|8xO]%'y5IqufJgv0a<132Rf{݅ {M:՘V]dݪ,mW_ 9:, T&״4HDf3nLK~*3rZIi֩w@隭L{q\H,3"H0l*MHԞBRy|9Y<*!3Mk^ X#Jk4kuz*JwdZD ^ 75~W w3wZ&|ۇ?a>;_|-cn> XuLiƋx- 4n<8V2mՋv+s.`F|/>͆/; P,|QF&3՝Si6ZQW/#ix۽|zO; jn`,sG蝍y;=t%Ƙ &mkoym1mpӗ^tuǩg'>Y~_qM7Czlj_9q׮_{'snt>\fu#Җ@O 'GxVv9+p 6s caqQ1ऄMժXƃa^JOO;PSH|蹋,1|" Q~3A+bEƶ?\jU=hz..sx[7jzaU{T(W$Vv[սɍ=)NИ4O\uxIM+"di?=-*hdD!Ϸٔ4Ma)p敆BN33qnMpͦ~☩P_I\Jz%ߞO9“/3M}ݒEѺ {]͎*(\C~ы uЖfid8xLƵivg+W3:V=iF~rF%܇=ꉏX6f|gIu*U &#!We]A-ZN蠒A:ODPXHA'Ԭ Yu{κZ$Ff98OQ4$[ af˜Tfax-YOeyD!-JUVcOL+ +t65H!RN9s+3VΫk d4o3-8JC̦Xu{#٬qv5 +Mll/-d6`k‡xUb\!636X-0J({-I=s^3ySUSŌIqW*=Jd52V`c|(aA7]/keE(4׷,ympT={*Gc 4#$ΜL?鰥LP6k-d78P$` ѽ9WPeUT8}fuyUƍ!H>4e9Ҋ0XL- R XQ8k={cuݪ~%k\ϕ3式{>nPv50qL]4@qf&fwGbQUKKJV^qHAXڄ>ܰY y#,5 v9Q!1&rfRLԤꙮȖH)D:9FI&!æ kȊpݘ53TICuX*oeR>%R s$7 T \gt0*H@q^%YhEηj֗p$fD{Ldk\?c/!b_` b'33;~4'Nƺ8+JμF邝]W}[nVhf ;KȑVQ+[ԥ#=IIؙ- _lбi څg/{O\O7+ t50cO+lMjŅU*-YyUgTѣ ǑP 8$cHDhNV_,W-QBpyyjX+߅@|U~H]^CItC 5dP<ʯܚKD+r. '.zDʍ'!U}`-*Т: U2KCc t;L4Qeb'Rڊ7S nCeΣsOw?JF ~ w;n/R1 t}ƴ :Ә)ntMf{Rt]ƼL掠rZCơ< 3|h΋p ,Ђt5\>6=mLB4pcnwM%֖N%zfzfA7->s @"ʚ)*);Fpcڥֲ dj%, 5i8w3C 533es唈!.8u8LjXvo@j蝽nNfN‡\ ,zgw3Pb|ozG.nb.9O>xQ6M 4QIENDB`kDd,L  C A"`b:k V1X"(c+k nk V1X"(c+PNG  IHDR\7[sRGB pHYs.>jIDATx^?ɶw$D玍<&:@'|%</BJ&0f)̉%-iRUUݽlտUU>W}_o|C #pD @:K{ڬZ>&S^ړ#r=Qt/tͨT?w_[U5 ȕ{?~Ly^Ÿ/rB5=k`n<\O*:fhʸדvu>:PlGiP*b&A{]XG]Bij}306f{R)UDҫ&ϓ#zԪ-Zl7W*:*du:^Ai4g|۷oR4Ղ#F9#wN@ΰGyHGdAkt)< Qx]-Kߋ޳0v2PEGSv;0HO>Enrz7o$鲷Xg :9;I%Q9G>~m$s*y}z9m$CzثjoZ v>c/g"$h]{^ݸ®J訮#E~mGw޳G DV~*~"5a}N"PF9~,Wkow1oJw.[ݤ|Wݓ%mt-oSw5I<\]sES~&&*j͞BknyD :FyVsXǧsq=J/c}%۷): ˟-mdoڔ+q |Y]i&|}<#+Ve/YhH4?5/?ԾhQ䩆n} ߕ%~`F}׷5?]?QV%"Ww6PwRQ=*t޸ҼwKUUѱ:_|9n>Vn\U\,޿| ϶#bc:O'OUu.%NM;\, xN?qԥUfKP+kFQCF^U wvu=<=*TV--KOb/Uɹ+7yГs0$ɰ_-I|'u>;?PrOxyˎ]۲>aN"yo׵]ޒ:,+ V>2LVy?MX:uEѫ`_˷O_&f,be׌Rsi Wعc=߫;庨1yǧ;ztgPS[zu㮭gzuZiX)IR@d;1E'|r܁߄%qY!^p.NkboaY/xىA#&+k*~s|+HvkJ^}K–?.7L2_3* w8\ ;%o=jG3bQ΃xZ_gS:Y+5 {]߾F3j5rG?D-n͑LQNM,cӰ4|\t>C٧Wz#"WClXO4¢c=}~[ѣZ7@K{lnٜ}Xذc;[:rו\v(GIt5%M4ůnm(S\#cJ}WɄCr) '".(a(R#{Q֚1h鯳XIQO;h/n_g_G)e9Ŧ:$~k?7?6 n~gz8Wn,/ZIڢeNW k$=djIDb4c6NQsz.|Qn为mv3׳fv/XjCuD4JCѣ%j˩ñ.Uam,>h"$,APoJYVJOUzn[Ic>Ks@gީb-dhKtrדvu<:PlG>e?7$5`306fջ`>{^5Q~ѣVmZf-_Uy֬%yYuMHUu5_s!]g`zEgugأ<$ ߠA]NE]ͅ zT<ˮޥ}X6{Úd 9-/,<ϏwO|2|,|66j4KVQRU^eP-8$j2 t4gدs󽿴_TPukd*=GڣMt-ٯ7|aaA@B{ b @UD^W \'*^( M @ puu @f__IRaﳐUv1 Cʲ7kK_E[ZW/} @WF[C[NêO_ɹV(^3$ZaՂ-BZYZ ȓ  pZu4đwj/,t(@ c<{@uhבݪ*\HxǴa^einX^/-ݫGj[6jzTһHv\ @z N~ )W #IH&IY/ iůɅv-;X{^KW  ,:b\UHQX'{_}GNRJW%5^M5=y-|rJ,"l @C`yNrr]uLh]35N҃䓄,Iy-r)RV^"tv&5 BǏ'f;'7HMbYLpw1 Mjg[ǃP ŜZӥhT忕m܈lH  8u8\o !@tul>+V56n^U tk~] އuoYQ{c@NJ@ puu @f:ɞTl @`E:ɮQ @'%pP]˗>T @& TS5K{8 @5`U$2 @دC@5NyAϞ=[A#y>=Y*1\+s^ލED^a]n5i!tYw.m33gWMg@wtw%%|сo/>ʖJrnI2UrNhtk]$#Lnzg61o*=7Y@ׁCn ?c9Gr  %uO7rgwÒDבwQ=~/,y0lDQeS~W5)8'^/F"TWϡ^@@m^Mf̥:gx+`yF~O>ɳv:tYHALZ#fIh1`i*K[g|xwEB <}Toy 3)~Qg.'_Rm|,fwK2Hw .+EKE=&wB6`( ΑZ4/O0b'L@iHӞi-DնƋ 6?YȽ, HuSgq-)Rjy6  hҫfdU m+f,! 00 ٧mW.[m7ŭvy*_V G~v-ч,XNno\ίkYtE'n>,PG=\.H:홶;8z&ϥZ7>6ZZ -9$Ji7ig#l-'"/]t~-х) XB}bn A-`5Ze c݇:{vh$|uME!n7$W+$1hi+)Na'P3ԦZ  z"57Kh/$ş`zM[İکcװc7h ޤ'hdˆv?)u`|u] rg2C]z4|$:'Ŝt=xV;ZfMO~Dj_PG _gNOWFh/$%O& \N :m#װueeY=**tߓk)f`]tPש)t*I&MX8sRm.o/&6Zܦx$'f>j-yB`g~"..6lǚ_HEy=%&_Dh-dĠ~ٵxɩvlq'O k)I;ZZbSSrU$~69Wiyu`&,~.tcMRTp` U$qذB"\^XKIizb+ >Z`uSTInV)&6Lqg"ZU-' "4]\XzQSvĠ9o$9ݝ~- J"篼#I@c}2$.Jy=j`Dkӭ= M &EM=I&eL)Jꕟ.y~ Z;aGQ_ݓ]o-+'ƾU>(b)ҭlyB&uvnmrXѶO m!QZ\NmcN-dB%߷Z}f`1YddI \@[t>mBPDh31],/_׹_|^xD۞={9]ŋ/_|:R EKO>d~=;+sQϿGZ_/% a?'Hqϟ?\6ӊ?}6Rel H`(1D|>Fޢp &d3t倮E#y_&hXTğ'OD,Y@\\6]Go?O阯TtzcHl|9XZl:i5B` p6!d W!/RZ7-DXq-P,ix]#iw~]Y=ۈ pvAN g'!00 3nE7RZWhxX|ӧvgRl#ʂ r8'*<$Q0Ybq >_`{^b<{xF*U*nUɏ4%{ڜxk ޺EG` p9ȉڂ](ܴS|Wj(uD5OIXI-uE>H>n-:yB ]g9^w WuvzG"ݲ(/EJk1(acsw w϶,\+AN "k`0n]xz.=m3IQ eo\uAa4ST]!p#)Z -jo!erR{bϽo?vWV|}|jGrmƧQVq048[̫ :ب.B`އ:p)GNW].ҷQExn&Kݢ8 'Y‚hрѣGkBNcVj."` ?u- |uuM68 'YÂFMLt͇ٷ[̻J][gpgB]v*p@-UZY˞:Hk]! 'O~{tޒro.{( ]:6!0Oi ;Ny(&Nv9뽫5Bٮl~!\X$a>~簷l1-s^|==_}KWb`ҭ:h*E` p9ȉ?|sVʘp~eOuT,uu lB` pƦxOM{(^bucY݃u^ƠtxT[7pSӼe]g? Gã>XǤz=.vY'/^5{EJ '/GuR^[F:޸HO _$ދעy#Y-^T]Wldo"voyI"\ˑ^ݶ iY\ j#;Yd׫^\ױH% }ABRmtmNb*^̬aGЍ}t{ֿҳj)::c'w:뵫\yIcl+ E=QH~}wq,ˊxXlHFll 2b[zqb[$e]#0:zr*$/1EO$썋q p(*E҇u$ZAS'-k9'_YQ2u6޽ rAuGSjvyww<`KRٺDeZpPk|s֟PmoƄ Xs-I y[/ʑ~{fK)zif>P@īȸpl4ROjј|-%ES%Nj#=Mr^\[Oh ni5Izfu@oui2#Wpm-: .++z^ġGՆk/^c֡y9b-tea%c,UYJuOO svm|'kHN6~dm9>8#Zl`O󭣥GF/[nttQic l`#B{]6,L}%abc?{X*k_;bsAnsH XEu׋ 9t):**rWž_Ӱ 4:\CC#i-~Sjx\#fu +EOʓχeƲeK6Fk&-~9YErVJ6)'IgTm^ ?IY"Oy Xͩqmaf|uUp' _Dlh,,m>#v?'9Cg+1 v>JrkTl/;/ ensl$3楼]GWɩK'ܤ%Pl'3%We4Re~$ s]?}''ī y+8#}0.V Da: 9kdk\%]qZ:m%/v|,NƟȓ}^b-4"VI箜z}Xz^d] z%9/]c^k$[Xw2Zc}#x*\dzhMRzW^o-_z9炋|lgِxOO[i—E+JO.;YcL%3FF־۞F{M2>}ÇfڱH%EB͛7kRT?Rrۂpcugi` }ÇҡǏn A:j .ie^ٳg2 G4։(:f2Z3F5upb&O@sٴQ2t@rC=HXSsc@9lzO2\(.1\ 0C` p9ȉ_=rӫ\M]6\:Gj82Q0X Ar5qk9>CH``lAr: @X$0İɆ(@Pw@W] qW Bn~6ZS_@m  @b> & >_úSk@;!_>Iَ>m{Z-bK9hKL۫E @f :,sudrW'oIL&#e3jI@@wΰƠkEF]f^r& p;Tx9E ;5= uc"zG^zP?%eUD5I'_N^$ye `5!$`SYdKdy U> w\#+/òWe O?r=Uk9l"}~Δ_z%.S)O}< Waqf3mw2F#t#9G<}X:rAjߣF_,.*4LQ&iWL66IX';YyE8@ 36u\ tmR*Ԟsm[,jq^YA NOA\5]GUM~:^:ǼuBt84tk-eQVUɴf/ tL|#cca 5(I\_ bqI%W:&%ŅEbot!TJDOkvVy)*Ks%Gw[lUؐs{@3iuTo,ju^5W{-jmfrb;ѪÊ܇5pyL Zv8lˈ!v3'E`f,L:untUՄEtW5Nx& Ek~u<͛71 K|5|csC`x,vW_G{իWdâk ^ Q1;A*չZ]۷"@F@K:(:YuI r4tӍY\VA>K?V?㊦_xa:uHMQ({҃5ρ~0z1<9wD?G}hx[&s4n圏clu?w53wB`]"XwnZ/'"ZcZ>G"R|ϳz<$ɑ.zcE5\᧏|)/InMF*+wq I-]'Z>zܴKۈ|)>{{^vM}XFEbk.HZxF{qhF\`f;;%s]VK0,jטC,@wf,IױFm8y|gi2=,&ɏŦ{5ܟTyO1ϊ]RV|򜋡^꒵՘ -U'_s)'Q9Ş3O?wFO2Lj}XӮ-ye5흝.C-@1(> kJ;{gBܝtkԆSQ]7[ ,8)I-Y܂_s,B5b8]{s^t6j[~{ĥőU,ԛ11i^᮲,6(d*[;f{qs&?$ G'{}L4n?#6k_.+PSdu"6 \]U8oZjjoɃH "9l۴3z՛)!6lQ%M{~06,~c=D^XDBױE?=+oOm5NlGY>Zxm,."}q [_`QYCmQsO~j"ӌ{(Ժaej԰[j&} "((Lծ뒴a]S*TNdcyIOKԝόdtSbd% s~}/rń?֖n82ZYI 4${85Cr嫐W$Xl;Vt>oE")tO1bC7-҈Os|sNFm^lɸ؍>QΌ(4#&sح)E䂞h9Ǜ/\}zBzS-:^X 6\-a)Ծ#w|${ryϟ c#h]g2^˝lT@m~ I|i﹉S[zꏷEZ+:i{uDms| <~Xo9SyWr<쑚! d.rJi3nok>ʑ7o޼}nӉA n֩'*±lH ţGjs/c!8x"%/jPFzI^uT bB)ipxʑw޽zJC<ݎ_Xxo{2goqoRŋW?*X$6]g>$@^6tc-" lb&ϙZRJi"ɥ.!u" pnk.]?dVT:I)Êkl6ʅZ⼠R VjON5~;APn )nMԹt@+I4U#s.ORuN܏?1BM7Sc@`mfάՓS9CO^u A&$ik/Sc IUV̙z|a˰|ZV-ko˹*IYdR |MUSk3=f,]?]c(;b3 `a{0I] gl-vZ+碤V|jRS#mBEJA?3宯^\7kT ,d/}jio[Se@H >kͬeki(5fO{}H=]=QY3=f,]_Sok?-'a @|FjiOnLϟ ru|u>W=|OFj>oʶ;eMD㹵1L}!'fֲFd50-nS/w]G.cIe[>GGǢ-3|ߑ~V& n@|JfY+~kckٟ㵞3c!^NndZs^J !fֲl2$Օ3c!^NmEGswWɨ pU\WKC_OGw$s]܍?3n(ݎwZF)|$ݒݢd[uB!fֲy.~<@~i}zK3c!^VHJ];r&іYNkRn-%"5! gl-!+t3c!^&NLF^7a,o!u!k"fֲۑDގu}'O!