ࡱ>  `bq`bjbjqPqP<@::Ӹ333835D@qz5GHHHH02HFH ppppppp$rh"upLHHLLpHHpiTiTiTLVHHpiTLpiTiTRmoH5 ^3mQn#pq0@qo0vST0v<o0vo(RH*IiTI|@JRHRHRHppSTRHRHRH@qLLLL## Compression of 3D Scientific Data for Archival, Distribution and Fast Classification First A. Author, Second B. Author, Jr., and Third C. Author, Member, IEEE AbstractIn this work, we propose a compression algorithm based on Partitioned Vector Quantization for 3D scientific data, and in particular those generated by remote sensors. One of the dimensions of the input data is treated as a single vector that is partitioned into subvectors of (possibly) different lengths, each quantized separately. Quantization indices and residuals are then entropy coded. Partitioning serves different purposes: vector quantization looses efficiency in high dimension and partitioning is a way of improving it; it allows a better decorrelation of the data in the given dimension, improving compression; the concatenation of the quantization indices can be seen as a projection to a lower dimensional subspace that preserves the original statistical properties of the data; partition indices can be used for fast (remote) browsing and classification. In fact, experimental results with NASA AVIRIS hyperspectral images show that the proposed algorithm achieves state of the art lossless compression on such data, provides nearlossless and lossy capabilities (that could be used for fast browsing and broadcasting on limited bandwidth communication channels), is scalable to arbitrary data size, allows fast classification and target detection in the compressed domain. Index TermsAbout four key words or phrases in alphabetical order, separated by commas. For a list of suggested keywords, send a blank e-mail to  HYPERLINK mailto:keywords@ieee.org keywords@ieee.org or visit the IEEE web site at  HYPERLINK "http://www.ieee.org/web/developers/webthes/index.htm" http://www.ieee.org/web/developers/webthes/index.htm. INTRODUCTION A n increasing number of scientific instruments produce high volume of data in the form of twodimensional matrices of vectors, also called data cubes. Typical examples are biomedical images, multi, hyper and ultra spectral images, measurements of volumetric parameters like atmospheric temperature and pressure by way of microwaves sounders. Compressing these data presents a challenge for the currently standardized compressors. Beside the three dimensional nature of the measurements, which exhibit correlation along each dimension, the single measurements have values of 16 bits or more. Current state of the art compressors do not perform well on data sources with large alphabets. Furthermore, it is desirable to have a single compressor that can perform well in all of lossless, near lossless, and lossy modes and can scale to data cubes of arbitrary sizes. Lossless compression is often required for data collection and archiving due to the cost of the data collection and that it may be destined to analysis and further elaboration. However, when data is downloaded by users, a lossy mode can be employed to speed up the transmission, since the final users may not need all measurements at their full precision, and in any case, can request further precision only when needed. In [1] we propose a compression algorithm based on Partitioned Vector Quantization, where one dimension of the cube is treated as a single vector, partitioned and quantized. Partitioning the vector into subvectors of (possibly) different lengths is necessary because the possibly large number of components. Subvectors are individually quantized with appropriate codebooks. The adaptive partitioning uses a novel locally optimal algorithm and provides a tool to reduce the size of the source alphabet. Correlation among the other two dimensions is exploited with methods derived from the image coding domain. Lossless or near-lossless entropy coding can be applied to the residual error. An additional feature of this compression method is the possibility to tightly bound error on a pixel by pixel basis. Due to the hierarchical structure of the data compressed with the proposed algorithm, it is possible to perform classification and target detection directly in compressed domain, with considerable speed and memory savings. The high speed classification and target detection algorithm that we describe here can process 90 percent of the compressed pixels with a simple table lookup, full decompression is only required if exact classification is necessary and it is limited to the 10 percent remaining pixels. Algorithms have been verified experimentally on the lossless and near lossless compression of NASA AVIRIS images. Section 2 briefly reviews the algorithm in order to introduce the necessary notation. Besides good compression performance, LPVQ presents a number of features that are highly desirable for these kinds of images. A communication paradigm that shows these advantages is outlined and discussed in Section 3. The problem of classification and the necessary notation is introduced in Section 4. Section 5 presents experimental results on the classification error introduced by LPVQ lossy and near lossless compression modes. Section 6 builds on these experimental observations with an algorithm that speeds up the classification by accepting or rejecting vectors based only on bounds of the VQ error. These error bounds are available at design time or can be extracted from the compressed image. In the same section, the asymptotic behavior is derived. Given a compressed image and a classification threshold, the speed up achieved on the nave algorithm depends on the acceptance and rejection rates. Experiments on typical AVIRIS images are performed in the same section in order to assess these rates for a number of classification thresholds. The acceptance and rejection rates only depend on the bounds on the quantization error and can be easily derived during the design of the LPVQ. So, these considerations also provide a tool useful to select the quantization parameters (codebook entries and number of partitions) in order to match a desired range of classification speeds. The Communication Model An application for which our algorithms were designed is the encoding of information collected by a remote sensing device having limited computational power (satellite, airplane, biomedical imager) and losslessly transmitted to a powerful central server (see Figure 1). The central server decompresses and elaborates the data, performing all necessary adjustments and corrections. Then, re-codes the data in a more compact representation and distributes the result among a number of users that may subscribe the service at different quality levels. A common example is weather information collected by geostationary satellites. The satellites observe of events evolving in a specific geographic area, such as hurricanes, storms, flooding and volcanic eruptions. The collected data have to be faithfully transmitted to a land based central station that analyzes the data, applies some form of enhancement, compressed the data further and sends the result back to the satellite in order to broadcast it to the final users. In this setting, the same satellite is used for both acquisition and broadcasting. More common is the case of the central server performing the transmission. While the compression hardware used in the satellite must be simple and require very low power to function, there is typically little or no limitation to the computational power of the base station. Definitions We model a three dimensional data cube as a discrete time, discrete values, bidimensional random source  EMBED Equation.DSMT36  that emits pixels that are  EMBED Equation.DSMT36 dimensional vectors  EMBED Equation.DSMT36 . Each vector component  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36 , is drawn from a finite alphabet  EMBED Equation.DSMT36  and is distributed according to a space variant probability distribution that may depend on other components. We assume that the alphabet has the canonical form  EMBED Equation.DSMT36 . Our encoder is based on a Partitioned Vector Quantizer (or PVQ), which consists of  EMBED Equation.DSMT36  independent,  EMBED Equation.DSMT36 levels,  EMBED Equation.DSMT36 dimensional exhaustive search vector quantizers  EMBED Equation.DSMT36 , such that  EMBED Equation.DSMT36  and:  EMBED Equation.DSMT36  is a finite indexed subset of  EMBED Equation.DSMT36  called codebook. Its elements  EMBED Equation.DSMT36  are the code vectors.  EMBED Equation.DSMT36  is a partition of  EMBED Equation.DSMT36  and its equivalence classes (or cells)  EMBED Equation.DSMT36  satisfy:  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  for  EMBED Equation.DSMT36 .  EMBED Equation.DSMT36  is a function defining the relation between the codebook and the partition as  EMBED Equation.DSMT36  if and only if  EMBED Equation.DSMT36 . The index  EMBED Equation.DSMT36  of the codeword  EMBED Equation.DSMT36 , result of the quantization of the  EMBED Equation.DSMT36 dimensional subvector  EMBED Equation.DSMT36 , is the information that is sent to the decoder. With reference to the previously defined  EMBED Equation.DSMT36  vector quantizers  EMBED Equation.DSMT36 , a Partitioned Vector Quantizer can be formally described by a triple  EMBED Equation.DSMT36  where:  EMBED Equation.DSMT36  is a codebook in EMBED Equation.DSMT36 ;  EMBED Equation.DSMT36  is a partition of EMBED Equation.DSMT36 ;  EMBED Equation.DSMT36  is computed on an input vector  EMBED Equation.DSMT36  as the concatenation of the independent quantization of the  EMBED Equation.DSMT36  subvectors of  EMBED Equation.DSMT36 . Similarly, the index vector sent to the decoder is obtained as a concatenation of the  EMBED Equation.DSMT36  indices. The design of a Partitioned VQ aims at the joint determination of the  EMBED Equation.DSMT36  partition boundaries  EMBED Equation.DSMT36  and at the design of the  EMBED Equation.DSMT36  independent vector quantizers having dimension  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36 . In the following, given a source vector  EMBED Equation.DSMT36 , we indicate the  EMBED Equation.DSMT36  subvector of boundaries  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  with the symbol  EMBED Equation.DSMT36  (for simplicity, the  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  spatial coordinates are omitted when clear from the context). The mean squared quantization error between the vector  EMBED Equation.DSMT36  and its quantized representation  EMBED Equation.DSMT36 , is given by  EMBED Equation.DSMT36  where  EMBED Equation.DSMT36  is the codeword of the  EMBED Equation.DSMT36  codebook minimizing the reconstruction error on  EMBED Equation.DSMT36 , and:  EMBED Equation.DSMT36  A Locally Optimal PVQ Design The complexity of building a quantizer that scales to vectors of arbitrary dimension is known to be computationally prohibitive. Furthermore, the efficiency of a VQ decreases with its dimensionality: the fraction of the volume of a hypersphere inscribed in a hypercube of the same dimension d goes to zero when d goes to infinity ([2]-[3]). Partitioning the vectors in consecutive, non-overlapping subvectors is a common solution to this problem (see the book of Gersho and Gray [4]). While a partitioned VQ leads to a suboptimal solution in terms of Mean Squared Error (MSE) because it does not exploit correlation among subvectors, the resulting design is practical and coding and decoding present a number of advantages in terms of speed, memory requirements and exploitable parallelism. In a Partitioned VQ, we divide the input vectors into  EMBED Equation.DSMT36  subvectors and quantize each of them with an  EMBED Equation.DSMT36 levels exhaustive search VQ. Having  EMBED Equation.DSMT36  subvectors of the same dimension (except perhaps one, if  EMBED Equation.DSMT36  does not divide  EMBED Equation.DSMT36 ) is not a possibility here, since we assume that the components of  EMBED Equation.DSMT36  may be drawn from different alphabets. In this case their distributions may be significantly different and partitioning the  EMBED Equation.DSMT36  components uniformly into  EMBED Equation.DSMT36  blocks may not be optimal. We wish to determine the size of the  EMBED Equation.DSMT36  sub vectors (of possibly different size) adaptively, while minimizing the quantization error, measured for example in terms of MSE. Once the  EMBED Equation.DSMT36  codebooks are designed, input vectors are encoded by partitioning them into  EMBED Equation.DSMT36  subvectors of appropriate length, each of which is quantized independently with the corresponding VQ. The index of the partitioned vector is given by the concatenation of the indices of the  EMBED Equation.DSMT36  subvectors. Given the number of partitions N and the number of levels per codebook L, it is possible to find the partition boundaries achieving minimum distortion with a brute-force approach. We first determine, for every  EMBED Equation.DSMT36 , the distortion  EMBED Equation.DSMT36  that an  EMBED Equation.DSMT36 -levels vector quantizer achieves on the input subvectors of boundaries  EMBED Equation.DSMT36  and EMBED Equation.DSMT36 . Then, with a dynamic program, we traverse the matrix  EMBED Equation.DSMT36  and find the  EMBED Equation.DSMT36  costs corresponding to the input partition of boundaries  EMBED Equation.DSMT36  and whose sum is minimal. A more sophisticated approach to partitioned vector quantization is discussed in Matsuyama [5]. It uses dynamic programming at each step to decide the current optimal partition, instead of designing one vector quantizer for each possible partition configuration first and then applying dynamic programming. Nevertheless, even this approach is computational intensive. The locally optimal partitioning algorithm that we propose here (LPVQ) provides an efficient alternative to dynamic programming, while performing comparably in most applications. Our partitioning algorithm is based on a variation of the Generalized Lloyd Algorithm (or GLA, Linde, Buzo and Gray [6]). Unconstrained vector quantization can be seen as the joint optimization of an encoder (the function  EMBED Equation.DSMT36  described before) and a decoder (the determination of the codewords for the equivalence classes of the partition EMBED Equation.DSMT36 ). GLA is an iterative algorithm that, starting from the source sample vectors chooses a set of codewords (also called centroids) and optimizes in turns encoder and decoder until improvements on the predefined distortion measure are negligible. In order to define our PVQ, the boundaries of the vector partition  EMBED Equation.DSMT36  have to be determined as well. Our design follows the same spirit of the GLA. The key observation is that, once the partition boundaries are kept fixed, the Mean Square Error (the distortion we use in our application) is minimized independently for each partition by applying the wellknown optimality conditions on the centroids and on the cells. Similarly, when the centroids and the cells are held fixed, the (locally optimal) partitions boundaries can be determined in a greedy fashion. The GLA step is independently applied to each partition. The equivalence classes are determined as usual, but as shown in Figure 2, the computation keeps a record of the contribution to the quantization error of the leftmost and rightmost components of each partition:  EMBED Equation.DSMT36 and  EMBED Equation.DSMT36  Except for the leftmost and rightmost partition, two extra components are also computed:  EMBED Equation.DSMT36 and  EMBED Equation.DSMT36  The reconstruction values used in the expressions for  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  are determined by the classification performed on the components EMBED Equation.DSMT36 . The boundary  EMBED Equation.DSMT36  between the partitions  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  is changed according to the criteria  EMBED Equation.DSMT36 = min( EMBED Equation.DSMT36 , EMBED Equation.DSMT36 , EMBED Equation.DSMT36 ) if ( EMBED