ǟ>}*ow I2'#޽{Ut,!.! ͛7/^0ᱰJ$www'z^~c=:G!@99PwW{g|#rxn:;jerS'k؟5yŲci#tyT82l[ױX7(Xlmuǯ)ٸ=n̜YK ܝM(X{X{CuNopn:>V_jÚjRCٵ0V}XyQ 0{^<-Uӄ%3̜YKE3=f,}Xv2sf-[K؆Lϟ ry6-O2gֲ:׳^㙞?3nRx%X pNyif-[K+;m߱}ޱf`h)w^5u/X "fzX~'1h~ߓb>@/!k&O=n<9b[^8b_siS<$?3n:ͥ$ 6*sfݿu#]ċ3c!^&D9^FHT%ߦE.=nJY1x|w@蝅޲/<嗷eZ5#<'q O6ޫ^_ {r{k9簥>s^*#9jm>|4<5xa @ N >L5wik_rxHs|*Y4񩊍\bKRim`orόx:7~GF^W->ΞT؈@|}x̹Q.Тbp۴8>/w:?^(lݸ + _eis &y39Ps1gBuHtG=Ǝ\ ҰFvCx @|.YqS2TH6&K[af@$gJvr\̧iׯ?~GOeW\t́/_iS޺y۷oߕ\Ol1 É$ HO8a?Kϗs e-$6cqɓ'ybpZC܃Kk6rݻw^tF3mlۚxE@: 4yŲ"]3 U eNr荨"lH]猭%0uCtzԥ\>Ug[@~*:Ye=xVNyp8 =:O奣Ӵ nO]g{pt@! 6]ԮjKѻBM K)tZAԺ;iu#E^Rmv2^kC 5ݢF ?3宯oG^w@/}<)wCـ:jn$ZbyZl2od1a2>-ڴFos__F.WUH%]Z}| xUGJY{񧔛i6]Kф'/~Hx|h9Xd[<"٫snyϟ ru޿/ohe{t|,#,'3R2pw-vx_vs1*͍gzXS[\*lTC(2gֲKūHJv1? hlWat$q2/NbGL1 ycYZmq/w]' @̙6wx)?%'сǓ4J?yy5; Zg\ej`xj]0b>ub( J-&o/w]GC,r[>#G/f%s"wcʹ(i!WC>_%WM򰩭x>>mJ#* l01yl=93c!^VHJ];r\sr2^rk9 5nbe#6y|xǏ$sZL @e^VȅYUۍ#d86" =YocwR+uZc@^.:cQXHHY\@P1~ U*{uV'MO& .o @&  ^v:^ʏ(\YҴS_9/}1?g]t؊TO`7]Ǟc}Jqyx$ ?& qNFvw%eixbNo\񸢣:k*  tEK8^ItD#9eQ42Nj4I쒋=EU)' :q֍1Շ K`ԁ<uP̴P :9F~Xz֙=j 2G(~TzLƋ"^}):WWծX'pM.WnSuq'@h<_'xJ/buh~@؇}Xm77F9\7::]x@:)3KTʝJJz`@Un |ы/8[@:l+3ٳgEm&e\b4ñ'm1Zabc(5|1"" ?M~`[g'3uݑ7,nmx[' ue3"nmĶM?S+y6R\ُ5"sLjiX9Yo]"S>78'% 8]kSTՋ_]'i}tdd^EnΊش_fm4I9>}?gO5g'%¼ug՚l5ómYz>u}du nGO|3\=YRV̬v+&I>geX>H2I: ;l'Ԫ{G[-u-WoMXܾ7$I+Z:cOL -4˟sNRwA^ ZL1ŵU@3i|:VZ)\уzDM~|k{Rag;;zăVڂ0$ y]9w0.%>iheȹAF,I{{hnFoip琼5&p#u=k-f$Gum|r##lL:v5'|dt]D2HuMfV6٘ՋXn9t?^3|jZ;*k'zhEc],^?krTo>?2;*sH6ub k"0x(B$oS9NA9N[<tm<J`XP)4z6eh6^x]'GP2:v||3vF:frXLϟ r܇n7 7,u hu ]8m' |yH+MtںsuTv#?K$sX@٭)+@iC\}X+Y5ڄ~zo@`E([1#S k={L}a~yoWEۑ4 D<mMA !|OnC @dUu~ xhHbْ @ݚlWꦙo69C6"൜~IJ);ׯ{ >h}aL?_1[=iS :.*=Yn ^>yP2v5Q嗺|e ͩXrw"**:);ׯL/@y&!*C1bGI= \6 sut;׶f_Jm-T^ [ 5!- \=l~+}];7йb$I|O EV'% @A pj1d[GLBF$K`qr  uZCO;mF:VɊL:8 \@s9Oھxrf/$ՂZVy7-4!E1B/_{s^ Ǔ8N0(t:Ic?7!Dv5 @kxQWVծu⺋ufS0fhLZC"ձ0 lD@tt}{mWJ.hzi?5 *Cׁ @uNl8 @@Nf ) !@@  S@9e4 @A:APA pJ:l6 @ H]' 3@NI]͆  Aa@) 뜲p t ( @8%tS6NC $ @$sfi@u0 @uNl8 @@Nf ) !@@  S@9e4 @A:APA pJ:l6 @ H]' 3@NI]͆  Aa@) 뜲p t ( @8%tS6NC $ @$sfi@u0 @uNl8 @@Nf ) !@@  S@9e4 @A:APA pJ:l6 @ H]' 3@NI]͆  Aa@) 뜲p t ( @8%tS6NC $ @$sfi@u0 @uNl8 @@Nf ) !@@  S@9e4 @A:APA pJ:l6 @ H]' 3@NI]͆  Aa@) 뜲p t ( @8%tS6NC $ @$GPjVO>/q 3 :F.w<)Om IzX4,k.ab~#v5_ͮڧv猜m幈XH#CO+Ŋc'3>UŃ~M/l'Sl '5^ĥym;~ߧOZ[!1|\:Nwu2.Ftrљ8.,!!snE'bYXjEAxjp3>7$R[I¯ YdDy鋝j<A[|~-V\=t6q̽K K@0)ԟ\kK.Qg 9`ǘ~Zd#/;6BRuZ;]g("9oǧꂶXWV-%nhE]=Hd1# E'1))?mtbh8嶘Cn/Мs8D }FJyn4zX:6ߙlh1в]aO<mI#[o8O=%e n$ɵ6LFumOxTء⋞D  y$}|N;z$/]Ģ< 0Y;oo Vyh-f2o"&\ mvF`u'S-v$ 0CK׹Ǘ/_{իW38.%ta@|I`$yDE^~͛oc\bhGj!ѣJ#Wua "@sUIe pDyA׷:D!?fj@8 t>| Gt~')ꉓ &Qt>| ]6  @\QKtS63NCn@\Q˛ux랽Z'%-?gGUܫ=9oj;.H])@aqcs7s؂NO_8PTrhSk7>-u^>[L s+I>>HbH|<}uzo>v9V{^##R/bk@@AxgoW>[漊}B[ϛ|V (Ԃ;ج-6V"@5X|57?u \=.]_} +! ze2޽=AVA p=CsUyIWb!@ ծ{www>u>Ao@B׏?H pļZ"tc@Q3F-y @7K@߬[@~%}ehG$*c@6RbTC;'qB~_W%D0E߿/$ՊaoC K QJLs|Y/q@wzثp8Z+;*c@6u4P# H9uęgϞ&tVR+cWUܲ+O @ N WJG"qO"q]G 9,d0y'bC @6s*oӽ9u,; uL1=FꕞZdC(: @ ՔeD::IsL-k @F`S]q?::> VLű^$[{,v# @[H$^{yeuď@PpM6RW6[(lTto^K%Ĥ(A%&X[Nl72ߪ\8F @&XN}:gJ@bNz{@!0tG 4ӴnCvT ߿(/߁o̷b#ѣ9 d!@ &70hAnN%ě' @ &@2 @h`LNC  G=2m8 @.Ds㻿 koq9"s$Ov5|NlN&uWͮ<>٩%լaE-Dm (jg>HR{Du'O9m2;s^džQJcPEU:+9ļrhNgY0thb^̴_C\j1N9\9v궮6hqIAD@ 8XJxP){{|5Ya_Ursoھn] JNt-CVodGWI?=_7B-?4=B Г }ω aA zbZ)b xסdSսDLJtX? ҋhAmRGC H}mKEߚKEȽLs{=7Kn{Dª7@N[)&IϠ?DFuq@+Eg#4ex~sU|.2 Oi°HCS^a'0 {ބ*Ek0"69]Hqz.|PVya+Y·3 pt5 [T+ij_6Q4W;gBG;~l-^a!%bm{ubE5s.!=_uRkw0n>jʿ?0F,\gh(Y18ϷbuH^O`-r^m3,Zf]U]G1; P4b!@"p%{!@x #h(:iw@&@) 7kCQ=%9֎Z p8@) b@/2''o9}e~HQ;DiQkC+|Eop砆#DZ m=Q צDҿDY"i6r GөCyn!?%Emْ"p@g s=iB NF=EgQS¨޽"7 w`2u8NA&sWN&*Fh g`S077ea@!7J;1,D> AnaQl{,߲=BX J*n]7~W$v;lkt&_ѯ7q9^rM;o̅!KFXY[( ]]dbk)_;jC w`۝xS[ł!,?'?f.0ʗ{UWUziOyU Ax%~Ni%(@'sL^Iʴ |Vgx3Q{F_͋=E%?B 9'.!c}!@W L40yxH͞ޣEf*vh`)$# ;՘ 0(v yC&t@O`w y4yf@7 JxGS?/)]%\ ]K A@KJ@n X2 p;4ۥ @BV  ܎m`m~I>ұn @l3o5L^r"7"/@#ӗ%; F3jU{(v!δ C9J5TY Rסi_ T.hvsj ~cvH4؇{BoT^W>S۹{?}_7B-?4=BZQجLo^8FCEN-Q?=|6 C"Qd9ItE[B&R6ɹր93R^aj7B'G>UUVH~ IS?5,/O/!Vf{[r@ϷAjb0c A4uWE& *? Ǩlr,B-w!"U l#LTAunx]X %@#h(h78& Ni`ߥ-2֚QmO呄 pg4;o /2^9n1͍zQB~eop$zhSXL^۾t·cv~?rGlK;ʗ( _&"s|k{6P/Rѣ^D0j2cMݬ?n1K/+tb~m2qp14ovD j_XZ/1CI`;0:G&9x\mbto&p6F" @>Vi'fVȇ33HՍV8|<,m[vB@ WI4݌lzЯ.zah‘ZZm'kt&_i~z=jknyc.zRfP1@& &ߍJg:m}WNjzV,srne< pgs`fVrޢ^/ vd Ax~Ny$ @?G%ğK9Cx;H~;w<'ʊ!O`|}X# KtW @&OɺcG7z;5_ @&~#Z} k=Li3% |DkNu)o> ?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-.0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entrye F A 2)Data /WordDocumentdS]ObjectPoolgnA 2 A 2_1173504314, FnA 2nA 2Ole CompObjfObjInfo !"%*/258;>?@CFILORSTUX]`adgjmpstuvwxyz{|}~ FMicrosoft Equation 3.0 DS Equation Equation.39qB x 1 x 2 "x k FMicrosoft Equation 3.0 DS EqEquation Native ^_1173504340 FnA 2nA 2Ole CompObj fuation Equation.39qB y 1 y 2 "y k FMicrosoft Equation 3.0 DS Equation Equation.39qObjInfo Equation Native  ^_11724878346FnA 2nA 2Ole  CompObj fObjInfoEquation Native 6_1172488172FnA 2nA 2} x i FMicrosoft Equation 3.0 DS Equation Equation.39qp> y i FMicrosoft Equation 3.0 DS EqOle CompObjfObjInfoEquation Native 6_1173504404FnA 2nA 2Ole CompObjfObjInfouation Equation.39q-(i i=1,2,",k FMicrosoft Equation 3.0 DS Equation Equation.39qEquation Native I_1168348040FnA 2nA 2Ole CompObj fObjInfo!Equation Native  _11724877781'$FnA 2nA 2Ole #i0! P(Y 1 "Y k |X 1 "X k ) FMicrosoft Equation 3.0 DS Equation Equation.39q> X iCompObj#%$fObjInfo&&Equation Native '6_1172487807)FnA 2nA 2Ole (CompObj(*)fObjInfo++Equation Native ,6 FMicrosoft Equation 3.0 DS Equation Equation.39qh|t Y i FMicrosoft Equation 3.0 DS Equation Equation.39q_1172488247@.FA 2A 2Ole -CompObj-/.fObjInfo008(  Y 1 ,",Y k FMicrosoft Equation 3.0 DS Equation Equation.39q8 . X 1 ,",Equation Native 1T_11685379533FA 2A 2Ole 3CompObj244fObjInfo56Equation Native 7T_1168349389J"8FA 2A 2Ole 9X k FMicrosoft Equation 3.0 DS Equation Equation.39q9|# P(Y 1 "Y k |X 1 "X k )=P(Y 1 |X 1 )CompObj79:fObjInfo:<Equation Native =_1172488283=FA 2A 2"P(Y k |X k ) FMicrosoft Equation 3.0 DS Equation Equation.39q8(  Y 1 ,",Y kOle ACompObj<>BfObjInfo?DEquation Native ET_1172492735;EBFA 2A 2Ole GCompObjACHfObjInfoDJ FMicrosoft Equation 3.0 DS Equation Equation.39q8> X 1 ,",X k FMicrosoft Equation 3.0 DS Equation Equation.39qEquation Native KT_1173245608YGFA 2A 2Ole MCompObjFHNfObjInfoIPEquation Native Q_1168329221TLFA 2A 2Ole V P(Y 1 "Y k |X 1 "X k )=P(Y 1 |X 1 )"P(Y k |Y k"1 X k ) FMicrosoft Equation 3.0 DS EqCompObjKMWfObjInfoNYEquation Native Z)_1166456530QFA 2A 2uation Equation.39q 4! " FMicrosoft Equation 3.0 DS Equation Equation.39q5ePa wtf t =Ole [CompObjPR\fObjInfoS^Equation Native _w f tf tff " FMicrosoft Equation 3.0 DS Equation Equation.39q5]b< wdl  =w f dl ff "_1166456560O^VFA 2A 2Ole bCompObjUWcfObjInfoXeEquation Native fy_1173504242[FA 2A 2Ole hCompObjZ\if FMicrosoft Equation 3.0 DS Equation Equation.39qG > k 1 =1.8,b=0.75 FMicrosoft Equation 3.0 DS Eq      !"