Equation.DSMT36 = EMBED Equation.DSMT36 )  EMBED Equation.DSMT36  else if ( EMBED Equation.DSMT36 = EMBED Equation.DSMT36 )  EMBED Equation.DSMT36  Entropy Coding The partitioned vector quantizer is used here as a tool to reduce the dimensionality of the source vectors. Dependence between the other two dimensions has to be exploited by means of an entropy coder. Furthermore, when a lossless coding is required, quantization residuals have to be encoded as well, entropy coded with a conditioning on the subvector indices. The index vectors  EMBED Equation.DSMT36  and the codewords  EMBED Equation.DSMT36 , with  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 , are sufficient to a lossy reconstruction of the data. When higher accuracy is needed, the compressed data can be augmented with the quantization error:  EMBED Equation.DSMT36  where, for each  EMBED Equation.DSMT36 :  EMBED Equation.DSMT36  The unconstrained quantizers work independently from each other and independently on each source vector and an entropy encoder must be used to exploit this residual redundancy. In particular, each VQ index  EMBED Equation.DSMT36  is encoded conditioning its probability with respect to a set of causal indices spatially and spectrally adjacent. The components of the residual vector  EMBED Equation.DSMT36  are entropy coded with their probability conditioned on the VQ index  EMBED Equation.DSMT36 . In near-lossless applications, a small, controlled error can be introduced at this stage. Figure 3 depicts a diagram of the LPVQ encoder. Computational Complexity Vector Quantization is known to be computationally asymmetrical, in the sense that encoding has a computational complexity which is orders of magnitude higher than decoding. Even more time consuming is the design of the dictionary which is also not guaranteed to converge into an easily predictable number of iterations. Numerous methods have been designed to speed up the convergence, for example, by constraining the search of the closest codeword like in [7], or by seeding the dictionary with vectors having special properties [8]. Adding structure to the VQ can also simplify encoding at the price of a limited degradation of the output quality [4]. All these improvements clearly apply to our method as well; furthermore, practical scenarios arise in which our algorithm can be applied as is without the computational complexity being an obstacle. A possible way of mapping the LPVQ algorithm to the communication model described in Section II is to equip platform used for the remote acquisition with an encoder able to download a dictionary transmitted from the central server (see Figure 1). This dictionary can be, for the very first scene, bootstrapped as described in [8], with a number of vectors randomly selected from the image itself, with a dictionary designed for another image or with any dictionary that believed to be statistically significant. Due to the nature of our algorithm, the choice of the first dictionary will only impact the compression ratio of the very first transmission. Once the central server receives a scene, lets say at time instant t, a few iterations of LBG may be necessary to refine the dictionary, adjust the partition boundaries and improve compression. The refined version of the dictionary is transmitted back to the remote platform and used for the encoding of the scene acquired at time t+1. In Section VIII we report on experiments intended to assess the performance degradation derived from a mismatched dictionary. It has been observed in [1] that if the codewords in each codebook are sorted according to their energy, the  EMBED Equation.DSMT36  planes constituted by the indices retain important information on the structure of the original image. These planes alone can be used to browse and select areas of interest and, as we show in section XX, to extract important information on the nature of the original pixel vectors. Experimental results LPVQ has been tested on a set of five AVIRIS images ([9]). AVIRIS images are obtained by flying a spectrometer over the target area. They are 614 pixels wide and typically have a height on the order of 2,000 pixels, depending on the duration of the flight. Each pixel represents the light reflected by a 20m x 20m area (high altitude) or 4m x 4m area (low altitude). The spectral response of the reflected light is decomposed into 224 contiguous bands (or channels), approximately 10nm wide and spanning from visible to near infrared light (400nm to 2500nm). Each of these vectors provides a spectral signature of the area. Spectral components are acquired in floating point 12-bit precision and then scaled and packed into signed 16 bit integers. After acquisition, AVIRIS images are processed to correct for various physical effects (geometrical distortion due to the flight trajectory, time of day, etc.) and stored in scenes of 614 by 512 pixels each (when the image height is not a multiple of 512, the last scene will be smaller). All files for each of the 5 test images were downloaded from the NASA web site (JPL [9]) and the scenes belonging to the same flight merged together to form a complete image. Several experiments have been performed for various numbers of partitions and for different codebook sizes. The results that we describe here were obtained for  EMBED Equation.DSMT36  partitions and  EMBED Equation.DSMT36  codebook levels. The choice of the number of levels makes also practical the use of offtheshelf image compression tools that are fine-tuned for 8 bit data. The LPVQ algorithm trains on each image independently and the codebooks are sent to the decoder as side information. The size of the codebook is negligible with respect the size of the compressed data (256 x 224 x 2 bytes = 112 KB) and its cost is included in the reported results. The partition boundaries for each of the five images are depicted in Figure 4. While similarities exist, the algorithm converges to different optimal boundaries on different input images. This is evidence that LPVQ adapts the partitions to input statistics. Starting with a codebook obtained by random image pixels, we have found that adaptation is fairly quick and boundaries converge to their definitive values in less than one hundred iterations. If LPVQ is bootstrapped with the method described in [8], which populates the dictionary with the vector with highest norm and then adds the vector whose minimum distance from the current dictionary is maximum, the algorithms shows a much faster convergence. Furthermore, the constrained search for the closest codeword is generally faster by a factor of 2. LPVQ performances are analyzed in terms of Compression Ratio (defined as the size of the original file divided by the size of the compressed one), Signal to Quantization Noise Ratio, Maximum Absolute Error and Percentage Maximum Absolute Error. The Signal to Quantization Noise Ratio (SQNR) is defined here as:  EMBED Equation.DSMT36  The correction factor  EMBED Equation.DSMT36  is introduced to take into account the error introduced by the 12 bit analogtodigital converter used by the AVIRIS spectrometer ([9]). This solution also avoids unbounded values in the case of a band perfectly reconstructed. The Maximum Absolute Error (MAE) is defined in terms of the MAE for the  EMBED Equation.DSMT36  band as EMBED Equation.DSMT36 , where  EMBED Equation.DSMT36  is:  EMBED Equation.DSMT36  The average Percentage Maximum Absolute Error (PMAE) for the  EMBED Equation.DSMT36  band having canonical alphabet  EMBED Equation.DSMT36  is defined as:  EMBED Equation.DSMT36  Table I shows that LPVQ achieves on the five images we have considered an average compression of 3.14:1, 45% better than the 1-D lossless compressor bzip2 when applied on the planeinterleaved images (worse results are achieved by bzip2 on the original pixelinterleaved image format) and 55% better than the standard lossless image compressor JPEGLS. We also compared LPVQ with the method published in [11], even though it reports experiments on a subset of our data