#$%&'()*+,-./0123456789:;<=>?@ACDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ObjInfo]kEquation Native lc_1166474091`FA 2A 2Ole nA? ?3"`?2K;kwvAmI `!eK;kwvAr@ |3xcdd``dd``baV d,FYzP1n:&B@?b = UXRY7&meabM-VK-WMc1sC VPZB8-NHq=`f̮<@ , **aFTn/(A(V%2qMw({gS> 2\J\ h0y{@qĤ\Y\ːC b> 1,u}bHDd hb  c $A? ?3"`?2ـ/MgP%YIun&L `!fـ/MgP%YIur @~ |4xcdd``dd``baV d,FYzP1n:&&6! KA?H1Zʞ ㆪaM,,He`HI? 01d++&1ܘ!+(|-͟NHq=`,b̮<@ Y@&TVQq > ZQR uN@e 2§=|(H&F0>?\ h0y{@qĤ\Y\˰dP"CD|b@3XR~Dd hb  c $A? ?3"`?2R.9X12ky>ϕ+.nN `!&.9X12ky>ϕ+@R|xcdd``> @c112BYL%bpu @c112BYL%bpusi#`g#71dBe%\`2M%I#RpeqIj.G @ ] U`n PtZDd t@?3b   c $A ? ?3"`? 2x cuv&mp @A~R `!xx cuv&mp @A  Fxcdd``dd``baV d,FYzP1n:&&v! KA?H1: ㆪaM,,He` 01d++&1ܘ!+(|-gLT T@:+a|=?|gd 2Bn&fT&ΟĄ* CClb7D(AKdG!.Q>8 Ma`ډ\L~m%41 Fh 0y{@iI)$5&\E.B ~`~zDd hJGb   c $A ? ?3"`? 2Zr }CT `!Zr }C* @|fxcdd`` @c112BYL%bpu?}2°Əaa$_{qLȆ ᣇ#Ei`L?lV?_Èwc*+]! %E \|" MU\Nh``j#RpeqIj.E.z0(қDd @h(b   c $A ? ?3"`? 2Qżha!yF -RW `!%żha!yF @ |xcdd``> @c112BYL%bpu @c112BYL%bpuϕ+.`[ `!&.9X12ky>ϕ+@R|xcdd``> @c112BYL%bpu @c112BYL%bpusi#`g#71dBe%\`2M%I#RpeqIj.G @ ] U`n PtZDd t@?3b  c $A ? ?3"`?2x cuv&mp @Ap_ `!xx cuv&mp @A  Fxcdd``dd``baV d,FYzP1n:&&v! KA?H1: ㆪaM,,He` 01d++&1ܘ!+(|-gLT T@:+a|=?|gd 2Bn&fT&ΟĄ* CClb7D(AKdG!.Q>8 Ma`ډ\L~m%41 Fh 0y{@iI)$5&\E.B ~`~JDd J5j!EC>  # A"`"jL@< &ήa @=jL@< &ή@C.RZx] |EWLf& HAy3 ((. @keePWr ?uQQ(*9,8e=A!ꞩ~1= N~wU}zt}_wqZp{! y& O%$({W"npIy(.Εu+LxHFz1e:+^9PaQfZvK?)ǻH~qØ߀#]G:.$đDוM8P=A}sL_#zBS g:Z7[WkB,"stMmFQ*=Nȧx;R[/IN;㏋>BVU㊷|Z<~?Pk|LG,[px \x+cg;_ jo\Ù9[y)H; !0Z6)M"̀r|fǛK6O =O?s?Kyy ΁ W{碒atcKAwQɁ.>x8Np '$ՃY^ô,&Ŷ'6}ӽL=YqP"xq<ƔzyTsƒax%Ϲä &+ݴN̔9Yl6 Gp.e c5WewxґhnoAnЀ B);19N q<Cr1O</vd z-b7;f߶9=o|*/.oO6T$?,RTWvej^zRW.^'w^ND|[&_+ϟ$l<ϦϷT\!f2]:;"tEDkxG?4?)#d#xg3vqʸ?ʽr<x72.+;iowb(YL*RGZ{zw:{32:?'gÄ%G:~m琥kׯOx=2 >,D^ gkOZ$p=Q2;_W o@>+u^tC,}p]x쩱Voʊ]LYid'dH "c2M'LIڊe bsxx 鲎.Fpj^ON+,Yre)H ΄N+GZgn)FOE'eشyq'P܃'uj<`W=B(7Iqʼ΅<~o㔘@^:*Pܝw#u\}cjn6 5KsƬ  wG\v> \,%sblbxa0O-'>y`>KDt2Fme0N1/qh1=0/˘rA^[|NkoC?AK#uLh1Qgds=\f uƱ>s}b>k1 0W8PXu̵QC ̗9yN\01~G!BZЙ֛:l }GdA}Gb^nVkv^>D}rTmz\|s5*^ךCm^5"|JNs1ٱl[0VU%1ॖT>W2LXedQ,HFoWȦm54YLS܍>;iS20B6Ml=xqX{}u]\ע 4\ f zAieڕƸeT3HnGFmi\?18q/h܂>%LIҝ[co-E{˫F|V1~1h7-ߓ3޲.[0ko9fo t[ܰ;ko-r\cgR/鱽JA#{KtA{+>sZs{A1LpM] P=-^s7ͮCfQ̙YZ9&h㣘K3x)Rtq]\ }x(Ik3-لbw;FLmю=nv}ds% *9o&1~Ybvuyз1ǡvW_r#6N!_pBc:<<3GM1 h1 =z0{3GvezuvDGf.L.[ܠG~scRi̓1 P̱A->kS)Zz9̨ 6b~<ఆ@&젃Yj7K{XPA̵b~ښ@mƹ2%*g-ne 1?|ɕ&۷%=N:Ĭ.´.<]\ s>yy`y-lA~M<^=HÄuEy|K{[M.s01 AyG v`p_ÝZ\& ~A_2 W =&9R!A֦E{&=4o 3fe Ɵye&R#E+] ]wE=VVtu0*d 0r\a౳ÐcVF+0g($g0s[_&;}: sG@0Ø`짘~-μc5ݶ( tFAmQvz I?,bNSL3jCkz%&9tFkރ=N+b9֨tF(fu@y=5dmHgTNXCqp\/ѱ=dZ5yd31?X|a l95\h3y-LٷIfl-6_\b#ԳSx_lCʹb~՚Qz@= )GB[ֿ^ͦdzG0ϣw JQqp(dC^S9[ljw :=H^9k *`$~4F;/q_1cIzW}<}u<\<\īOHȺ3I)wt'_B֕y]\\_PEzi;7oӃe|"$~ΌZRf6/=Nmo$K<~8bW!+-[i-+i|Or,7]EAgŮornf~cn>JI')7q"i|{.y ׋OrfV:'&ʎrN :E -se_":8b]ggź}+t+N-s?&l;÷*n %5'|z ĹZ\SDo!} ]P[?1Q7LAo/f/lGv> sxkH!4|>-[KYg*-M33nr&gL.8c=cu(Y*7 iW~ WMusa9}6v3DZwo|Q5u׭ԩg ivlr1e&neA4h7g3|`η ismAl>9կ{h{-:_3{xQw[™>QܙU1 Vo1;`e[n;X'Z)%c[4W"$Hv60+9lVc0qDD9yz#ٗ=@@\lm[bܹ>IJPUM/ y:vn̍/qGfY伲}ZйkPE{S)3ٰ5p(|fmsx\YW %u,K66c1^o_aВEym3߻ s%c28X2s~cAY=8cc~3yélӱ\x A%}{/95օ*t6c]Z>&]nʵ8.0W?f\ qbkb)ws>)?5v|Kւ tNOصsS} FI}5p/2;r}|ȇmeU~K;S9yR982?GDd HhGb  c $A? ?3"`?2V ڇZ1Xgm~ `!eV ڇZ1Xg@@"|3xcdd``ed``baV d,FYzP1n:&&V! KA?H1Z h7T obIFHeA*CL N`A $37X/\!(?71XAk޴,@u@@ڈc? Hgv%ļ?m`< 2!Wba {сۓᣇEyx~ClgZQRõ? vO%4)\h 0y{@iI)$5a7\E.Y`cTLDd $hlb  c $A? ?3"`?2[f;(>^Tr[ `!j[f;(>^T@HD |8xcdd``ed``baV d,FYzP1n:&lB@?b  ㆪaM,,He`HI? 01d++&1ܘ!+(|-•T T0g̮<@ [Y@&DT >= $b#F=:p{2B@0f5~&! %E \\j > |&>\Xsq,6F&&\& @ ]` bg!t?03WDd ` b  c $A? ?3"`?2/5aSWY  `!5aSWY@xڥK@KJj-BttSQ'AEK@뫅0PI͡S'urN#p|{! uRg!q[mB%j VC-8L qEjQDa8OY&#F gIA%O-d̬1b捱Lzn^tk/6{@B+Uܡsm|3|a,x/(GTY?KzudЯ]ۏ6?Tiʼ/C2O+_[:XEQǃ;e \()[QaBS3cRp1xT0 ["Я6^e^W6\pȭFk<~JpKJ=!g&,3qVfze `.]Kno8?ݍC9'l#!V6gfS$G/{՝GDd HhGb  c $A? ?3"`?2{y1.K.m `!e{y1.K.@@"|3xcdd``ed``baV d,FYzP1n:&&V! KA?H1Z h7T obIFHeA*CL N`A $37X/\!(?71XAk_4:]b m3b^?cd60dY +1@0n=p{IfÇ"<`v!63 Qp()Jdj]Qhק \v.4=Ĥ\Y\2C D,Āf~FLDd hub  c $A? ?3"`?2ԟZ.3O+6TrӇ `!jԟZ.3O+6T@ |8xcdd``ed``baV d,FYzP1n:&6! KA?H1Zl πqC0&dT20$ͤ `[YB2sSRsn\LKd*F\tfWBkGy , "*a n[f:p{2B@0Ƃ5~&! %E \lv$J.h,sSع8``N#RpeqIj.C"\E.Y`R[Dd b  c $A? ?3"`?2Qg:m^b- `!%g:m^bYxcdd``eb``baV d,FYzP1n:A! KA?H1Z MĒʂT`35;aR&br<??b dF\ǘ@Z5+ac ~ ׀P{OgBV_xY}-ZFkFzF/5w # +a|6T~+32.qvT~$.F.FfLȆ Qo ?#EudB<]x@H8_ Y=0p!e<ƆcA_˸1KaCpenR~DO+8>t+/sssm.fZk\׊DĻ$8% 炖"\e;T6121)W2ԡ"b> 3!\Dd b  c $A? ?3"`?2;7~9ͻ& `!7~9ͻ:@Rx=N=Qf/R萈 !!T [()n@D%%f3``^L*?D!"@Qp2ܤ%9, 'AeM1HCN|}/z6& {PY;= "jHXz 7FB}E̥?ǀ(X( ~?M6FiDd 0b  c $A? ?3"`?2нRb[یD `!нRb[یD (kUxcdd``bd``baV d,FYzP1n:N! KA?H1Z@zjx|K2B* R>t/ @2@penR~3n.n\(EUsi#, O+!5ejq. 12ps0~ba 5W?\( ܐbr<5ؽi>S(HK`|Vv<vGIHsjؑ]dby' ◤rA nj`!``S#RpeqIj.CfE.B fS}`Dd h0C?b  c $A? ?3"`?28Ɲxl5 =Fv `!~8Ɲxl5 =Fv@ kLxcdd``vbd``baV d,FYzP1n:f.! KA?H1Zqu2@0&dT20|_&,eBܤ3 1ܘ!+(|-̟vPHqdlfi}Z 1.6$pm#7wc3[ aDYd++&t'\F: *!|8NT~!;?vPԥ~l s8|f?%' _\8eq; +KRslfPdh-߅C,phDd TTD:b  c $A? ?3"`?2T^uwwCCĖx `!T^uwwCCĖx XJTxcdd``Ndd``baV d,FYzP1n:&&.! KA?H1Z UXRY7S?&,e`abM-VK-WMc1sC VPZ--NHq50frdn sr҃=A|8 ?`-fT@|CS{^FE &I9 \p\Fxx #n#Eo`g lS.hlsS8q-EX#RpeqIj.C8E. ,ĀGf 5Dd'J.R  S A"`b:5btqtYA85H n5btqtYA8PNG  IHDR`NJsRGB pHYs+4IDATx^Mr%)FIi'0zrBpdie*]p|D܈ @w&wv!@xKW @^^^^ßrnt- @L7f9!@`+Vd2 p&̸3k@؊rU8  I9?3"oN&@g@ό;  [@3ά!@`+Vd2 p&̸3k@؊rU8  I9?3"oN&@g@ό;  [@3ά!@`+pe\lQi0v7vsǼJ?H\SړS<(r>3:o?ڃjQ?,3Cyl;&:g~]Y6ȍ67,z5'7 (Ptg(?t.٩Uf 7fҟ:6r>G'#p3fָl nf%!cw`=x{4,7.Q0ȊIg^59PMQ\50'` ng57aNW6S۬ٽ NqGnyeFJ5x3;)MKtfݽy^j/e^0s.][܂nrV}Q_+x%; YpD} @ #`o_ڴw7 !