set. The algorithm in [11] uses LBG to cluster data. An optimized predictor is computed for each cluster and for each band. Prediction error is computed and entropy coded, along with side information regarding the optimized predictors. The resulting method outperforms LPVQ, while sharing time complexity similar to the design stage of LPVQ and lacking any extra feature provided by the latter, like lossy and nearlossless mode, fast browsing and classification, et cetera. The last column of Table I reports, as a reference, the compression and the SQNR when only the indices are encoded and the quantization error is fully discarded. As we can see from the table, on average we achieve 55:1 compression with 25.27dB of SQNR. More interesting and practical are the results obtained with the nearlossless settings, shown in Table II. At first, the introduction of a small and constant quantization error across each dimension is considered; that is, before entropy coding, each residual value x is quantized by dividing x adjusted to the center of the range by the size of the range; i.e., q(x) = ((x+()/(2(+1)(. This is the classical approach to the nearlossless compression of image data and results into a constant MAE across all bands. With this setting, it is possible to reach an average compression ratio ranging from 4:1 with the introduction of an error  EMBED Equation.DSMT36  and a  EMBED Equation.DSMT36  to 10:1 with and error of  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 . While the performance in this setting seem to be acceptable for most applications and the SQNR is relatively high even at high compression, the analysis of the contribution to the PMAE of the individual bands shows artifacts that might be unacceptable. In particular, while the average PMAE measured across the 224 bands of the AVIRIS cube is low, the percentage error peaks well over 50% on several bands (see Figure 5). Since the PMAE is relevant in predicting the performance of many classification schemes, we have investigated two different approaches aimed at overcoming this problem. In both approaches we select a quantization parameter that is different for each band and it is inversely proportional to the alphabet size (or dynamic). In general, high frequencies, having in AVIRIS images higher dynamic, will be quantized more coarsely than low frequencies. We want this process to be governed by a global integer parameter  EMBED Equation.DSMT36 . The first method, aiming at a quasiconstant PMAE across all bands, introduces on the  EMBED Equation.DSMT36 band a distortion  EMBED Equation.DSMT36  such that:  EMBED Equation.DSMT36  Since the  EMBED Equation.DSMT36  band has alphabet  EMBED Equation.DSMT36  we must have:  EMBED Equation.DSMT36  The alternative approach, aims at a quasiconstant SQNR across the bands. If we allow a maximum absolute error  EMBED Equation.DSMT36  on the  EMBED Equation.DSMT36  band, it is reasonable to assume that the average absolute error on that band will be EMBED Equation.DSMT36 . If we indicate with  EMBED Equation.DSMT36  the average energy of that band and with  EMBED Equation.DSMT36  the target average maximum absolute error, then the absolute quantization error allowed on each band is obtained by rounding to the nearest integer the solution of this system of equations:  EMBED Equation.DSMT36  As can be seen from Table II, the three methods for nearlossless coding of AVIRIS data are equivalent in terms of average SQNR at the same compression. However, the quasiconstant PMAE method is indeed able to stabilize the PMAE across each band (Figure 6). The small variations are due to the lossless compression of some bands and the rounding used in the equations. The average SQNR is not compromised as well. Similar results are observed for the quasiconstant SQNR approach. The SQNR is almost flat (Figure 7), except for those bands that are losslessly encoded and those with small dynamic. The PMAE is also more stable than the constant MAE method (Figure 8). LPVQ and Random Dictionaries In order to evaluate the proposed approach in the framework described in Section II, we performed the following experiments for each scene of the Moffett Field image: we compute compression ratio and RMSE at each step of the LPVQ algorithm, starting with a random dictionary; same except that the starting dictionary is the locally optimal one for the previous scene. As we can see from Figure 9, compression performance is quite stable already after just 50 iterations. The compression ratio is slightly lower when LPVQ is fed with the dictionary from the previous scene, while RMSE, on the other hand, is much lower. This suggests that the entropy coder needs to be tuned-up for this case and that the lossy compression is slightly better when only few iterations are allowed. A different experimental setting uses six different schemes to generate random dictionaries. One hundred dictionaries for each scheme are used with PVQ (i.e., the dictionary is not refined with the proposed LBG-like scheme to obtain a locally optimal one) and with both uniform and optimal partitions. Compression results for each scheme are boxplotted in Figure 10. The results suggest that taking a small sample of the input data is enough to obtain compression performances reasonably close to the locally optimal solution produced by LPVQ, and that the non-uniform partitioning is much more effective at high compression. The Problem of Classification Let each image pixel be a  EMBED Equation.DSMT36 -dimensional vectors  EMBED Equation.DSMT36  divided into  EMBED Equation.DSMT36  disjoint subvectors as  EMBED Equation.DSMT36 , the  EMBED Equation.DSMT36 -th partition having dimension  EMBED Equation.DSMT36 , for  EMBED Equation.DSMT36 . For simplicity the boundaries are omitted here. Upon quantization each vector is represented by an  EMBED Equation.DSMT36 -dimensional vector of indices  EMBED Equation.DSMT36  and by  EMBED Equation.DSMT36 , a  EMBED Equation.DSMT36 -dimensional vector representing the quantization error. Each component  EMBED Equation.DSMT36  indexes one of the  EMBED Equation.DSMT36  codebook centroids for the VQ corresponding to the partition  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36 . If  EMBED Equation.DSMT36  is the quantized value for  EMBED Equation.DSMT36 , LPVQ decomposes each component of an input vector  EMBED Equation.DSMT36  into as  EMBED Equation.DSMT36 , where  EMBED Equation.DSMT36 . Given a target vector  EMBED Equation.DSMT36 , a distortion measure  EMBED Equation.DSMT36  and a classification threshold  EMBED Equation.DSMT36 , we wish to determine, for every image pixel  EMBED Equation.DSMT36 , whether  EMBED Equation.DSMT36 . When this condition is satisfied, we say that the pixel  EMBED Equation.DSMT36  is a member of the class  EMBED Equation.DSMT36 . The Euclidean Distance  EMBED Equation.DSMT36  is distortion measures commonly used in classification. The classification threshold  EMBED Equation.DSMT36 , plays an important role since it determines whether the classification is used to detect a rare pixel instance, eventually present in the image (problem commonly referred to as target detection) or whether the objective of the classification is the determination of a thematic map, where almost every pixel belongs to one of several classes. Effect of Coding Error on the Classification As we did in the case of the compression we used AVIRIS hyperspectral images to assess the effect of the coding error on classification. Classification and target detection are algorithms frequently applied to hyperspectral images; when the images are lossily encoded, it is licit to ask whether and in which measure the coding error affects the results of these algorithms. In order to assess the classification errors introduced by LPVQ lossy modes, we have experimented with the standard images for various errors and threshold parameters. It is important to notice that, while the lossy results are peculiar of our algorithm, the near lossless apply to any compression scheme that bounds the maximum absolute error on each vector component. The results for three images, Cuprite, Moffet Field and Low Altitude, are presented here. Results on the other two images were very similar. In our experiments, the absolute error introduced by LPVQ in near lossless mode is respectively bounded to a maximum of 1, 2, 4 and 8 units. The classification uses Euclidean Distance with an average threshold per band  EMBED Equation.DSMT36 ; typically, values in the range  EMBED Equation.DSMT36  are of some interest in practical applications. The figures show the percentage of errors in the classification of 8 targets selected from each image. The targets are hand picked from the reduced resolution image constituted by the index planes and are meant to represent targets of common interest (buildings, roads, water, vegetation, etc...). The lossless encoding is used as a reference for the classification and the error percentages reflect the sum of all false positive (classifications that were not classified in the lossless), false negative (vectors classified in the lossless and not classified in the lossy) and misclassifications (different clusters in lossless and lossy). As it is possible to see from Figures 11-13, in near lossless, the classification is fundamentally unaffected; even for an error of  EMBED Equation.DSMT36 , the percentage of classification errors is, in the worst case, below 0.1%. This is sufficient for most applications. Smaller coding errors, achieve even lower percentages. For the compression parameters used in Section VII (i.e., 256 code vectors and 16 partitions) the indices and the code vectors alone (Figure 14, lossy case) achieve a classification error always lower than 3%. While this error rate may be unsuitable for some applications, as we show in the next section, it is possible to take advantage of this characteristic to design a faster classifier. Speeding up the Classification Since the LPVQ indices provide most of the information necessary to classify the pixels, we designed a simple algorithm that, bounding the minimum and maximum quantization error for each component and for each cluster, can decide the classification of most pixels without inspecting their error vector  EMBED Equation.DSMT36 . Bounds on the quantization error are available at design time or can be inferred from the encoded image. The advantages of this algorithm are twofold. Firstly, the number of partitions and the number of centroids are typically small, so pre-computed values can be arranged into a lookup table. If most vectors can be classified without the need of the error, this will result into a substantial speed up. Secondly, we can imagine a user on the field, browsing and trying different classifications on the LPVQ indices instead of the whole image. Then, when a parameterization of interest is found, the user can query the server for the error vectors necessary to resolve any classification ambiguity. Given a threshold  EMBED Equation.DSMT36 , a target vector  EMBED Equation.DSMT36  and a distortion measure  EMBED Equation.DSMT36 , we wish to determine, from the quantization indices of a pixel  EMBED Equation.DSMT36 , two quantities  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 , respectively lower and upper bounding the distortion  EMBED Equation.DSMT36 . Since  EMBED Equation.DSMT36  belongs to the class  EMBED Equation.DSMT36  only if  EMBED Equation.DSMT36 , we can observe that: If  EMBED Equation.DSMT36 , the pixel  EMBED Equation.DSMT36  clearly does not belong to the class  EMBED Equation.DSMT36 ; If  EMBED Equation.DSMT36 , the pixel  EMBED Equation.DSMT36  trivially belongs to the class  EMBED Equation.DSMT36 ; If  EMBED Equation.DSMT36  then the quantization indices and the bounds on the error are not sufficient to decide the classification. In this case, and only in this case, it is necessary to access the error vector and compute  EMBED Equation.DSMT36 . Figure 15 shows the pseudocode that implements this scheme. The main assumption is that the quantities  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  can be computed by summing the individual contributions of each vector component  EMBED Equation.DSMT36 , upper and lower bounded by the minimum and maximum of the quantization residuals  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 . This is clearly the case of Euclidean Distance, here defined as:  EMBED Equation.DSMT36  Changing the threshold  EMBED Equation.DSMT36  into  EMBED Equation.DSMT36 , we can focus on the contribution of  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  to the individual terms of the sum. For a given component  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36 , in the partition  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36  we have:  EMBED Equation.DSMT36 . If  EMBED Equation.DSMT36  has been quantized with the centroid  EMBED Equation.DSMT36 , EMBED Equation.DSMT36 :  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 . Summing these bounds over all components of the partition  EMBED Equation.DSMT36  we obtain  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 . These quantities are precomputed in the algorithm in Figure 15 by the two nested for loops. Since there are only  EMBED Equation.DSMT36  possible values and  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  are typically small ( EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  in Section VII),  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  can be stored in a lookup table indexed by  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36 . The computation of  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  requires a total of  EMBED Equation.DSMT36  lookups and  EMBED Equation.DSMT36  sums. If one of the two comparisons  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  succeeds, then the pixel  EMBED Equation.DSMT36  is classified from its indices only, with a number of operations proportional to  EMBED Equation.DSMT36 . Otherwise, the error vector  EMBED Equation.DSMT36  must be retrieved,  EMBED Equation.DSMT36  exactly reconstructed as  EMBED Equation.DSMT36  and finally  EMBED Equation.DSMT36  computed. This requires a sum a subtraction and a product for each component and  EMBED Equation.DSMT36  sums to accumulate the result, i.e. a total of  EMBED Equation.DSMT36  subtractions,  EMBED Equation.DSMT36  sums and  EMBED Equation.DSMT36  products, a number of operations proportional to  EMBED Equation.DSMT36  (for AVIRIS images,  EMBED Equation.DSMT36 ). The speed of the new classification algorithm depends on the probability that the two tests fail and that we have to use the error vector in order to classify. Let this probability be:  EMBED Equation.DSMT36  If  EMBED Equation.DSMT36 ,  EMBED Equation.DSMT36  and  EMBED Equation.DSMT36  are respectively the times to perform a subtraction, a sum and a multiplication on a given computing platform, the total running time of the new classification is:  EMBED Equation.DSMT36  Figures 16-18 show an experimental assessment of the quantity  EMBED Equation.DSMT36  for three AVIRIS images. They have been classified using 100 randomly selected targets. Average results are reported for thresholds  EMBED Equation.DSMT36  and quantizers having 256, 512 and 1024 levels. It is possible to see how the percentage of vectors on which the simplified test fails to classify the points reaches a maximum between 7.5% and 11% for a 1024 levels quantizer and 12% to 18% for a 256 level quantizer. Practical figures are typically better. Problems like target detection are solved by applying a low threshold, while classification requires a high one; in these regions of the curve, the percentages may be much lower than the maxima. These results can be used in two ways. Given an image encoded with LPVQ, we can improve the speed the nave classification with the use of a lookup table. Alternatively, we can use this as a tool to select the number of quantization levels and match a desired average classification speed Conclusions