@` m-$Œ#~1sA4keN, 悍sɤ$I0 keO{9tך<ވ A/U*c΍ɜ+Eq m sC ;1gx'痗 &-fZjjke}p wJ7W[ɵ`90\G ;7}0rx5-mFλ8 @[lE aa Ћrދ,v!@ P3 @YB@·f @@/y/؅  #C@ ^^d @F9 @ b 0r> 5Az@{. @`|j ""]@00 @E9E aa Ћrދ,v!@ P3 @YB@·f @@/y/؅  #C@ ^^d @F9 @ b 0r> 5AzM~~~~yy5v!5#V5B `k7rnW1k Yʇ`~?==w@ɫbۙٞ|\y.iK/BZ<-H lrV~L^/68+3uhR+y)V!9tU*"]FLO!v!:(Df Xc,YN-Ĭc8e/"[YfЂTȹ|5*닜q+WȵqK'U#?},"N'pزE~ᜯ0`9l9j9䚶Uaq͊둺IGNE΢,ɸSGb3,25l7)55s{y&2Iͣ(:L|8B ꃳ9 |cL+-բ>X!y+]e:ToE怨E|eb-ŃthS۳:^6_6Jr>"/1uH1it9k-3xԿ1. : o_P2y_y,XG˯MiE \k=Aq5}\Ou>>l-WjẍhnmbWr!t!&I# 2y'sا+dxf͋ ѧ̑mz܂A>Y.Y["rV-uQr4gZM׿jflo;ZEol-+xoy[WD)Нmv>9`L^}{~]Cank?79ˎzrޖQW~>=EW #e^掖M9AE/4:>Aق#Hq+-G-,͍Upy5+4k9zy6z:_}rbg$C&hW_˵\Eo%QCR-cg*EBStE伞a@WS6JOmMK)4l 'I\=F~u^!y%DNaU G|ƹ9peeep)E'xrEϓmޝq#6.'DDY߾ɽ4YrwTLQYRdbEwb-GыR೓NB|9"{ ZӶݑZE[SҼxҵ[1ë4^mEvfOuyshh˷ɟx]NyeV#t/S|ժ47_ 걖ZL\yzf%|.kgaC9J ]&]R^USY+Gx:Z\ky)#@ RGh~.Rѳ Nl;譒FRk6F*zd.r%F))L?ŸzDՕۍH v.9//2ݷ>RkJ v]cSgY]gi9Ey]Ye]Oɜ67z{{S2|4n;̖reODo Y*M઱mv%ʬazOrF`ܱSd 6{0C cJsep |,,޺9 I:s"lk7y2' $Zz^mk+ ga6ғPPIȮ KstLϺ%Q]˵#"C1W- vv׻E}in*y(rRxi.[)z WZ7\jlS;_Lo/ʳ8;{:tMK+2bg2Y2z[yB~< s9f-V4l;Ey7c>d,B-/+[La2(z@\(8rKY{gQ-w Yhx]J9{<~;ez'E>Yʄ]4s!f;`ӝF%|<^;'ķ+ͷ,˿XL!]s]TE>m0_Z)Ƙ~t葄<=d*ժЃpyų֔^nQazmJU{ǿRcnMT{"i*RDѝ5_[zQ?؊$$/ Wƺ*wL@wN㼛JG\7)%5BNi<;TH!g 'sY}*u~-zl!=rFڃF޴y~z^DȵKy?9tV79'@Α |Xyq^U߿_}{ovƫ;r#J7ir^wf{*eM- ,x֐gGxiU%R]uw#ƚ ňy2Kɹr}"t{Xs]6DrYu3b=vyI:=ΚK+R}l97\io,S#xMi^\l9_DYd8"~r,(9$)G(Qk8 :KVt/ {WkUs,kUȹIqZg ,2E\Q4;G0)еu#緐s]f£#ތE9oHF%iyc*9҄yG6iYÙ^sb ŏLE <>rT8:S=Vjﵤ¾i} v(:%9K(}Qk<ÒM)@@J@n[99?4M.ʎiC_ +drB90p>nk.%Op@ѻD%(rCK=1Il{Pan^rN\i:ZRv>{g->,;.D7y;"s9ZeՅs&5@onj ,8{%Ϭ7^s ݥcSg9u\㬽0 hހ/۞U*m40)5_Y 6@ "PY\\Go:yN8RrJDΫrݐyJU<@>&!Ȱ#Ҽ l/@EWoB˫k瑊SZ:S>Kk:/r~f g[[[֘^n(QtJ"eܦȵ[Q:yYv*#;]Io^qEGˋ#)zLPau\8GEpѨ_c +EGk29wmJoE>ZDlVчhy2 JGtGˏsI6Q)W}  ·]hyIھ@Os&.#hwJzy=Ã[V@?EsM4UJ]sw} |>Iz_,lVS7e~h Mv*ɐ_t@[ܰFG'bsE ۾+u{,(Ui>~ i!yC(rKf]2J-vaE_&8=a20Tm%]w/rZȹ5=.Zտ:iR([AE/y#U{PW{):EyDŽ7SF󑛨\q]Rv/rg{Uwiw]` T*z%j}>Nd^=Z2 _3rޏ+lȃ1xg"mQS/!mLuޛcҾx_P`OoffD2xsmF3K`ύcA{h]V{y,'\JIYlӵq!@Y^^^hlkZrı @;XvJq6wJ0@@ <ӆ  ;E@Cϴ!@`'Nd. p(3m@؉rS4  J9?4Lv"cd?wlmi3'`7ɜ&FHSRLqtQo~@' ~.@]a=$ܫߓfgNŷb8t|';`a!̆'s Vv&zPn؅z@{ \%"z`2"p_mTS>l:i5hg-6lr|eݎGY@|Xk5'B&d_Qr̉R-K޽6x8 \S;0NdU_]-3-+enǏ#{d, fse5'E8F$YiEƵ}MyRU6+f\Nip_HD|ӾxK?9nW۹' Шc1&\s$Q8jlߕl.Wj$$@Y=x8kr5/Wz83$VkN'%oI78HFQWM+: ,8!9qo :.ל&x'Q H9/L }ˆp>w8Zs`69ܢ|f.άqtIek,qYDzb4Ar\v '"W]SϠfNW]^Lz7Y=m#jȳE֜"IfvFssGgUOec֝I^\5]ϒy #}͊uƁ לH&/WAcDRKe풉q룄QmjDT&B]̰oWɜ, F0ƋLvcu9/fMG@` u1J\dnsB\i;`j $@F6:I @`T;D9@V#fj@@6< @@W@ @XrZD M9FF@j"? ly62:@V# d@γ ՟.|Y~FӱՍ 3Gt28&c"yẔAOýi lvZF7+ R_^,]n &CɖUh~O+x.hqX rrG#I?ν8ꮼ=ْQ]q@\*ݢv953Jkd 6L+w#p8W 6j2WѴ}wKmhI#y 'C1f@;xz(NZjl#˅r̉d27ݧ2f\]e)4bL3HDdyU%^# Z6'A5N>H#~7 Dsܯc&Bv3W"{ƣ8_ Wy%L0YO _nӒ̩ ʹorN:k&F}JҲ}}mLIK2a:ZΝNWV YrpLu%uIyC`Ypz.ބoa89{P?h?5^wsc+oCo;V`pRY*]4ܹK(xwUfȓ?7@+ _< s#WrVMa3=#'|H8İ]!:3۝xW׷!} }lE725BGəĸv8}[YʃUܱ3Sřuqv$Q?N܍e;Iu1~E:?Ygn.(` PF`A9?k,l ,N9_<@@H@ӌh@Xrxp & 8| @ M9O3 @`q=@@rfD @{ 4<͈  @iy- @@AVAֿ_<4Mc6kD9}u]Ժ0pflHř=e;s.R{,Y, -zg8X>fk߱Uy?e$#2丶o&Xt>OjkK!psy%7hs FY+B%s\6>{?u@7ЊגV[L$K+1`%XC9KgV-~ /щY&wYǒߨuFM抔tS+&_r5G{5퀳;˳dsioϥT$~挌` : Ő+scqhׄl׀5uSgžH1_;f,lAn;}p5 d\dFrɿ_a=þ N按`eq`z-*짖( pΉ,,FE`yƕ!Nr'S.Mۍs7ۧdBz@{P=˦ޜoQEgl+CP}o?o'f!A@Xw @[`}02 @8r~v= lA9"L&f@[@η# gǟC-$ @lg |02 @8y!N׼g6ci٫QkFoCW!4伈򺺜 _|eʀE/ۻ!I;O~M&Wk-g^drI!E %OH_Uhf+H{V?"Zr37ƅBj_RX 6ՙpsf{j7/rQt~Z01 }GFQ>1bȹ`dT!iovntVnOVν|;HΝݯםja&o#9@o+#hgKM wui"⟒wPAL̩;r5;7+`?p˿'{e4-.%NL͆~WʪS((Mj&LYr06q;Avv7BTs;@CY7sn@%3ȐP:|dd, ]3srgA9$C}$0rʌ- ,r,nt|˔bR7">yw* 0r>=8p yB xt A9_#x@ W+ @` q @@]!@k@׈^@*  @XrF PA9GW@5 y*U^F;V2%sMAdƋʝ^|#O?"J7=s^Lg a3BY$m(>?k#ɜDʳ*X]o!z"5AlϸM j.Lq-_vFZW|>,O8-$o,tn7!@$PwP9W'Ѻ^׿5(OM}f\ mܓ(_1 xn?;姫Ic?CrT5Q bw" 5ƯR{ѳ*|YOMFѲk,b_fʎk-x2~I-UucQli690.lBh;\^9E>5fMBe1R3toM~!_}Yg3n`%s,G2gΜ4u\]8;r"o1X Z 13h8qg\7#uuyccnERM_[6/eK"]AVռ&nL_Λ7.(^[#5~7qoeg95NVit; IMVN ƚ]-XYg|vqO>\N!=*kMLZN6̥o6k^wfgǷcþqѩqlzr 9Yѕ"#Y!jOfRq3ʹ!;|ìw?Gv|O"C;8;wL?QlI' 7A$LI@΁kuCr:uIrdUZ#v^#^ w\M@^}^қ~ɜbcFn.=€M@@ |8 p/ⅷ 䜴 ܞr~2@rN@nO9} @9' @'>L  @oB&@@@ !   @Vew\/KwὙB\4 pިVח+ &?"%q #f-̈4cd@K`[9/QCV%$Wz]`G~,g`"s}M]1mRuw|j*k7KB -1q^O5d-] @!a~uT;(za @ 煹`_$oq\B6 $'ui6|Q1 @VΓTXjޜ<!@`[9ס3)&]nt| AAnvxOrjc34تlv#Ӈ p &:lq9p %@@y "tD# 1Evck'S 9@=!d  ܞZw' 3EvF0K@@ކ#V @" @@9k6 @ 1!@hC6@H9!@m m8b 0iIENDB`$$If!vh5 5^ 5^ 5` #v #v^ #v` :V"l t6,555/ /  /  / T$$If!vh5 5^ 5^ 5` #v #v^ #v` :V"l t6,555/ / T$$If!vh5 5^ 5^ 5` #v #v^ #v` :V"l t6,555/  T$$If!vh55 5 #v#v :V"l t06,5P5R/ T$$If!vh55 5 #v#v :V"l t06,5P5R/ T$$If!vh55 5 #v#v :V"l t06,5P5RT$$If!vh55= 555#v#v= #v#v#v:V"l t0,55555/ T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555/ T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555/ T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555/ T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555T$$If!vh55= 555#v#v= #v#v#v:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh5; 5 5 #v; #v #v :V"l t06,55N5V/ T$$If!vh5; 5 5 #v; #v #v :V"l4 t06+,55N5V/ T$$If!vh5; 5 5 #v; #v #v :V"l4 t06+,55N5V/ T$$If!vh5; 5 5 #v; #v #v :V"l4 t06+,55N5V/ T$$If!vh5; 5 5 #v; #v #v :V"l4 t06+,55N5VT Dd <0  # A" Nf_X;  @= Nf_X;46]9 x[klTC 0B(ZƑ K!&7y<$\,HbYR6(u *84&nRm vJD\SH}Ĕ{3`Yts|ݹ3s;挱 )$lSX3 _X1rr>eCXk6|-^x.i V,4 b7{H?}4Nië TL)o2H  "QmüoѻXC~c)D]{vKM<}۳.H_NڃײQaLtYoX`v)1^g1#=1tt9 )BKW1X}.Uo| FVtliMUj~ i7=PϾcMC`1o:6+ʧSxRnc7Lƿ%ѷ0sW2"s:.2:1ƵS7Qy,R <]  _ϝ(>%[5!K˓.R;y! 94o#˺2T|>ܟgiY++g5Y_5NͿS4PǨ9-VsG$1%~{{Ajc,~.rNsH"fv%9t=KnG*fSr!6趰%7o)=S'#H!]M / _H–grLȳF_BZcy\6 -lpESo"QlmaK|J!x)(6趰%;V ϬFQ GH>@>B@o?E H BG-/8lķV·SH+–+ ᯀ@[Һ\:QlmaK|J!]H" Yȋ _ LWVL^-$o$|NO#B:Aw2e![WXu_94'N> cEJlCv!%c8>|;t E<- tq.л-K6!ړ9VHam.dlv*OdJ+5~\ˍi{7oAǼy?" ޏ< E)`Yy[5^Ix}[>9mc6εCZAR\a\sᷠT9 :nҗ~O~mSӧS9iT^zNLr3xN鳒@ܺGeWxo3o>ʶx<񞧲W =(6;ʿ,w~x1ޠ .Y ,=*سov q!'L>u9ʻt~.n#CwF.ժY4澟x,8'wp^݂C']^~W~] rOo.!kPٽk*S^6'W țx@`h9d~zNבzon soe9 ObЯߥSLVRCwXks4ʛO߯ '1{?;eQW'{8mNX&Q]捳),;Mkw:s}x;_ ݏSǗ&y;yXwjIߔ$kj=[ B,~dX?psZbVHY#7le[X{J>{WokwX+3Beeq|ysH2V}xZܥJLTŵFz\D{[P8GwAzDmfj]"+7";XJMbG9GDd("uL  C A"`b g7n )FћS ng7n )FћPNG  IHDRG fsRGB pHYsjrIDATx^흿%'Ӈw^ir{Mb} ߓjy (~&X=(~ s믿jx Jο4| @ʒ '@՞Xur p*Tʒ '@՞Xur p*Tʒ '@՞Xur p*Tʒ '}'#g@#18C<@*PU*A 0NUg@XJ% j @@Vq@8Tm! @`* @'3 BU[@Pqx UjT8 @`6 TmJ @V!0W~U%@x@ W5g P <AFPE P!1D @9$@y W[ @:P5x j+T @@/ Tm* @t8V Pb G@ P@  CU@XBtj:@+@V1@TM#^ @` U @!p  @U[ @:P5x j+T @@/ Tm* @t8V Pb G@ P@  CU@XBtj:@+@V1@TM#^ @` U @!p  @U[ @:P5x j+T @@/ Tm* @t8V Pb G@ P@  CU@XBtj:@+xl5hm= @`jn .!@`TmXB z&]Bث a@@6)aB&  lB@ڌf3>w8a >ƌ6A$.0IOjx!^Or8)dzb)p†@UEɼyg@g`HiV5V@s~Z<2##\@Urg *uOwk^V|r`[@!>eKa g(?;3Zأ:$ yLD.;:qY P$h '4I.Ϩ_" 3!0='&ܡJ~n&jlm$*ICe$$$$%~:rƹA~?ޒU. VJ#C`={5I?1;zf1^y=g%"]SxM”p$ fz)Pz=6;uwbgU_b/-5L&^qGn VuGlJ`i-{G Ͻ [D^)&caq[yE]6쐥k( Es[_كB^1rktNJ9cSxMh䫭f_R/!GUvg+C\kD/4de~gCٔ.+Lg {$swZI*{d-t$!V2ˤ ZwkP9)T{iB4qZ)ω ˄%^^=E>Г[MSxMW^=hR6>>'O 7=xB4%bɭpU.էV}7M!iZ%59G|v{劔=$M5ٗvyJ< wvWՌ$~UU5@`jBU FPVUcHZ;jS"E^J"Œ:K6 ]xjV5/sc)iJnDInzea [X[]P3&FJHpC"^.. @;@v1kxskM-> |/?(+'3?]ؗOܲCS@BjSn!݂N!@` Tm VB-&Ws_v% S@LY~FTttD[tj 0H`q7XC`L]*8$2}ʽWObpMW;6&50 e|mGG,v]5ka9rVAiQbt70 ,\J XH^K!C6U\).׍q̏GFW֫BL&J|#ӑuj`)_hu;FHd @tj  Ũ7rSh$/N)QQ"95XnMUrbXB@H@Y$37ʓͱɶB@vOAҥPƱ7Q5}a9BI?3hZ|r"_2,H&~C!M@Yպhjش4y/NAI_GH$#᪩I  `]ۨZp1$+Nrk u uG`*lARh:e* >oT5;ciV~ۤ,ɯ ヲƅzW):?[=%kZD>h@;UMq<3_ŸYGeI"luYZS*1h)VRl p{T-yϩUŒE)H?Wt4oG]kGMv@8Z[S/0ZAL@ -8053ǽ{Oc 3H) "=u2ma<SYU*?+rGfy%= rTJU 2xNUWgG2g3gOdUwm5ȂکĜdDLu^@I)a5|^E|̗1$m$mܬj u6 ]I̒;`dH;˺#&P;TjJA[yXr*+šY1CҚaE@nelp 5VvV" p^r no0닫WJjP-.Œj´ ZU`4"]L`^zÀ@;`jXi-2'H@8y'_#" )TMJ ;@X~P5)) @`}5"B@@դ j׈!@@JU'_#" )TMJ ;@X~f4 @JfRx 4@`a @@/A@T 8<'8ƒ c|CҎ($@` fҶbm1IO OC. ){- CU;$@_P5 CU;d@c!SK2 T1@@Ω%@ sjԒL @Uc @9PsjI& ƯML?>>Q粁 fȽ ث].D&@PqxlD0 '1:#]ǚ Tb+>Y![6P"P2wܛ@a \@`9U/ w4:MjU[c.t%\.*؄Y$]5i(9+u @Jk@&OrV91WlĸJv6ۿ*^ͨL"T`.UM GW.5"v{ՐA2߅hZ2u+@`#ZpdVR q& "ZtXNë#n4GRs^"x''_tD",3@` ezPF?BcS~Ugo[^V<ϴH XNժr8؆w݊F >ggRk ikօ E`o*\ۏrk}AGU疦 +I _-I/ZbIKj33olxL&Kۛ0M -jpIJKC(Q5UVWPZ.ZaC̶XV9w6u(FJ QRbBZ!g3 OC`#TX$ jdPrrrm1RHV=¢,r[gr  A`uD] p%UN ȾzH$KU۷vcm>d Kԫ0!4 T|U^JU;FFG d'u@`v\II W{pIx*o߾Կ~zjt  f j0!hp2f%I;6j6!vi߿bv؎ Ud$&  FÜas9D4 > C(ۥcCD @@;6TMiVNݱj[ C G7@@`7P BP%p؎ US8 !vlچ!h8fdži Aؓ;6TmG&8`džM -wlڶC!! U3( @`gPCF`6mDvܱj:‡ 0v;6Tmp7 cCqd@`2vl䱀{@Gedž1H@ߍ۷o5[Pf>V~߱|~~61 B̌)#6c<| cj3ܑ#]uӶ U u-K{dž5c@ǃɝYܱq_CX?]zol{lʪcQ /mg4&̽/M=1Y 0#4<N¥JI=i xx$vȂ\Hwf;<t ](yM.RoW>.U.ۏ?uuϧv+*Z8sO9{^p-8lr5h<2>{3x<%3ٵc^߹H_HOC|-n|mdI؆ecn/;tm ].)?\ =CZ A@ H'' d4{B/D 3:3xnIL޾ג\4bDF6?_L/y租~J`vf*AFvwxZrTį;`'+J@< V#̉g] [Arr%Ԃm()D$^~d @ O0H: x\("_p);NV2FW8 @6jR}àP¸:̍gjS+QZZ3*h=xu]䝞HC$ӈ ";1A'ud6/Lp%]5#0k^$DŽ%Wr릟EqUlѓJXXzAVAąr-7$4s P%KnT&ǤRUC\\,?w' ͮʘ4"0~[5=pƨtUJ0?6k( yKС+$8CpoLrÈwZN`W6jnvBWTGmDIuǼ邞KgskT[߃ւ)M˖(*tV_06y#*9j}hA/3|5J. )H*bUa5=sJPVaĠ99 mn۬G9ie&7>ٔq)*pewfIܧGKo(Du@ u:F+Irvc$crWoݱZ'h$ʼHK~]>c|-8 C-[f|v a;6RǾ.Z6{&q0|f  ϹU#QUqJBu;-#Wf? aժ:p ߤ0>+-%Q~rfEz~T&C/\I!@ؒuUkA Z]f$&)A,AݑKas}0c .܃eaxP ]_IHCo"_{ڋkTG=xB@B@LK`X7տA1E5=jL<<@pz=|<$YVhK޶؄(Kђ"L\A@D-1'}uIo4ں=zcguUo$` o >[krArǏ{WGI(8yyIENDB`VDd%TXXR  S A"`bUNerZ:qаU  nUNerZ:qаPNG  IHDR[!sRGB pHYs.>UVIDATx^m[\Ђ  $K ٥`I!kZM I!6: BSR'ihB7l ]p 8V b  bXA _sF6RVsoF:sZ}kcB+ !Pj;RJRn_>v/쐹oE|u.{)ANS|a2_<;xYҩ'${S_, Huqec僕y(eR흃'd/7%ӹiwB]vbۛk~DN+U{xw1$@ PS^AtH;k/|+ϗ*rmg.Wϫt>o䠵Hv o_0W&$ӑuvfy^:Т5$F BKo_j6 ,T*RԽ>M}ou7;vSon|kye )ur?uhv9m3dl?C=rsKD𭰯''7Z Sf9* V:L QR=\u q pVVjV0[/?n?.l;91,stmȮXY:|#EI {bOh7y韗ɧN^|ҁzɶ\= xTis0&z;WB~w 7ʎPW+rsw*=:JϦv;G˥{_z!K-?r/|a:0L{7IBssW=X !ξB"w #{k+ˏ,o~?Z9Bii 3x:kfKE%lͤssqXO[7q]*>B%n4.Ub7_L۷oK l幕̺;Nծt{Fs9g[X;W_O˵ߞEcNVs,6>c#˕[FD-a咪0 xic n;Xë"u ]y'E`91_=vO=19MY \^?l[Ǝ|#h91#Zn⁀㮃Knu?=|{ Or+ך5qxËKTjݑi`d).t?'kS~镧VvMꞦaYx*sr>ߞm&ed?I(|e-h3M(]od8wwe'7jεy쮅.["U~.F;/36fYxG51Zqgsԝ7׾{LtJ zԎ*"c,q*3gkߝ)%]9ݏLw{"homv7_;n0_~ޚpl5Eėn_@F6),y,bӌo^\T˧DmVwW]rL٩^<_|תT%tW4GrLyFdY[Z?\9ձ'#u9n⮔lXm Qa&̗Yډq >k^s"=w2,ġ']k :r[Wj:W[o^h;Ad;+?ztd{sbg'Ug-Ԯ8V0,ˮ]{Ӎ?a4}"oNw]/@n};J}vY׶vVo;7XtߝYrb z cpgǥPڍKn],^< F$=U?:z̺[zr @9c] v,:Sh F IP ukXO-VϾ#>Gt(BJSﭹ֋pZRiڟo] ?o)]^(A"*&9o.|$:r>wI%ӈC]vd0\yw;뫇I! ,0q^!ϵωpϛb!sH^70 T)Ws(`qQnIaQ"ds [okIl{v?#%mwo˞Sxh\Y|d-"^(6\~lYm7E|h1ۗ? _tCRE?I mV3r&)^$$ysOg_?rݏ@âum3wW$_\KZūޕ;n%Rƙ#|XxY"1zsի.^|r(&gr^"~, :؊y 1mu=vOO.Dˍ_#VS%m\Ô+Qn.W("$^ǐ $ݹŅƗ7z]}e8q炞20lc,d[n;m ?dvtN [:׾>M'jњ[Z+;u?kXFcH "8e$yD9D=z;ֿ76$zj/(5d ^[P]-b=+67:5 {Eworm<[wBYwɓ[9j{GsCUZ[R{kNC"*,Ko"_ 9\{e"d?/a: :E^YĢ `QvO:sN ;잀| t}>Gv=˘v!w[~_ԫwAl֟Eቷ1J_>v*1[A[ >Ը&3ep|N^ء !Hj!NY7W3Ӯ,En.+AGo"eg[/ĩ|bJTsr)l8HƝBEL|"0%ʈq'X/z ʇ~4\~jvf{ae衕ӝpBwPy7ۗpaUn|e$wjb ?_|L[| OV#{RXw'\wT@%oV5L|GG8ϫ2mAjUl>{z'ŘGo|@4_6Z.EOiSO~c9ooNj}l/lnn{U+Jm>Z!?̅HU[aS+a ɼTttNG|5zSLtƛ|iS8߷ڶ;^v/ԟq'V+ _ΪODz?gu^,/YeqC~sYh8>n4޺PK},-+KǸ@%A|1drlW)HذeAyR19Z,1z%u*6:RT3"lgfk4_tUM_Eɽ7|oCՃ+bZuբvg]눊zɺ7ZO!O|UF9٭91׍հHĠ=˝OWb>ZbL;[ߞ(,"BrwzX^@ρl+b~ Vxao,G CsCkrݩpwkGGDî\ӹbSD|V#7(Ԥ]T#m7= {]z]ꮎGV5qsum+///2G 3WN:UTԶo,>^NW/^(~JoGa.EM~ZI8cLb"_aXԏ#1+]sZݍ;x6}s>w_[O?Sq嵙v^]_^-/}"1 Wr1.%l{5&XZFW0Dޭ_*b6_VKuW' ?c|p\o+yirNih|z>0`, WB0v^xySYjYov! &nĿbs?ocjS;.vb/.UT.&N$ZM #<~H {lqpvZ=*ϯ`Zl%ĵzc abŵpy:Q:k{)Yi< v7[ywg_I5 [a-w+ [R˽>nNصjmSJ\1!aUP>Jׯ~vJvbsX,RI&b[ ﾿Ę`3 6帯\O-7ѭm9:be<aqZh>'wba-gjGlÆe.>t-:+r:7%IJ)٬x})>Jݚ:LĐ`+5ݑ^DkbNɸ>Ӌ/nj^VW.)Nm(H,4cT\Iy/ﴚh b*pZ/Rt(AhZM/_Q t ~"v l7[ aW7vCb[MJ@Ϸh_)޿%dnn/cWuM߈4.)~a|Co㷧"+7~WΙӽrW<woKb`o|FӕB楷\6oc0AliH`ܹs'l_rk}ɤfSE`eF tO1_.cY\>+`Gq&.AqI=@z4WBko6bƓ<2#~zX[XvD!tk.=w{6 _ϖ2 O/-.]oNt%} b}"-"Z|3/0cGWo/n@ks W8?ׯ^XIЃs HS\\Weޔ.wD3l(n&Hw}U-/s^-qٹJ^Ѿ`ls;8xKli+6 '-W/_0^;)IJQO^^[DzŁ"bZ+֎,;]k`c g_^zg~zrD~V_yjE͖ ,hA7̫U[px0^8p<;..te˽@ }1 bƆWYp'mS+Ok!2s}F@[)vU&T}۷ϟ#j=`w HXň2.Ts PHqfp*6,_9sk 0,qN)usS1I/%0/K/)L(ʏ_|ܱ͵ʯv%FHM`G,>W/P!Ѻ_+ Ona7Ml}930޲KUڵ C.uār$b._ۅEaBH*qdBjכoaM06T}b=;.M9w|RAVss#X+5^֕MH2´\)R?RV}ʎSq~f]Eb+K߾*]%}%ҴڝO-JX@w};B0lgfWV/-W\Y]ܘ-U\R&4Ͻ|6 ĻC.v|1op`UZ8y^{k!F_Cl9P-#"X6]̲Ƃc2U.3X +xƍF@gXh|R͛RVk"21*,aμX*P-[8UDI ,El՛YLv^- V;'p|@P(a*Aw!Sb:gѸќnbbPZ^Bd,plx;Ok̓I E) Bt?Ci~15Vm1K]T~ϠOr_yPu~EψJnaALš? FvtSBM#*4*46>Xwr;%$N,ٝ#fk#^LCTAJ )/-83xܑ[^tw{B%z}M%6ut;׊8ܵ ǃht ˏ,—)FJJ<5rM?o5.b`gAJߜ/DN3D+f+;QoǕvSvTT?/?b&ݑq/cn1 Ue=Ze g@%Lr ˜JVYsn6;_t\{{qƼʺxp#Dd|(/`.]|hQucILc޸- QevGQixW^Fow*)2* bdX %ffar=@V \\Q[1JtAӯdTfq~z$ C[d5rҽUJ8ֲ'$q0cDy P_7SYв#ZJr`>YDUZ DdLnWڂC%6ڔ[TLn˟bC Gg3'Czꑦ ș"u#\X+q;>UƧ]N̷΢{r&a'0a]o0&GH4`c" D+aAHL@c&X5/wރb;7ۍfhz>rtT/L{v*"&㑂_*'$0ađFH`^틘WF'0P*/5;"B1Vth =&ˎkɸ/lPb 0^}O:NFzG/q^]Y@$8  Ly 6YFXQr;DZ2ms|m * %! X}ku5Sb$Hkdq|-w]vE@̖괃OWY㕆b7s#  fl PGfIHH  $Qb婢9G$@$@$0(CaIHH`llqت +% T`aIHI z=1g}0/  Y# ½fY$@$@c#`;ͳV̈HH`Xij%   3"  Y#@%gyIHEJ7$@$@FJGVMyCx~7Cbd[۳Jiu' e'FAדTF{7#/1&z T^;"[xmZp,[eSo%zv܃~q!j3O*!Oz.?򙉃PS6 78$z8U;C3LEP:n"!NP?S=- rS /ouWHԉ}Kb% !e$b Y!