We have presented an extension of the GLA algorithm to the locally optimal design of a partitioned vector quantizer (LPVQ) for the encoding of source vectors drawn from a high dimensional source on EMBED Equation.DSMT36 . It breaks down the input space into independent subspaces and for each subspace designs a minimal distortion vector quantizer. The partition is adaptively determined while building the quantizers in order to minimize the total distortion. Experimental results on lossless and near-lossless compression of hyperspectral imagery have been presented, and different paradigms of nearlossless compression are compared. Aside from competitive compression and progressive decoding, LPVQ has a natural parallel implementation and it can also be used to implement search, analysis and classification in the compressed data stream. High speed implementation of our approach is made possible by the use of small independent VQ codebooks (of size 256 for the experiments reported), which are included as part of the compressed image (the total size of all codebooks is negligible as compared to the size of the compressed image and our experiments include this cost). Decoding is no more than fast table look-up on these small independent tables. Although encoding requires an initial codebook training, this training may only be necessary periodically (e.g., for successive images of the same location), and not in real time. The encoding itself involves independent searches of these codebooks (which could be done in parallel and with specialized hardware). References G. Motta, F. Rizzo, and J.A. Storer, Compression of Hyperspectral Imagery, in Proceedings Data Compression Conference, J.A. Storer and M. Cohn, Eds. 2003, pp. 333-342, IEEE Computer Society Press. D. Landgrebe, Hyperspectral Image Data Analysis, in IEEE Signal Processing Magazine, January 2002, pp. 1728 M. G. Kendall, A course in the Geometry of nDimensions. New York: Hafner, 1961. A. Gersho and R.M. Gray, Vector Quantization and Signal Compression. Kluwer Academic Press, 1991. Y. Matsuyama. Image Compression Via Vector Quantization with Variable Dimension, TENCON 87: IEEE Region 10 Conference Computers and Communications Technology Toward 2000, Aug. 1987, Seoul, South Korea. Y. Linde, A. Buzo, and R. Gray, An algorithm for vector quantizer design, IEEE Transactions on Communications 28, 84-95, 1980. Huang, Bi, Stiles, and Harris, in IEEE Transactions on Image Processing, July 1992. I. Katsavounidis, C.C.J. Kuo, and Z. Zhang, A New Initialization Technique for Generalized Lloyd Iteration, in IEEE Signal Processing Letters, vol. 1, no. 10, pp. 144-146, October 1994 JPL [2002]. NASA Web site:  HYPERLINK "http://popo.jpl.nasa.gov/html/aviris.freedata.html" http://popo.jpl.nasa.gov/html/aviris.freedata.html B. Aiazzi, L. Alparone and S. Baronti, NearLossless Compression of 3D Optical Data, IEEE Transactions on Geoscience and Remote Sensing, vol. 39, no. 11, 2001. J. Mielikainen and P. Toivanen, Clustered DPCM for the Lossless Compression of Hyperspectral Images, in IEEE Transactions on Geoscience and Remote Sensing, vol. 41, no. 12, pp. 29432946, December 2003. F. Rizzo, B. Carpentieri, G. Motta, and J. A. Storer, High Performance Compression of Hyperspectral Imagery with Reduced Search Complexity in the Compressed Domain, in Proceedings Data Compression Conference, J.A. Storer and M. Cohn, Eds. 2004, IEEE Computer Society Press. . G. P. Abousleman [1995]. Compression of Hyperspectral Imagery Using Hybrid DPCM/DCT and Entropy-Constrained Trellis Coded Quantization, Proc. Data Compression Conference, IEEE Computer Society Press. B. Aiazzi, L. Alparone, A. Barducci, S. Baronti and I. Pippi [2001]. Information-Theoretic Assessment of Sampled Hyperspectral Imagers, IEEE Transactions on Geoscience and Remote Sensing 39:7. . . JPL [2002]. NASA Web site:  HYPERLINK "http://popo.jpl.nasa.gov/html/aviris.freedata.html" http://popo.jpl.nasa.gov/html/aviris.freedata.html Y. Linde, A. Buzo, and R. Gray [1980]. An algorithm for vector quantizer design, IEEE Transactions on Communications 28, 84-95. M. Manohar and J. C. Tilton [2000]. Browse Level Compression of AVIRIS Data Using Vector Quantization on Massively Parallel Machine, Proceedings AVIRIS Airborne Geoscience Workshop. Y. Matsuyama [1987]. Image Compression Via Vector Quantization with Variable Dimension, TENCON 87: IEEE Region 10 Conference Computers and Communications Technology Toward 2000, Aug. 25-28, Seoul, South Korea. R. E. Roger, J. F. Arnold, M. C. Cavenor and J. A. Richards, [1991]. Lossless Compression of AVIRIS Data: Comparison of Methods and Instrument Constraints, Proceedings AVIRIS Airborne Geoscience Workshop. S. R. Tate [1997], Band ordering in lossless compression of multispectral images, IEEE Transactions on Computers, April, 477-483. G. Shaw and D. Manolakis [2002], Signal Processing for Hyperspectral Image Exploitation, IEEE Signal Processing Magazine 19:1 S. Subramanian, N. Gat, A. Ratcliff and M. Eismann [1992]. Real-Time Hyperspectral Data Compression Using Principal Component Transform, Proceedings AVIRIS Airborne Geoscience Workshop. G. O. Young, Synthetic structure of industrial plastics (Book style with paper title and editor), in Plastics, 2nd ed. vol. 3, J. Peters, Ed. New York: McGraw-Hill, 1964, pp. 1564. [1] [2] NASA, AVIRIS home page,  HYPERLINK "http://popo.jpl.nasa.gov/" http://popo.jpl.nasa.gov, February 2003 [3]. [4]. [5] M. Manohar and J.C. Tilton, Browse level compression of AVIRIS data using vector quantization on massively parallel machine, in Proceedings AVIRIS Airborne Geosciences Workshop, 2000. SPURGARE LE REFERENCES ED AGGIUNGERE PAGE 1  EFjkl}~'()*+}+-6T7PouӾݱӨzrh*hnp6hnpB*phh*hnp:h*hnp:CJ8EHh*hnpCJ8EH h*hnphnp0J5CJjhnpCJUhnp0JCJjhnpCJUjhnpCJU hnpCJ hnp6CJ h3hnp hnp6 hnp0Jhnp,UV ').%%P'(2))+*$a$gdnp & Fgdnpgdnp)gdnp$d & ,D`gdnpgdnpgdnpgdnpgdnpgdnpgdnpu?@AEGP{%|%%%%%%%%%%%%%%%%&&&1&2&3&4&6&7&O&zjhnpEHU jC hnpUVmHnHuj hnpEHU jC hnpUVmHnHujhnpEHU jC hnpUVmHnHujhnpEHU jC hnpUVmHnHujhnpU hhnp h3hnphnp)O&P&Q&R&t&u&&&&&2'3'K'L'M'N''''''''''''''' ( ( (Ⱦ٭ْwm\Rj,hnpEHU jC hnpUVmHnHuj(hnpEHU jC hnpUVmHnHujL$hnpEHU jC hnpUVmHnHujLhnpEHU jC hnpUVmHnHujhnpEHU jC hnpUVmHnHuhnpjhnpUjhnpEHU jC hnpUVmHnHu ((?(@(X(Y(Z([(g(h(((((((((((((((((()))))2)3)ǽ󳭜sbVjFhnpEHPJU j̳C hnpUVmHnHujBhnpEHPJU jɳC hnpUVmHnHuj<hnpEHPJU jƳC hnpUVmHnHu hnpPJjhnpPJUj6hnpEHU jóC hnpUVmHnHuj!1hnpEHU jC hnpUVmHnHuhnpjhnpU!3)K)L)M)N)a)b)z){)|)})))))))))))))))))**** *¶ukZPjJ`hnpEHU j۳C hnpUVmHnHujZZhnpEHU jسC hnpUVmHnHuhnpjhnpUjUhnpEHPJU jճC hnpUVmHnHu hnp6PJj8QhnpEHPJU jҳC hnpUVmHnHujhnpPJUjEKhnpEHPJU jϳC hnpUVmHnHu hnpPJ **&*'*(*)*+*,*D*E*F*G***************++++++0+ȷȚ}sbXjvyhnpEHU j곒C hnpUVmHnHujthnpEHU j糒C hnpUVmHnHujJohnpEHPJU j䳒C hnpUVmHnHujihnpEHPJU j᳒C hnpUVmHnHu hnpPJjhnpPJUjnehnpEHU j޳C hnpUVmHnHuhnpjhnpU +**+,-N-./2"22236;)?W@EEGFFGtH%$Z@]Z^@`a$gdnp)gdnpgdnp$a$gdnp & Fgdnpgdnp & Fgdnp0+1+2+3+W+X+p+q+r+s+++++++,,,,,,0,1,I,J,K,L,,,,,,Ⱦ٭ْwm\RjhnpEHU jC hnpUVmHnHuj hnpEHU jC hnpUVmHnHuj׊hnpEHU jC hnpUVmHnHujhnpEHU jC hnpUVmHnHuj:hnpEHU jC hnpUVmHnHuhnpjhnpUj}hnpEHU jC hnpUVmHnHu ,,,,,,,,,,,,,-------0-1-I-J-K-L-M-N-O-g-h-i-j-㵩㘌{o^Rj]hnpEHPJU j C hnpUVmHnHujhnpEHPJU jC hnpUVmHnHujҤhnpEHPJU jC hnpUVmHnHujahnpEHPJU jC hnpUVmHnHujIhnpEHPJU jC hnpUVmHnHu hnpPJjhnpPJUhnpjhnpU j--------------...'.(.).*...........// / /ύ|rύaWj'hnpEHU jC hnpUVmHnHujhnpEHU jC hnpUVmHnHujhnpUjýhnpEHPJU jC hnpUVmHnHujhnpEHPJU jC hnpUVmHnHuhnpjдhnpEHPJU jC hnpUVmHnHujhnpPJU hnpPJ" / /!