ʌˇ@nqp2Tdע~iAc+$X'2ihi ~bľU>}]| ԍ2D5M9+iPaoɫsq͵iT.zB~p7 hTM ^-) ߗU8|oqBgg魻R qMY [ˑt<X% DP6ڤT{Lzo(b->7}E%R\3A KmI# x}b5atbt慪 Q+!o 8[ueH2n9Fd s];U7%%6Λ[qFX2''n6T](~T&1~A9z?D,Vb_k+e *{ks.wL"!h%56K: khu1mZ*Ş,|zԃ} w8euA1oһRO u6J<%.y=(>QMoJ~&tCO0)l=N)/i:#tzU<ϢBpFD;jS-x꫘$ H5z O}$@$@&@%Nu \ gka >!YI` 58,X$ljqKUj)rK-_75b;wf^R-'dIOȵR |b$諸CZhFW|팎cѱe5!^oۉ8PBu%Վ^ě\beb}a%#MF].Mn=9DT!L))oMM}%6E|4He&ݸC_Ob}du:aɃS!U,Id%F."-0APP #AmT|c([ay98nGňRу\;588.4C0.(։1H6an wU+'i$B3gl^yDc h“E(}D ~Co5,'Au_{% L"ۃ/:6]ЯOu\ysOx.afRSTI}d|H2go!KՔJ-Eȝ.ȱ|W˰^}f#Iľx.})i;JɆK77bxg&IbiбIH*H"įbL<* bb# }C7o0GvW~dt7" bzl\e(/!`q{%eSw&:K~ @J<3"e49Ɩh8`VmJ  CasHϜ9cTU! 9gY`  T: \$@$@"@%NUu  #@%*gg f1x *E  0*q+Aq0 d8FIe͂2J;{d6 " H ٭;zN>bx,H s;*$O̖A%8GIHc)HHH 8M$@$0%SR, @F P3ZqtHH`JP"Y   g6 OIE$@$@%@%hm  )!GSIf@* Q(%%䡃 H@x )1Y vDclH +xqVj~N )+ B$EG$0|T3E4)&  *1 Lx7 PH`HC"LAJdE;{ $L޹s'2}x$L e\Wc]QL#%xxi<D4<*Yx-%fj㼄'A6J *qJ*ndVQe_?kV|^\_ ÛiUzzYBL$0*`x48 VGH鍹ˇݭ5}qI]eVaz$3bbǀŤIW`8;+6!d.=ĆYԻEF+īg'!jL}n[\幽KʇJF^晌scyBZNR) 1W}5*q%RcSe4b;=(A^O$@$@`L<=^KC&{t? d81'v%{  )>F$@$yJ,g#fY  Ꝧ , UL> L *T$A$@$QTV& T)HHH 8M$@$0%&{הd1HHH ULR/ܹgO:k0K7;|-8nyHHbĺJ'UR-5_ Afc%fc"  J (fx6.K mdX[J= UbgRK=4B>pkȦ'<ĂzPק\$@$@ */_Rz]^ڌ yX6   !2<~`enFÑM7G"b  x(}eX |;a90;  )&ě{6e@7 Xx#7*=cBYURVC.By|#{˕AgF*X4 'ASzB$@$0;c pjXJJWIHSS%, Lxj% TGHH Y9M$@$0Q, @V PZsHH`:PY   g7 tq:j  HxJ L+%f(<R wLa%  "`CompObj_bofObjInfoqOlePres000acr\Equation Native .uation Equation.39q!  .   & & MathType-&&-    Symbol" ؟ww waf-2 ySymbolm, ̴؟ww waf-2 y2 m+y2 +y2  -y2 z+y2 =y Times New Roman؟ww waf-2 ty2 tyTimes New Roman؟ww waf-2 ny2 Ny 2 wtfe 2 avwdl 2 wdld2 bd2  bd2  kd2 zkd 2 z wtfd2 AFt2 FBMTimes New Roman؟ww waf-2 )M 2 log(2 )o2  )o2  1o2 B ((2 z)(2 z1(2 z((2 25 Times New Roman؟ww waf-2  152 D15 & "Systemaf !-NANI5Ht  BM25F=wtf t (k 1 +1)k 1 ((1"b)+bwdlavwdl)+wtflog(Nn) t "Oh+'0  , L X d p|HHKJ<^̍HHvE $ pxL#s!c"lHH #XIQ *tprF P7Loާ!=ގ1G  JJ~F4FA¬+UR KrA ʲol$@$@3B`[Juz6j|{;|*;XT #2 /2lrO7 hiHH`R PxR|dž ZhIJFdĦP*#A2qiHHL`$J<ɰqZ:ch([I,VFv4a^B$@)'0%A^K^J<4ÌV=& HX騧hh6lWΆT2u+x#&?/r^.{=Rf(<3% 1wz<-q>=$(TKP"^B$@$K`8LUǯt 1M & i"014"B$@$]g9 `L Ti Vڷ0tM=NP%nH2F 1.H!Q'P{7em7 C. zUo(nS7Y.&  'U%ZX|jq\CIrXE}ZYTG$@$0S2ĉe8ˑ3In9"4K$0#ģȈyBג r%Ϩo`Խh WX6#_#r_lo`(!IksDY  Kً7lJ-lOd|7H,Bz&q2YH;OT^We)[ HH`x f GgYR L^L tHH, 0&d$@$@$0GFIHH)(Vǐe2 @ dwXnG1WBOn% xM)W|[ĹǪ! XpLҲSػCCw_LCbxjj! KVMvqCJFeP~囯e5Kb6ZH|Ck#}` SA17tSl^xf!Y6&Be٦x0rkpNVH`dXU,kqkTFo!*4hL{ ;Fo雑񡯟JU_64$@$@8(RҒ̷G׫оRhO썆g܃suJS$@%a%փ*_ ̡SH`|h)6A[CnArz @c7wjAc &bvT2|QPPxA $@"Y,J?yU _=$ `L&  *1 Lx7 PHHH`qmy @xhHH&I JPxMy w*N#  2,# @z P[7HH`PgYF  8qzflٳy唻d=F]0G=7V(fቲtW+ZB(]hi|5[!{,Fb>XpUx0=D4E$`C +0~^}%6^-Wq/HC 枠솢xdnx'k!/FCBJےG?`y}Ji])Ã0$0189rj"ŋid cG¦=)|,ә;fbCbXɭdP̂ 1XE!!fҺe6W7 |@f(iRO!z O 1Vn)cUԋ?"ּzo1u#X| W>}n| )o,1t,qa6%QȶZF}Q ,3v!' %Y!!qƖQb PL#$@ P@%=zxzD2mI$iTLW' <ę/1 @$@$@i"@%NSm  #@%:gIHDJڠ/$@$@GJ<{u 8MA_HHfx%& HxJ-TwHH`Si(1@$@$@i"` TeHH`p˩NHH sb̕ @VPRSHH`: $Wb7{6*Bybn %GCw6eAZht'i7兑^M$A{V8 8` Q K*,JĒda$b;lGd[oهuZϰ, .ɚtUipOF:<9*lr(n22,݁JĉbJ6 ĸ?h%)>=FOH~D=#]b!tb̝UyƳ7DT"xtL>Rްld7&Gkm^^b2Y[^\"? A z7JVI6W;T0lUVe7YdA|ቷ{^z`PIh~}÷ UmO o6Z޸"[j o8mo ? B~AUoZ*:?m-{>?c@YFt˯TM ^pǼ׆'"x2,x#4W3^Iܲ(?odko$ֲ{z 2{Q}o%[*mx: .2٢so&GHtp6;ǺQ2ߤ2+FD2bYVA%q`Dcp&qHV^5 xwҟCnSJōP2K2;#WbqƃXidɲcTbвzB\ y5%FFZ}b}A,ۘIc5H!joX>Ǻ;@TAbe+20W7)#@3D9Y/iG f|!$M=>L2|U1y?`^0n|ی6HҔtzޒrkBLvtN<ᛯ5˶Q_-.z{Z/T۝{Eo{ &?H!׽Bm$i+B<}jjR$!>{֨\#:+͉~>("0QBf#muAK`zY!I8^H @HHH Rbc(;K{X9єKvXruiH&Kw:|&X1IelN"rO歚'79B*fgYS3QxB$@$0jV1- M͌gMx`IF*& PI}*T/qIH4I"EBڻBxsfa[%z2OnV0^ze^kȾ V~{UycqP`+bF$@$0+%V7SmcJ|o}Rdhwɼnӷa@z T FJrj=KF RH ˷r-MZ i FD!>dA)}M_bSGOt#J;%ȼDkK>KEfJ "[%Y* @&)لb}M13Prȿt (X)q{abS(|zlZp, <'yg6   X)ꗖ]޻q'I/Q,t Օ7fL`YIŷ4Q^߼ 8qHk<|ze0C 팭q@,+JL3U,, PXCiF#鉻GHHf@b٬'HH`Z PLs d$@q ĈL{#<#n P=U!{qc. ģp|Eަ2Ҿ8${2>Fl*Zզ&ÅN÷ރڶM1\/i)%_nZ$@b(ńH(ܽ85oo*6!q6[dP~(oc 1$!2ˎ/ yfٰ}4>ībdb @ % F% aA'}U6ßyDl"ʺ36i"Z]N36g>9PiEr 'ݿ,kAĕe(HDd i|{^"AZ 1=8=u hrɈ"?/LlQzw> )ecuJO1& uĦ[AhSYAա?!֯M-\|fh31xkײoӲ?Si Tu1qHWP\JP̎ql !QM !4?{[̊D+zj E 1q201 d@br$@$@$87^E$@$@!@%GZ!  dɸ*  l^`rIENDB`$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 / T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 / T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 / T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 / T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh5 5 5 5 #v #v #v #v :V"l t06,5 5 5 5 T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l t0,55555/ T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l t0,55555/ / T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l4 t0+,55555/ T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l4 t0+,55555T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l t0,55555/ T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l t0,55555/ / T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l4 t0+,55555/ T$$If!vh55w 5b5d5D#v#vw #vb#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l t0,55555/ / T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555/ T$$If!vh55 55d5D#v#v #v#vd#vD:V"l4 t0+,55555Tl Dd Q5i$>   # A"`"e7V51 @=e7V5<3=5x?QZ|x[klTUݽb(Җ b-BL$j1M3.(?Ih|EMH"! nP1:s9s{¶ndo9s=U`RB$BWV2=ׯ@>T"༾ 0J8ީae(#5"yJF؞Nxר6-H﬌73c۳QleTKnK2< D{tt*VoH ~c a$}ҠB^8H *2pNvĞ;Y Nw~k@al?0]ϯ#;DFN~񱲃냭6& v`s`g~Zokڶo pڸø_L3Ս{Ģ=W2Uƹr/ B%PvδUJ2PM3W)D]=IIr>ca|gWdK;/`(CՋIFzt >G| 'N}w1'K;N8qsĩT'N}SLp\[sPlme?.>l~>šnă?t29xO/#GUB&AKmuk^auiAOι"9!:y~P=xSaiA(gV%l䝎J}[E9ŢM?n ۣuz:F3G@' fc ʙD ɿ0s2YL(gvZ+.w#jQv&US5X"ʙ{}x_}, TQ$R_A}xQE5!ʙSK)Ѿ(g-:h Dh2%r&81|,HwD9|J}߉~b\:?iQ\bѡx">r3_:Tu(HK8#Y/qG^8dq^z>/\_d͉SJԉS7ǝ89V‰S*'N}=7YRt5d^#dGFb4vjkB^=HI[H)Sbz^KfÒ['9"@HslJ8琘bBn m-Pm8@Xذ8j*굁0=`~9VD6Ɔ^lj^CWxU@X`(>kaz(^&=Z8N0s+IԚ >mXu-gA;Ĵstгa_b^K_k}`= :뛀n1O]Zԁr,-ƥ//04zH\_>\nzX9Ze+/eדU+窿8.)^_q\S"\y?.!0f礜~X\ %m%%Vސ{v}%)g讉9.kT^>| (zt2uv˱U;K[. ]ޛ`w`5L-L VNgP?!ώs]%%65zzkef!^j0[8[y:[ynN"acˇu{‡ck*rg*x+uf&&bvL>݊.'& kw~xwc6y'Eۿ?Աɦ);o%Y.t2,Db1VaFsɯr=sֲ)7KtŎ ;vzG'$VeD/D/w,{=bؿy7. ]gJ =uJ& .)[0WQӏ{gfnvn@-heYCڧ?彬aՉ6Î/E qfI;ZƗJWb\mBvLpg&׈h)| H"<7\I,;}VPZgb|h=棍>,Ǵߓj|7D҅~\䷮> n q)?rY!f#yKM OS=*xfAy \Dd 8b   c $A? ?3"`?2/lP^^-R4Qp?`!z/lP^^-R4Q p/~ Hxڥ?hQ]sFiJ CI )\--ELGՊSh8ġ I38:MP'(6s`=vP|blTYi: uLʽTKS'/5 P >^2iJjPktזuIݟ,Zģ5KD<֟[hM0V^i!-Z~N C_OE0 6HϪNl8-;wvl9oy.P^㜹C tx޸xkgxk7vC< _[Eu`4+!u:U'Υ-oY|\h[ԩ:hx]e2Ub`oҘwW=<姥qw TZ*骹mzW"37/j-䭌IQQ*~ }L> Shv.Uw\Q&3N 񂉕˪ 8˹g!}Dn~'yu؂Myi@ )GFkA8ws0:ecvc1TableSummaryInformation(fDocumentSummaryInformation8CompObjq*F@F Normal $Pa$CJ_HmH sH tH V@V Heading 1$$ & F(@&a$ 5CJKH8@8 Heading 2  & F@&B@!B Heading 3  & F@& 56CJ818 Heading 4  & F@&^^ Heading 5*$ & F(@&^`a$6CJTT Heading 6 & F<@&6CJOJQJkHLL Heading 7 & F<@& OJQJkHPP Heading 8 & F<@&6OJQJkHT T Heading 9 & F<@&6CJOJQJkHDA@D Default Paragraph FontVi@V  Table Normal :V 44 la (k(No List P&@P Footnote ReferenceCJH*OJQJkH>O> Author$a$CJOJQJkHPOP Paper-Title $xa$5CJ$OJQJkHFO"F Affiliations$a$ CJOJQJJ@2J  Footnote Textp^`pCJB Bulletp & Fp>Th^`p4 @R4 Footer  !,Ob, E-Mail<LrL Abstract$ & Fx@& a$5CJ; List Number 3p & F8>Th.^8`PP Captions$H#$+Dp/0$a$5CJ>O> References$ & Fa$.)@. Page NumberHC@H Body Text Indenth`hRYR  Document Map-D M OJQJ^JB"@B Caption$a$5\^JaJtH TBT Body Text#/H&#$+D@/0$CJ6U6 Hyperlink >*B*ph44 Header ! !t@#t C Table Grid7:V"0 "$Pa$828  Balloon Text#aJB'AB wComment ReferenceCJaJ:R: w Comment Text%$a$@jQR@ wComment Subject&5\>O!r> 2=h 2 + kMR: 10 x'x>+> a Endnote Text ($G$a$>*> aEndnote ReferenceH* eUXeeٵeZ4ZiZZdepCZpc|}$ t u U Exf3!l h!##;&'2'i'R(8***M+,I.K...1345l67<:i:j:l:::<<<=='=/==@@@@BCPEWEEsGHHHHFIIILLLOOQfQoQ"R2RRT5UU*VV*WWXiY}YY[\\]I^c^^abbLcecfcgcc d,dXddef3g5g_g`ggh$i,i1i4i;iBiCiHiNiTiZi[ifilirixiyij6jQjqjjjjk kkkkkk+k-k/k0k:kkkmm=nlnnnnnnnnnnnnnoo ooooo"o(o.o/o0o7o=oCoIoJoUoXo^odojokolosoyoooooooooooo2plqqqqqqqqqqqqqqqqqrrrrrrr%r&r'r^r_r`rjrqrtrur{rrrrrrrrrrrrrrrrrrr tfuuuuuuuuuuuuuuuvvvv1v:v;vCėwj{џܟ} C1Ef u_9TW1'e1ȴf0000000000000000000000000000000  00 0 0 0 0 0 0 0 0 0 0 0 0   0 000000 00V#0V# 00&0&0&0&  00*0*0*0*0*0*0*0*0*0*0*0*@0*@0*@0*@0*@0*@0*0*0*0*0*0*  0 0<<0<0<0<0<0<0<0<0< 0<<0D0D0D0D0D0D0D0D0D0D0D0D0D0D0D0D 0<<0P* 0PP0OQ0OQ0OQ0OQ0OQ0OQ0OQ0OQ0OQ0OQ* 0PP0X0X0X0X0X0X  00:]0:]0:]0:]0:]0:]0:]0:]0:]  0 0bb0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b 0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b 0b 0b 0b 0bb0+j 0bb0j0j0j0j0j0j0j0j0j0j 0j 0j 0j 0j 0j0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j0j 0bb0n0n0n0n 0n 0n 0n 0n 0n0n 0n 0n 0n 0n 0n0n 0n 0n 0n 0n 0n 0n 0n 0n 0n 0n 0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n 0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n0n 0n0n0n0n0n0n0n0n 0n0n0n0n0n0n@0n0n0n@0n0n0n0n 0bb0u|0u|0u|0u|0u|0u|0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|@0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u|0u| '0bb050505050505050505050505050505050505050505 050505050505050505050505050505050505050505050505050505 0505050505050505050505050505050505050505050505050505 050505050505050505050505050505050505050505050505050505 05050505050505 0bb0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+ 0+ 0+ 0+ 0+ 0+ 0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+ 0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+ 0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+ 0+0+0+0+0+0+|0+0+0+0+0+0+0+0+|0+0+0+0+0+0+|0+0+0+0+0+0+0+ 0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+0+ 0+0+0+0+0+0+0+0+0+ 0bb00@0@ 0@ 0@00000  0000  00y  0 0 0 0 0 0 0 0 0 0 0 0  0  0  0  0  0 0 0 0 0 0 0 0 0 0{00 YZ[depCZpc|}$ t u U Exf3!l h!##;&'2'i'R(8***M+,I.K...1345l67::<:i:j:l:::<<<=='=/==@@@@BCPEWEEsGHHHHFIIIJLLLMOOQfQoQ"R2RRT5UU*VV*WWXiY}YY[\\]I^c^^abbLcecfcgchcc d,dXddef3g5g_g`gghh$i,i1i4i;iBiCiHiNiTiZi[ifilirixiyizij6jQjqjjjjjk kkkkkk+k-k/k0k:kkkmm=nlnnnnnnnnnnnnnnoo ooooo"o(o.o/o0o7o=oCoIoJoUoXo^odojokolosoyoooooooooooo2plqtqqqqqqqqqqqqqqqqqrrrrrrr%r&r'r/r^r_r`rjrqrtrur{rrrrrrrrrrrrrrrrrrr tfunuuuuuuuuuuuuuuuvvvv1v:v;v_Cėwj{џܟ} C1Ef u_9TW15'e1ȴصٵڵ/0235689;<EFGHIabcf0000000000000000 0 000000000000000000000000000000000000000000 00u 0u 0u 0u 0u 0u 0u 0u 0u 0u 0u 0u 0u  0 000000 00#0# 00'0'0'0' 00*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0* 0 0==0'=0'=0'=0'=0'=0'=0'=0'= 0==0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE0PE 0==0fQ( 0fQfQ0"R0"R0"R0"R0"R0"R0"R0"R0"R0"R( 0fQfQ0iY0iY0iY0iY0iY0iY 00I^0I^0I^0I^0I^0I^0I^0I^0I^0I^ 0 0cc0 d0 d0 d0 d0 d0 d0 d0 d0 d0 d0 d0 d0 d0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d0 d0 d0 d      !"#$%&'()*+,-./042356789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|0 d0 d0 d0 d0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d 0 d  0cc00k 0cc0k0k0k0k0k0k0k0k0k0k0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0k 0cc0o0o0o0o0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o0o0o0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o0o0o0o0o0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o 0o0o0o0o0o0o0o0o0o0o0o0o 0cc0~0~0~0~0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~0~0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~  '0cc0 0 0 0  0  0  0  0  0  0  0  0  0  0  0  0 0  0  0  0  0  0  0  0  0  0  0  0  0 0 0 0  0  0  0  0  0  0  0  0  0  0  0  0 0  0  0  0  0  0  0  0  0  0  0  0  0 0 0 0  0  0  0  0  0  0  0  0  0  0  0  0 0  0  0  0  0  0  0  0  0  0  0  0  0 0 0 0  0  0  0  0  0  0  0  0  0  0  0  0 0  0  0  0  0  0  0  0  0  0  0  0  0  0cc00000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0000 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 000 0cc0Q0Q0Q0Q 0Q 0Q0Q0Q0Q0Q0Q 0000 00j 0 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0 џ 0 џ 0 џ 0 џ 0 џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ 0џ0X0@000U@0h00@0h00@0h00@0h00@0@0@0h00 0000H YZ[depCZpc|}$ t u U Exf3!l h!##;&'2'i'R(8***M+,I.K...1345l67::<:i:j:l:::<<<=='=/==@@@@BCPEWEEsGHHHHFIIIJLLLMOOQfQoQ"R2RRT5UU*VV*WWXiY}YY[\\]I^c^^abbLcecfcgchcc d,dXddef3g5g_g`gghh$i,i1i4i;iBiCiHiNiTiZi[ifilirixiyij6jQjqjjjjjk kkkkkk+k-k/k0k:kkkmm=nlnnnnnnnnnnnnnnoo ooooo"o(o.o/o0o7o=oCoIoJoUoXo^odojokolosoyoooooooooooo2plqtqqqqqqqqqqqqqqqqqrrrrrrr%r&r'r/r^r_r`rjrqrtrur{rrrrrrrrrrrrrrrrrrr tfunuuuuuuuuuuuuuuuvvvv1v:v;v_Cėwj{џܟ} C1Ef u_9TW15'e1ȴصf00000000000@0@000@0@000000000  00000000000000000000000000000000  00r 0r 0r 0r 0r 0r 0r 0r 0r 0r 0r 0r 0r   0 0<<0I0I0I0I0I 0<<0#0# 0<<0&0&0&0&  00*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*0*  0 0<<0=0=0=0=0=0=0=0= 0<<0E0E0E0E0E0E0E0E0E0E0E0E0E0E0E0E0E0E 0<<0U* 0UU0FV0FV0FV0FV0FV0FV0FV0FV0FV0FV* 0UU0]0]0]0]0]0]  00b0b0b0b0b0b0b0b0b0b  0 0ChCh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh 0Xh 0Xh 0Xh 0Xh0Xh 0Xh 0Xh 0Xh 0Xh0Xh 0Xh 0Xh 0Xh 0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh0Xh 0Xh 0Xh 0Xh0Xh 0Xh 0Xh 0Xh0Xh 0Xh 0Xh 0Xh 0ChCh0o 0ChCh0#p0#p0#p0#p0#p0#p0#p0#p0#p0#p0#p 0#p 0#p 0#p 0#p 0#p0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p 0#p0#p 0ChCh0s0s0s0s0s 0s 0s 0s 0s 0s0s 0s 0s 0s 0sProceedings Template - WORDEnd User Computing Services Normal.dot Yunbo CAO18Microsoft Office Word@@ 2@:@Թ> 2$՜.