/9/:/;/MVMWMXMYMMM N NNNUNVNnNoNpNqNrNsN齳题}lbWhBBhnpB*phj6hnpEHU jC hnpUVmHnHuj0hnpEHU jC hnpUVmHnHuj+hnpEHU jC hnpUVmHnHuj#hnpEHU jC hnpUVmHnHuj hnpEHU jC hnpUVmHnHuhnpjhnpUjhnpEHU"sNNNNNNOmRRRESaSSS5VPVVV[VVV7W8WPWQWRWSWmXXZZ\#\>]]]]]]] ^ ^$^%^߾|k jC hnpUVmHnHuj?hnpEHU jC hnpUVmHnHujhnpUh9hnp6PJ] hnpPJj|;h9hnpEHU*jUC h9hnpCJUVmHnHujh9hnpU h9hnp h0Thnp h/hnphnpB*phhnp*%^&^'^ c ccc&d@dAdBdZd[d\d]d^dtdudddddeeeeeeeeeeeeeeeeܷܦܷ܉}ܷl`jZhnpEHPJU jC hnpUVmHnHuj&VhnpEHPJU jC hnpUVmHnHuj~QhnpEHPJU jC hnpUVmHnHujhnpPJUjHhnpEHU jC hnpUVmHnHu hnpPJ hthnphnpjhnpUj9DhnpEHU%_a ccAd^dte%fBffgkrf?f@fAfBfNfffffffffffffffffgggύ|raWjuhnpEHU jǴC hnpUVmHnHujohnpEHU jĴC hnpUVmHnHujhnpUj3khnpEHPJU jC hnpUVmHnHuj0dhnpEHPJU jC hnpUVmHnHuhnpj_hnpEHPJU jC hnpUVmHnHujhnpPJU hnpPJ"ggVkkk=l'm(m+m,m0m1m4m5m2n3nKnLnMnNnUnVnnnonpnqnnnnnnnnnn퓇vjjޅhnpEHPJU jдC hnpUVmHnHuj?hnpEHPJU jʹC hnpUVmHnHuj|hnpEHPJU jʴC hnpUVmHnHujhnpPJU jhnpPJ jDhnpPJ jhnpPJ h<hnp hnpPJhnpjhnpU"nnnnrrsrrrrrrrrssssss-s.s/s0s:s?WXYZ<=*+CDEFXYqrstĺ׳נ׃yf\jKuhnpEHU$jC hnpCJUVmHnHujqhnpEHU$jC hnpCJUVmHnHujmhnpEHU$jC hnpCJUVmHnHu h *hnpj;hhnpEHU$jC hnpCJUVmHnHuhnpjhnpUjchnpEHU$jC hnpCJUVmHnHu 2345:;STUVʖ˖̖͖ĺק׊mcP$j*C hnpCJUVmHnHujhnpEHU$j'C hnpCJUVmHnHujOhnpEHU$j$C hnpCJUVmHnHujՄhnpEHU$j!C hnpCJUVmHnHujUhnpEHU$jC hnpCJUVmHnHuhnpjhnpUj~zhnpEHU$jC hnpCJUVmHnHu !"#$>?WXYZfg—×ėɗʗ鹯霒ubXjhnpEHU$j;C hnpCJUVmHnHujthnpEHU$j7C hnpCJUVmHnHujhnpEHU$j4C hnpCJUVmHnHujhnpEHU$j1C hnpCJUVmHnHujhnpEHU$j.C hnpCJUVmHnHuhnpjhnpUjWhnpEHU" -.FGHINOghij23KLMNљҙәԙٙڙĺmcjBhnpEHU$jKC hnpCJUVmHnHujhnpEHU$jGC hnpCJUVmHnHuj`hnpEHU$jDC hnpCJUVmHnHuj*hnpEHU$jAC hnpCJUVmHnHujhnpUjShnpEHU$j>C hnpCJUVmHnHuhnpGH`abcϚКњҚךؚ67OPQRjkĺק׊mcP$j_C hnpCJUVmHnHuj}hnpEHU$j[C hnpCJUVmHnHujthnpEHU$jXC hnpCJUVmHnHujmhnpEHU$jUC hnpCJUVmHnHuj<hnpEHU$jRC hnpCJUVmHnHuhnpjhnpUjhnpEHU$jNC hnpCJUVmHnHuΛϛ FG_`abde}~鹯霒ubXj|hnpEHU$joC hnpCJUVmHnHujKhnpEHU$jlC hnpCJUVmHnHujChnpEHU$jiC hnpCJUVmHnHuj>hnpEHU$jeC hnpCJUVmHnHujhnpEHU$jbC hnpCJUVmHnHuhnpjhnpUjhnpEHU"ʜ˜̜͜ל؜:;STUVWXpĺmcjwhnpEHU$jC hnpCJUVmHnHuj hnpEHU$j}C hnpCJUVmHnHuj! hnpEHU$jyC hnpCJUVmHnHujhhnpEHU$jvC hnpCJUVmHnHujhnpUj5hnpEHU$jsC hnpCJUVmHnHuhnppqrsuv ./0167Oĺײןׂxe[j7hnpEHU$jC hnpCJUVmHnHuj3hnpEHU$jC hnpCJUVmHnHuj*hnpEHU$jC hnpCJUVmHnHuhnpB*phjd!hnpEHU$jC hnpCJUVmHnHuhnpjhnpUjhnpEHU$jC hnpCJUVmHnHuOPQRǞȞ2345KLdefglmĺק׊mcP$jC hnpCJUVmHnHujOhnpEHU$jC hnpCJUVmHnHujKhnpEHU$jC hnpCJUVmHnHujGhnpEHU$jC hnpCJUVmHnHujBhnpEHU$jC hnpCJUVmHnHuhnpjhnpUj]=hnpEHU$jC hnpCJUVmHnHuԟ՟֟ן$%=>?@UVnopqvw鹯霒ubXj^lhnpEHU$jC hnpCJUVmHnHuj+hhnpEHU$jC hnpCJUVmHnHujchnpEHU$jC hnpCJUVmHnHuj^hnpEHU$jC hnpCJUVmHnHujYhnpEHU$jC hnpCJUVmHnHuhnpjhnpUjyThnpEHU"w àРѠ*+,-23KLMNhiĺmcjhnpEHU$jöC hnpCJUVmHnHuj܀hnpEHU$jC hnpCJUVmHnHuj{hnpEHU$jC hnpCJUVmHnHujXwhnpEHU$jC hnpCJUVmHnHujhnpUjqhnpEHU$jC hnpCJUVmHnHuhnp֡ס*+,-BC[\]^xyĺק׊mcP$jֶC hnpCJUVmHnHujhnpEHU$jӶC hnpCJUVmHnHujhnpEHU$jжC hnpCJUVmHnHuj<hnpEHU$j̶C hnpCJUVmHnHujhnpEHU$jɶC hnpCJUVmHnHuhnpjhnpUj1hnpEHU$jƶC hnpCJUVmHnHuJK!$AT`\˱~J˳۴mݶ*gdnpgdtgdnpgdnpgdnp$a$gdnpgdnpgdnp()*+[\tuvwţƣǣȣ+,鹯霒ubXjݻhnpEHU$j涒C hnpCJUVmHnHujhnpEHU$j㶒C hnpCJUVmHnHujhnpEHU$j߶C hnpCJUVmHnHujhnpEHU$jܶC hnpCJUVmHnHuj1hnpEHU$jٶC hnpCJUVmHnHuhnpjhnpUj̤hnpEHU",DEFG $%=>?@BC[\]^cd|}~$%ĺmcWjhnpOJQJUjhnpEHU$jC hnpCJUVmHnHuj;hnpEHU$jC hnpCJUVmHnHujhnpEHU$jﶒC hnpCJUVmHnHujhnpEHU$j춒C hnpCJUVmHnHujhnpUjhnpEHU$j鶒C hnpCJUVmHnHuhnp%=>?@A !9:;<T`%&>ǿǬǚo_XQFQjhOihnpU hOihnp h_$hnpj@hnpB*EHS*Uph$jC hnpCJUVmHnHuhnpB*S*phjhnpB*S*UphhnpB*phjKhnpEHU$jC hnpCJUVmHnHujhnpUhnpjhnpOJQJUjahnpEHOJQJU$jC hnpCJUVmHnHuhnpOJQJ>?@A Zڱ45_`v}~Ѳ)ɳ۴789klmŵͺɵɵ­¦ɵɵɗ~qj hI$hnph~O7hnp0JB*phjh~O7hnpUjh~O7hnpU h~O7hnph:!hnp6 h:!hnphz(hnp6 hnp6hOihnp6 hz(hnphnp hOihnpjhOihnpUjhOihnpEHU&jC hOihnpUVmHnHu(ŵzƷҷI{k!PIgԽ&ԿտƹƪƹƚƇzƹƹƹƹƹƹƚƚzhOihnp0JB*ph$jhOihnpB*UphjhOihnpB*UphhOihnpB*mHphsHhOihnp6B*phhOihnpB*ph h~O7hnphOihnp6 hOihnp hnp6 hI$hnphnphI$hnp60R&ypuh]h&`gdnp & Fgdnp*gdnpտvмhthtmHnHuhnpjhnpUhnpB*mHphsHh<hnpB*mHphsHhnpB*phhOihnp6B*phhOihnpB*phgdnp:&PP:pnp/ =!"#$% DyK keywords@ieee.orgyK 2mailto:keywords@ieee.orgL\DyK yK jhttp://www.ieee.org/web/developers/webthes/index.htm&Dd @<  C A2tjUBEBr6`!jtjUBEB!d xcdd``d 2 L0" c,NFMc`cjl@?H @fM\PZ@pT9 P\H+21,P1&F 0H=LP-! 6n3OXa*?qb*TF  T&n'έ @Syu bݪ4SXJJĺu%n*"iFf0uECǛK="o( Ҋ!d xTn@^I UZR9URO­rubZ^'MT*@  8r>@];N*U&;3zdl 2AcOExޏI㺀sCq(Ǩ+J.Y}5vyH !<"tSa4W@Ț+Tu\1eI 1鐘5̪G# wv ۤbb^]oCjJM3ߪfɓ.씏;SP?^/p^n58!abF{Y8 h"w uBzS1&B'ЬCr̼ 8K}>(7b^ sESmd/ + Ecem]^"_N~9/9y1C2 O6Dd <  C A2tj[Xs)6`!ztj[Xs8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?ȱ ~x V8p( TY(t36C7 M-f86?3v .@2l$H:`3CbׄE;ۆ` !!XxTn@BE UjOh)ZZ)rȉp^'ήqJH 8r#G0vGB0;3|3Ϻu‚/+ ;_}VJ?,z>^v >yvV6|6d !T\һcZ7kv2~:'Jqa ìa2#{7јkʲ2@(6~k/n}Aa$6豅 T2QO(,kk6)4E\C9o {6kՕqBD7AWFk}rB "m(sD&85y}crXq$ n#^L61.U/PuG_&+W21MR/fy03=t#}#,#Rb eO5^prg̿vQh솅v!WEijf;Va骰L!5{T4R.D͍}.8˽a4J~A#NefmuXwgǰjCq(.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wS?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ Root Entry F~@Data WordDocument<@ObjectPool$;=~_1133687689F;=;=Ole ObjInfoEquation Native : #'+,67AELPfjnu| "I(x,y)  D_1133687692 F;=;=Ole ObjInfo Equation Native (_1133687695F;=>Ole ObjInfo Equation Native 7 I(x,y)( I i (x,y) 0iD-1_1133687698F>>Ole  ObjInfo Equation Native  D_1133687701F>>Ole  ObjInfoEquation Native :_1133687704F>>Ole ObjInfoEquation Native =! Times{X iT Times{X i =0,1,K,{M i {}_1133687707F>>Ole ObjInfoEquation Native p_11336877350"F>>Ole ObjInfo!#Equation Native (  N  L d i_1133687738&F>>Ole ObjInfo%'Equation Native (_1133687741$,*F>>Ole ObjInfo)+Equation Native 4_1133687744.F>>Ole  ObjInfo-/!Equation Native "tX Q i =(A i ,F i ,P i )= d i =D 1iN _1133687747(82F>>Ole $ObjInfo13%Equation Native &Y_11336877506F>>Ole (ObjInfo57)Equation Native *p A i ={c 1i ,c 2i ,...,c Li }$ R d i_11336877534<:F>>Ole -ObjInfo9;.Equation Native /@_1133687756>F>>Ole 0ObjInfo=?1Equation Native 28 c jip P i ={S 1i ,S 2i ,...,S Li }_1133687759 `BF>>Ole 3ObjInfoAC4Equation Native 5_1133687762FF>>Ole 8ObjInfoEG9Equation Native :@$ R d i S ji_1133687765DLJF>>Ole ;ObjInfoIK<Equation Native =8_1133687768NF>>Ole >ObjInfoMO?Equation Native @sW #S ji =R d i j=1L U: S hi S ki =_1133687771HXRF>>Ole BObjInfoQSCEquation Native DV_1133687774VF>>Ole FObjInfoUWGEquation Native H. hkJ F i :R d i A i8 F i (x)=_1133687777T\ZF>>Ole IObjInfoY[JEquation Native Kf_1133687780^F>>Ole MObjInfo]_NEquation Native OTc ji" xS ji  j_1133687783PpbF>>Ole QObjInfoacREquation Native S>_1133687786fF>>Ole TObjInfoegUEquation Native V(_1133687789dljF>>Ole WObjInfoikXEquation Native Y8 c ji d i  x    4! "%#$&('),*+-/.0312tv56789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrsuxwzy{}|~ <  C A2k!G?Tb(TMzG66`!?!G?Tb(TMz<# xcdd`` @c01(Y,P1=F (`d?3%$ lʛ27 27)?ȱ~x V8p( TY(16C7 M-f86m61pSbzXM&֭qĺhnL[5] 3y7XC#\HbozI'6#xTn@;iKK TT*'4[I6="!h@UBNLuDjOH7/@ ̮ĕ̛7of׻r܁mЧ0"64% Q*/qk-yWk XW~' J<*zOq4ϿWr)T*>pIxGO $N6$`s*PLCZTqKZzedU3ZFȪ زfɝ.}3d~<棽^a݀hS$ Cr%dZ]Ѷ.= M4qᾍ7 H$(܀{^ s:d.z"]̅(.D4h_?2a a!]2M{Xǀ{l4sC&*{7эq6y(,:ςi3SqѬBCYutwo԰No1~0|ZOyS6UgQ~!!;(;>2)\ÉHs|ID/:Fƪn; a,l0G6ƒ=su0šM-W'Fs.Tp`QIJpw.5av寅|#ztK+k~*6l~a#_ ?WtDd <  C A2[\bA*ci716`!/\bA*ci<<  !xcdd`` @c0q3؀Y,P1=F (`d?3%$ lʛ27 27)?Y'ջ@pT9 P\H+21,Pfl4M-f86M]v@.Fb؇ˍS.ʅa*v&֭@3uk?)n$֭wu Teb:XMc 2 %֭@S.PJu<~v C("iD M=hN<  !xTn@^?F U*j)F q~^"("ᆄ{:k˻ISJHHUE}ܠg:#}3^"m[&F#WmOn|%) 2f$/TD z: 2qt.Ff|"C+9,y=yEGШA2R$0ֱ N(7`MT~T\od?dBI1׏Y:bt\sQ DuE*2&EQ_G/NN2UդЍTdL”z"NOxL!]!›W쬲+7S^X~]iSUnUf.$B;.65#*%S!2MS d[^|f Jx}AP ܀JHPݚ~YUmg31Q4$tBO 6ΔB7ՠRZ>v_XY7s<4߅2ϊ9^x}]O7}tDd <  C A2Н$)B^6`!Н$)B^ xxcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?ȱJ~x V8p( TY(16C7 M-f86+ i`TF[nTqb*#L`r#HJu<EC +2n[Q- xxTn@;)[T\@TqM A {,u֖w 8!P$PfNJ7ߌ3[ Fր{ $0f=|־uZ[WbY#Vfa6<td[PIoZY w9,{jP Vgl與76*8JS.DYUN[7HG;Uw"dˢa =GxQJB,Ȭ1L+j:M4y 9DNTdp2s7139/f6, ?+JѡkqcM (A@E!JT\13IVtIT?=){mvjv !-8Bjl&$ b Gxpjrr8jnǝoӕQ[\qbPQuf?_]NDd TT<  C A2pGk׉!sqږ:L6`!DGk׉!sqږ:vQKPxcdd``fb``baV fx,F^equ02Ȓcjl@?H @fM|C\+JAl?z+8U`fҊL M$ 27)?皆4F0 ad3 tT\n5!~5Ĺ!ĄЭӈu-ЭLĺh)y@d:Kf+[-L-3u/T[bS,RRRLbZ t$W yzH 3 hC#I_OA6QKxU͎D.{fR *8a6c{vR5"ȴe]h ,J !Z=c)0Eߑc,%cʳat^GuM0д*u,2KJk r2DS˼ LI1UeuiV)w<kD8 +F_rҕd>!Z@lmEU *WuQE}fO3Sߜ5Ց}/{BR/4Dd <  C A28^=,"$6`!x8^=,"8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Pfln!Zpl?61pSneBs7#HJu<|4%C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp5Dd <   C A 29.hr^v(6`!y9.hr^v8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9  n?z+8U`fҊL ad3f%4F0n*#ЭLhf R?3;3JCsb$c?ɻM` h!WxTn@BȢ"UjOh)ZZ A NL]kw 8!p<G0vGB0&7kýsXHEr :I: q,Ote|/+mع*`vC[cO 6]経ߨR\TF0N\,fn~pMUVhCrf7NZ}eOCb|5zAc >$aS?rn{1_'4: G=vV6k)R4R%D7AWR>K͸Os%6}Q*R4Ɍh_"3N*r8D){˂q1C3F"j"1t {їUfe(M2/fE0,*=<TW9Ʌ\4Sj(~楓g75F/fӅ_6R팵0XKy"r@ag,I>>d\pK`n4S,C@(%[pZ;={ÖەgƋ#6kQc2NS,⚮;g 'SNo+wg1O0ކOPlDd <   C A  2*ǕGT<47!,6`!*ǕGT<47! !xcdd``> @c01pY,P1=F (`d?3%$ lʛ! 27)?