+,D՜.+,H hp|  ACMJ\~ Proceedings Template - WORD TitleH 8@ _PID_HLINKSAyQhttp://by14fd.bay14.hotmail.msn.com/cgi-bin/compose?curmbox=F000000001&a=76816560e9aacdd215c43639a924648095e57c27454f6c0233a7a7ba135bda76&mailto=1&to=lteng@cse.cuhk.edu.hk&msg=MSG1125992514.13&start=1650911&len=1935&src=&type=xF  FMicrosoft Office Word Document MSWordDocWord.Document.89q 0s0s 0s 0s 0s 0s 0s 0s 0s 0s 0s 0s 0s0s0s0s0s 0s 0s 0s 0s 0s0s 0s 0s 0s 0s 0s0s 0s 0s 0s 0s 0s 0s 0s 0s 0s 0s 0s0s0s0s0s0s0s 0s 0s 0s0s 0s 0s 0s 0s 0s 0s 0s0s 0s 0s 0s 0s 0s 0s 0s0s0s0s0s0s0s0s0s@0s0s0s0s 0ChCh00000 0 0 0 @0@0 @0 @0 @0 @0@0 @0 @0 @0 @0@0 @0 @0 @0 @0@0 @0 @0 @0 @0@0 @0 @0 @0 @0 000 0 0 0 00 0 0 0 0ۊ01cT70 0 0 0 0 00 0 0 0 00 0 0 0 0  '0ChCh0 0 0 0  0  0  0  0  0 0  0  0  0  0  0 0 0  0  0  0  0  0  0  0  0  0  0  0 0 0 0 0  0  0  0  0  0 0  0  0  0  0  0 0 0  0  0  0  0  0  0  0  0  0  0  0 0 0 0 0  0  0  0  0  0 0  0  0  0  0  0 0 0  0  0  0  0  0  0  0  0  0  0  0 0 0 0 0  0  0  0  0  0 0  0  0  0  0  0 0 0  0  0  0  0  0  0  0  0  0  0  0 0  0ChCh00000 0 0 0 0 00000 0 000000 0 0 0000 0 00000 0 0 0 0 00000 0 000000 0 0 0000 0 00000 0 0 0 0 00 0 00 0 000 0 00 0 0 0 0 00 0 00000 0 0 0 0 00000 0 000000 0 0 0000 0 00000 0 0 0 0 00 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 00000 0 0 0 0 00 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 0000 0ChCh0t0t0t0t 0t 0t0t0t0t0t0t  00 0 0   00b  0 0l 0l 0l 0l 0l 0l 0l 0l 0l 0 l 0 l 0 l 0 l 0 l 0l 0l 0l 0l 0l 0l 0l 0 0 0 0 0 0j00`   j h 4$ $)(c+P-/1956p8=:BCDPHILaNO)PPQ\TUX\_bdehjkkm:op6ruFwxzN{E|m~_*ՊێڏtuAٕgݗoS <W"پe`dfgijklmnpqrstuvxyz{|} !"#$%&'(*+p $K6'EQ*^k,qCqlq6r ss/szvw x"x7xUxkxxxxzzz{%{^{t{{{{{~~~:o֌؍EReЏ2ِWi|ˑBVtΔ$Bٕ fږ-Ko<Pn'beacehow~     ),dbF"F$FGF[F]F`FtFvFzFFFFFFGGGG HHH(H*HTHhHjHvHHHHHHfIzI|IIIIIIIGL[L]LcLwLyLLLL\] ]bbbMcacccccceX::::::::::::::::::::: !:l,2$/lP^^-R4QZ@(  n  0A ? ?"`B S  ?Lce qV(t OLE_LINK1 OLE_LINK2ffB2w/쇤 3w/#4w/\!5w/̴ 6w/ 7w/48w/ 9w/,:w/T;w/aw/7 ?w/{ @w/,#Aw/Bw/܄Cw/ Dw/Ew/Fw/ Gw/ Hw/OIw/܍ Jw/d Kw/4lLw/#Mw/4w Nw/\Ow/"Pw/|Qw/<Rw/{ Sw/< Tw/Uw/\ Vw/Dã Ww/ Xw/ "Yw/ԅ Zw/'[w/DL \w/ ]w/$"^w/$ _w/ĩ `w/T#aw/ bw/vcw/dw/S ew/ fw/t gw/? hw/iw/gjw/!kw/( lw/ mw/ nw/"ow/d pw/ qw/#rw/ sw/d<  e*38]f );;NUUcKT-U|a|aў!0SS{{gۮۮ f      !#"$%&'()*+,-./0123456789:;<=>?@A n)15;;do(3MSSZZ{ KT1Uaaڞ*6[[of   !#"$%&'()*+,-./0123456789:;<=>?@A <+*urn:schemas-microsoft-com:office:smarttagschmetcnv>=*urn:schemas-microsoft-com:office:smarttags PostalCode:A*urn:schemas-microsoft-com:office:smarttagsStreet;B*urn:schemas-microsoft-com:office:smarttagsaddress9?*urn:schemas-microsoft-com:office:smarttagsplace8@*urn:schemas-microsoft-com:office:smarttagsCity=9*urn:schemas-microsoft-com:office:smarttags PlaceType=:*urn:schemas-microsoft-com:office:smarttags PlaceName9>*urn:schemas-microsoft-com:office:smarttagsStateB5*urn:schemas-microsoft-com:office:smarttagscountry-region><*urn:schemas-microsoft-com:office:smarttags PersonName ѱ011.312185 Bojuan ZhaoFFalseHang LiHasSpace Jinzhu Li John ChenJun XumNegative NumberType ProductIDpt SourceValueTCSCTrueUnitName Yunbo Cao Yunhua HuBA@?>=<@:9BA@5?<AB@>5< <+ :9BA@?5@:9BA@5?<:9@?5?+ + @?+ + << << < @?@??>??@;Ccjk{~ KU"<CEMOWY`F K !!!!$$%%%% %%%[%b%%%w''((('(8(B(((((D)I))))))A4A6A?AAAAAAAFBPBC C CCTC_C{CC#G.G0G9G{JJJJJJ"K*K,K1K0:ntyԅڅhn:@blܟKQvCOR\]a9<KLP\g1;ioSZ`ifouy4=CFu}9>krxTgjwx|WpsŲe=Fȴڴݴڵ0022335689;<GHfc{~ZeK.R.S..<:C:D:h:l:s:t::<<< =@@@@BBwFzFHHHHMM5gYėj{џܟ$CRuڵ0022335689;<GHfڵ0022335689;<GHf K c&J.!jo@.@.@..@...@ ....@ .....@ ......@ .......@ ........h^`OJQJo(hHh^`OJQJ^Jo(hHohp^p`OJ QJ o(hHh@ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohP^P`OJ QJ o(hHhh^h`CJOJQJo([]"!joc&         `hZl$,. <EQSWne !56oHpK6h8_GWqA!7 :R%!?`gp~]o  S9gpA &- !U / = J , F N P ~W X } [4 ]L Rp ]Pmv $5EH+UVf5_]8m%k'7w} !W w+55%Tl t+Z7<p=d/cY-6PGCN\S} >@e ~AO5Wpss{EP|9W| &^g~3^41<L?9[p2v0NZ^`kkH%{)+zr~W S k 1!`^!p!t! "^ ""#{#m$!J$e$Hl$lu$%V%O %R8%{@%I%o/&E&nj&z&''&'-C'l'(p(+$(/(14(?(]G(MQ([(m(()7)4^)d) *$*<*K*u*c+++ +o+6+?+P+,, ,2y,f-}-. C.JW./[/^X/080|0z~0H+1sC10D1_S1id1;2U22kE23Y3wA3n3w34$4855X5X5\_5tb5 6+6Z6q6x67H 7,767$;7KS7u7!8v89 99(9xf9o9%::%+:yB:[N:^:i: ;+';D;N};Z+<Y< g< ===2=5=8=b=v&>:B>U><[>b>k>9?axj y#yE/y5yIyqNyyXy z'zyzo{{{({!{6{2G{P{V{e{h{7i{k|/|>|-^|,e|ai|o|_+}X}>f}e~<)~g~.x~,B7=MfhB~<:W@;(-/<Lzbd.dogyv&L4\Oe7mRi~.UUZ+ODhj+bdGj>ps9x7rm}O2R_f 6BHM?^";'0q0 R`awD ADM!FG`IQ !v)*+1ALS`n}19HYUVdyzkUr9]cNhC#BCKmO`Iw'.} <sj 1n$-<\L`vx |*iJpgY'+0GkT5mp!6#%}\%~z *+U^v);!6QCIO6u`8efVLmX~n2G[HJ&Za`oi;.j/pJBUhEF`yku8-DzJ[{9 #`jy-sb *R0^Zh_jfs390`djN1Rz0CGfkOy\^v-/9\]r +]@fI7 DJqo!uV1 :GL*MBFjGla  b(S )$)0KT\@%/3Q<Wgs_q#8@7Gf,~a=fF N86NaTY.46BsU_lwk,:O]X$4*Wn *hmx;X`6op9r<>F,a7>9 %XJzL?Sz BkFpY|RjB&]n+W0<JNPw-csk )=,r1hOg":>K`L~d*jo@8ipD<SDe\aE9gm [LMO5ghh$i1i4i;iBiCiHiNiTiZi[ifilirixiyijjk kkkkkk+k-k/k0knnnnnnnnoo ooooo"o(o.o/o0o7o=oCoIoJoUoXo^odojokolosoyooooooooooqqqqqqqqqqqqqqqqrrrrrrr%r&r^r_r`rjrqrtrur{rrrrrrrrrrrrrrrrrruuuuuuuuuuuuuuvvvv1v:v;vwq~~~4<ăʃЃփ׃܃ v|ƄDŽ̄҄؄ބ߄ ۅ܅݅ !39?EFGRX^deņІֆ܆ %+12opq{LJ͇ӇهڇۇABCMTWX^iou{|Ĉʈˈ$%0BHNTUVagmstċˋ΋ϋՋ"#$/5;ABLnjٌߌ PQR\cfgmx~Ǎ͍Ӎٍڍ#*-.4?EKQR]ou{ *<BHNOP[agmnP0ڵ.f;Q60- =V0=V0=V0=V0@\\msrasia\AdobePS Print DriverNe01:AdobePS Acrobat DistillerAdobePS Acrobat Distiller\\msrasia\AdobePS Print DriverS 4dA4PRIV0''''\KhC\\msrasia\AdobePS Print DriverS 4dA4PRIV0''''\KhC ϱ@{(noep@pvp@Unknown Gz Times New Roman5Symbol3& z Arial;[SOSimSun;& z Helvetica3z Times5& zaTahomaU MiriamTimes New Roman?5 z Courier New;Wingdings"h-&.|f-|f $\J$\J!4d~~2qHP?i2Proceedings Template - WORDEnd User Computing Services Yunbo CAO