ڴ@pT9 P\H+21,Pcln!Zpl[Mc22uOݸ!2Mde@ F 3y;FCTOWbTV7# !xxTn@'.TԞqHEKEሄz"#rdky7nP Rp|@ĕ̛7o]7.fj`nky/`4IoZ vGr$Wy~LF_o) .:-^yy6ChMZkkUNs7)&QB?SMغc4j-0؄+uGXq_d"48C!5.w0Uqz2@ɇ,{Qi߫P4_4i u|zm%J"In"YWqŒpn /xH쿵oLhCX %*dǙQ-SLDzÀt]6I0rhF衐Oaj 9Iv^P%g%dbh7y[3%SШ9TTa*14޽mͬ"ZǍ"/Ҵèܶ| ("Yq@CMC s΢j/4tT$wG@=aXuۥY^(y- y*Q5m /f?~?N #Qo:~ہ5rl;\V Dd x<   C A  2)]}EK3e16`!)]}EK3@F Jxcdd``^ @c01Y,P1=F (`d?3%$ lʛ27 27)?ȱg~x V8p( TY(16C7 M-f861pSne'^SnFpS3xLρn=K[mnf*6V  D9b4Up%%\U [V[Ո [uT-B ĺĦxb cbEb4՜4@J%6 \"֭@:_CA*gqW {CZɳ6F xTn@'-S*q@$q6* 8Tj!HkN* !*!!O0<@];JNo]9A.&gaB24W9+cw۽@ET .U8 KQP ЩuG>Nj{XV**+~&_AwIxGOuBlHNV`T73A]֡R0ȇSeR_dvMMi PzֆL,Eq*Zo朳s~LJΐvM|ak L0D18@)2薥CVFnOlb8\NpƦ5z-^ g1?JTn^a t\zz{HI]ǼPŅlF!#ٺ=KŃv%C'VwD<$M8+GGy45S0uGw(xX 37rt9}ZO)A*Y^~T.$#DyeLJS&K`Vn$s8wefƚ$M0poƾIVEThn`F>{-k@.t#0Q~dl4ޤ&E|aO2fҎRpEydmgckoްgI0F!a^6'ogoݽC3.V:/ip&\ҫKt^N:~ o-L~қ!vDd 8<   C A  2A*I BV\7I66`!A*I BV\7I>6v 5xcdd``gb``baV fx,F^equ02Ȓcjl@?H @fM\Pື@pT9 P\H+21,P4&C $3784mN1pSRxġnTr#TF  D.bj t1!tb4U#k7n5d``f*H[Me!*n!֭@S9u#RM[{uT1n}˨@*gC\΃h& [e6v xTMO@~eAW*bH &7ܯ셃(tn;m:Mx 7D1=2sQ*OwQsyQ0"fLF>E7)(Jůw0^3DCƿ&`!LPTkQs=sO̪첶q%,h!gVQl3'ёۅIgڹϔQi P\UFwl/MkZ8 r֦8`T֫ QN`فQa60[VF)n 1/$n7acCT\*닄ē0vIH٬ngL>H;6n/€v*eCv]Ig::at kz &v`d|Dd @<   C A  2j׭9(?6<Fa<6`!>׭9(?6< UXy mxڕJ@$HB**BM;AqUP]"ЕoE7Ri\ŀ1ɜf|"P4C$F4,ic7t19WR*&'TJ1Ӆ*X [g 8}5WiҶ׮ߡ^Gkф :@2e\'--Reݲcڜ֫k%Z]v R[*=;Z`WTAі %PLf+Q?'›1T_ ѵ0EWC|jc19rvAa8{pvXv鏕[^V"r.nmi"p^$ܔ^cՀ\ΘOC 9-,$mD'J> d {T:A-P4ZX~į ԗ*5ǮF*Y+eX#K\$P)o-[1JSus$N DǔN2b:|u8ˁbd|:}PTȁ%wu89ٮH%cMӉy0;=~&R8=De|"Qa e1({cwf?Crt!C)TsAynCY0H0glPPp̍$*\0+1 IG)(jݩa|}~$ [nNvZn{([culu7h۴cI PN,\)W]<<|;Ľ }֝҇MXyÁ*96?TyҴ]ѓDd h<  C A2:&|+x=DF6`!:&|+x=D R|xcdd``> @c0 1Y,P1=F (`d?3%$ lʛ! 27)?ș+pXr.P5;VdebX0i{ 4@Hfnj1qi9\?MedHƟp72a1E1nj?0aL6 5l3  GC>3 hHɤ R|xTn@')O␈i4 BP=YM]˻ISJH\x 8 x&2vWB0oOf ܆UUX ɪdr!\:IR:/s9r! QTȞgrJNrߚnt~*Lϔ4OEʙQIFr$ձDf ʲ#kyg̿|""ϋT;c e',UaB k36 I6>qK;cn4B,CG9@(8;WǙpm5Sii-tٙ#^ aunbjMZf{ߪ-߱kڻvs@rBwCu+g% ^z_kDd @<  C A2cfLJ7 lзZ8x?K6`!7fLJ7 lзZ8x V lxڕ?K@wIMSIPcIppRBɿA0~w.Rh:H^T.q˓'fhD2OeI{A|4ɨBDU6!@Ll@&K=e ѨZl\[m>?{gl'X-{9瞙/ :< &\рbZDŽR/ w$qԐ iuؼv.PI ܯU@Y5oUm ߨiIZE3sHj}53FuA=9SfT$QZiHfUeFNT_3axG^xĎQ&&Q!gǔY$ kг;ip-+E 6"6 H4Q2"124Xfr:a.Qr"{H\(Ru4I#FфQ(\^*8ƐtcEx4 ɧnhÞqJɒy+;Q07LN80S2uGw"w5yg'0~rb:sS6Ug3-\HDH&!9d4;>`RKwl(R{!(X qvbdm>%uE4 ڥB A~94RlUĖe#F= C n_dto^gJ״xK]󺲙݂NT{k4rL-l6io/Ž<7 eFʾbumQl%2/{ֶdDd h<  C A2 G)/(7S|Q6`!G)/(7S |xxcdd`` @c02Y,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Peln!Zpl /v A <А@7.FpSS؀a17!zL>0f R?3xWCwή\][ |xxTn@'i %TNhH%n(G$K gxRgmy7iP }8PfNJFdow]\؂e.`n[6g`p4IwhU ۇ*/ҏK-E Pf}!\W~a*J> d {T:A-P4ZX~į ԗ*5ǮF*Y+eX#K\$P)o-[1JSus$N DǔN2b:|u8ˁbd|:}PTȁ%wu89ٮH%cMӉy0;=~&R8=De|"Qa e1({cwf?Crt!C)TsAynCY0H0glPPp̍$*\0+1 IG)(jݩa|}~$ [nNvZn{([culu7h۴cI PN,\)W]<<|;Ľ }֝҇MXyÁ*96HpЪ]Dd h<  C A2( oeL%V6`!( oeL% x|xcdd``> @c0 1Y,P1=F (`d?3%$ lʛ! 27)?ȱ:@pT9 P\H+21,P4&C $378ش@>Mc223@C7Ox!2Mb3䡆 #>u<GyCB]$@2 x|xTMo@'˗*B#-i7􊄠zJ8#^':k˻IJH/(@];̛7oƛ]W>Vb.`nfr!\:I3hmw;G/-bUgsxE]XtMb2l b޴\7s5hQqDH 8.\+1䊺lAR&*<!kXj9G#1. y!6qYǨF]G`^Ec q]! ML)cE/&M%rq-R$$$ޡ p>IS-ZFrr~;g3֡k2҂LV>ӌzU WvztDM;p,N+܄vzxg@e;t "`? >xi @T]]0˸$_;Pp.w@W.uRZRd.HWS1?϶~ɷY4[n9بPDRYll({'RTb=jӓވ xNqDܶ6nژ%qrP,`w;B̀%d]o62"&ـ"(f BmASG'j[n+rɜM,$FyBٲA2䟥c1#1=5a1y['(,sD4s:M,KaYA"򿢌y"OELH&4TTG$^VY_X/0 y0C]HiSUݲِa](DH-xԸ>Yߵб]lSRzBv;rPH~hK745*"%ni7 CȉOhu m%BJ6"Im"+nkŽ65QVrCd M-lVBw\)gYt\sV!qI/OMn^!KM/"\gWU8T^h2^u+Kk * unMMQU7bli Uob$Dd @<  C A2cKyv뾪2&N=p`6`!hcKyv뾪2&N=< xcdd```d``baV fxl,F^equ02Ȓcjl@?H @fM\Pq~x V8p( TY(L`ln!Zplk^_Mc223@C7O!2M`g2؀NL8]X!.Qtbݚ tJx(.r[c@!Zzdҧ?3:sCPb-gF><#< xTn@'mJT)Q s$Tʡrul- *!p=g5xI\l|}ߌr]- =IQ|eiQB˳}~>}Р8Vh;51̩+#9,(CJ .Il-Xo{}&*JT)KH5O! R?ASEW-!CuFmϊúG5G1 g]Bf^۩K7k ,W2}v1L.|3f40f^ QN2SoF8oCf־Il{"&uS030(YٴL2W!wcq&i(]Eu$e4bZM M_OK9NZ S/bVD'Iy0i&PRқWjfoDwOrU,'$JՅd$$ 7YߤFǤ% Ĺc'H漢pRdeb8v m`ETǦYRL6 \%h4+Ԛ*qg֪IZMNUeUF֜'A:M6>mG/3GGOڇt.1_(jw,yeN~/0u~Dd 0<  C A2"WT`xd(e6`!"WT`xd( `mRxcdd`` @c0A,F^equ02Ȓcjl@?H @fM`[\PL:@pT9 P\H+21,Paln!Zpl<'(v @2ƣ1nN DUXfME cF 3yMXCK7nͶr@l `mRxTn@;FU*qJ4ih A9D!!^&ήnTR%$pD>@];+U0We6`վYʆr><9NRaLJ݃wcX^ߒG#Yp`,Þ*B}h0Yg`rf>FiʸJDĕl"FC&Ta[Z%\}ɺ;ek{]8J7ch~|`{Scjwqcr'tQjv ^hscvP.QJkn;8}HD7~WR9Df̣T]~$e`AfJPo(2F<r8x)c*Fg>JFo@yv=h= %#E̍T^ ^F)HsG6Sj(^&zQUh@ߗ'5%R5 ל LӅFCХ1`%SIh QJ qR;ܫa6erà:tuLM/o7&XMZ`)tee<|=ǣ>keaٲOC}J˹'w b ^Dd @<  C A22 =st?S]0j6`!2 =st?S];hn )xcdd`` @c0 0XY,P1=F (`d?3%$ lʛ27 27)?yŰ@pT9 P\H+21,P4&C $3784^AMc221@C7OȈˍSf23p`96>!֭V@r1@2 |DЭOuk ߄L|r0Hn}F[C@pe!֭@S ĺ,$aT3 C+anV@\atIAE;hn xTMo@;iM[ЈS'nJ(^B\{ݘ:k뤭J%D%$$ #_3̮ĕ*Xiyo<;u r,,WpEFcK!.vŴ.Ҝ9=Р4`eq/ O NtHT3A] N4_y%R6R:8e;K)X҉/jVō}*v:#nWCeZ5t}'=k |`brOI0[qݬ驴p]WLbN6-IlTcF4qMrKB8OJ̡RMŮ/a&Q_?2'ppؤk.;Hb+ mXc($,ځS;#?fNǩ9pd&g(P`+5*Yg;މ\2[`ü~1 d06[er>2JRhʘlnmP$}>w.h03`GXt@U0rڇ_s+=_`x4ezqrX gJ&Kndl-La.<7Sj(P2f;c _. )A*Y AT]HDH!(?>3)elQSDCQJQD/>ȣ ufnz؆nQ-g4ra͕.:{JsSnF_}]dM [)mN{RyG,$΄QӼȟ.^|qS&1CJ\|A`w6=,W4ztaF?1~boV32(=\EDd lh<  C A2J-bJMC&t6`!-bJMCHp|xcdd``Vad``baV fx,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(Lblj& Zpl3v  <А 1n92aw n *5'^Sn"֭Y @3aL6ixd[ҧ?3W8j C/pCx'Up|xTn@;/5JHtS"ڦIܐACO 7$ĩ4UT 9q<K68G!@"뢷5G=%,M*o^0.(Rw⤦8[s͠UYPJ1U+.W'ԑPg `1gt(s\:j P<棽^JuxUW(P^ }P ̟(ŅId$ df9G޻׏\S=TRf{< opؒ3 @c0 1Y,P1=F (`d?3%$ lʛ! 27)?1~x V8p( TY(Lblj& ZplZ#v @2BF&,ˍSf1d`QÆlH:`EC(YΦA- R|xTn@')ODI+A9D8#gN6uv-M*!q@Hp ''x`$;;3|3lX pt$0uBut^)|/s=x,y!| I;p5﷥3jS?g4$JH[b1[ Wb5eـP#cP)eƒղ7$ T=aY˦qAѽBa{C x_KIq8F=}, ~B3{Y 8fA/4: N7R&J4E*X9Q,\h  m(z D%KQfyF*/ܱ\vB.Y@~8N0oB!C>A6}>2h=w%cCi~*Lϔ4OEʙQINr$ձDf,gGiy^d̿|""ϋT;c '>qK;cn4C,CG@(%]pZwzs۠'*1j5V6>Ole ZObjInfomo[Equation Native \4_1133687795hxrF>>Ole ]ObjInfoqs^Equation Native _(_1133687798vF>>Ole `ObjInfouwaEquation Native b(  NX Q i =(A i ,F i ,P i )_1133687801t|zF>>Ole cObjInfoy{dEquation Native et_1133687804~F>>Ole gObjInfo}hEquation Native iS7 "Q="A,"P,"F()M "A=A 1 A 2 KA N_1133687807@F>>Ole kObjInfolEquation Native mi_1133687810F>>Ole oObjInfopEquation Native q4 R DM "P=P 1 P 2 KP N_1133687813F>>Ole rObjInfosEquation Native ti_1133687816F>>Ole vObjInfowEquation Native x4 R D+ "F:R D "A xR D_1133687819F>PH@Ole yObjInfozEquation Native {G_1133687822FPH@PH@Ole }ObjInfo~Equation Native :_1133687825FPH@PH@Ole ObjInfoEquation Native (  N  x_1133687828FPH@PH@Ole ObjInfoEquation Native (_1133687831FPH@PH@Ole ObjInfoEquation Native (  N N+1Q b 0 =0b 1 Kb N =D_1133687834FPH@PH@Ole ObjInfoEquation Native ._1133687836FPH@PH@Ole ObjInfoEquation Native m_1133687839FPH@PH@Ole ObjInfoEquation Native (  ND d i =b i -b i-1_1133687842FPH@PH@Ole ObjInfoEquation Native `_1133687845FPH@PH@Ole ObjInfoEquation Native 4 1iN I(x,y) i th_1133687848FPH@PH@Ole ObjInfoEquation Native 7_1133687851FPH@PH@Ole ObjInfoEquation Native 7_1133687854FPH@PH@Ole ObjInfoEquation Native : b i-1 b i -1_1133687857FPH@PH@Ole ObjInfoEquation Native ;_1133687860FPH@PH@Ole ObjInfoEquation Native ^B I b i-1 b i -1  x  y_1133687863FPH@PH@Ole ObjInfoEquation Native (_1133687866FPH@PH@Ole ObjInfoEquation Native (_1133687869FPH@PH@Ole ObjInfoEquation Native (  I "I oUNVX5unwR;|=n\޳\J@/1Op~_GlDd <  C A2̈́cfP9ib#.Y~6`!̈́cfP9ib#.Y !xcdd``> @c01pY,P1=F (`d?3%$ lʛ! 27)?ȱ`~x V8p( TY(16C7 M-f86Oi`TF[nT=qb*#L^` R?3ECTOWbTV7# !xxTn@'.TԞqHEKEሄz"#rdky7nP Rp|@ĕ̛7o]7.fj`nky/`4IoZ vGr$Wy~LF_o) .:-^yy6ChMZkkUNs7)&QB?SMغc4j-0؄+uGXq_d"48C!5.w0Uqz2@ɇ,{Qi߫P4_4i u|zm%J"In"YWqŒpn /xH쿵oLhCX %*dǙQ-SLDzÀt]6I0rhF衐Oaj 9Iv^P%g%dbh7y[3%SШ9TTa*14޽mͬ"ZǍ"/Ҵèܶ| ("Yq@CMC s΢j/4tT$wG@=aXuۥY^(y- y*Q5m /f?~?N #Qo:~ہ5rl;\V 1Dd <  C A2[Bb'cE}6`!u[Bb'cE8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?)cb~x V8p( TY(t06C5mfM-f86Si`TF ݌`#A*g*3hC:ONlP*4~;^ 0RxTn@]*@ԊqJ4iZA/9 ǻM]kw: C܀ok'q$#3x3+b!wkYH!_#³ڰ>9 _:cy>'r# G7_&#E EqMwk d>p^Ih0U Xj?S!~0pUHk"Wm֣`{0BXq,6ZP2I@OG =kAy9o YZ+m+Ra^B&7 5~Kx쿹o,$JeZN2#1Z7G{ND5Ģ|lBDЅX0/M1Fb@VO7/2D*N ^'WвTIaƊFYAr!!T*Cih}䰬 ׅ_5R퐵8ѶLׅEǦb~и47d<,R,C%@([G0/;{A-KLS;:5^n?wݫFozyNO2Dd <  C A2y jT1͜~6`!vy jT1͜8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?j~x V8p( TY(t36C7 M-f86 Wv ~@2l$H:`T74CORQtj[3Z=` R!TxTn@B"*'4J-i~r NL]kw& 8!' x N<av$`$'7L* XJE2~&R.n8gC22xC|YU8۰wl<1##>Wr:eF}O$2Ywdb1p vo1Te*eQ.li^}<1opؠ5P8F=CiSjfʧȹ|(O3\۬DHJ5&2\j}J -9c̉He^cQ Ǒ(}xY0I0bhFH|4 @mT$d92h^irz1Ka^Ioe$B^rL:WNʞ7v^*, jJP0X]Y"2@awlQSl}857{yXsU  VPN~FW^R}/ϰ=ROoDd x<  C A2ݒ).ϦM6`!ݒ).Ϧ@F IxڕJAϙͅd)S b*O X ] -lHqvL>7UDVȌ"fEVUU =nJ1Ϟ`&*KVm0oRndqt}wKIR࿼ΓfkS绸uhښ{Ң~5Ҟ v}ϨLpe/_)3B,rK]py& z"u=)͵Lz*u݁k[\CkmQp==P=p.uAH{ Z[p-Rs5C0H'85Q^6F xTn@' ! 8 [$q6*Q rB8TBNLuVj%$$p<@];J7̷ݙY9@.. d#xpaкBi3eQY%oާ j07f 7@<^U9 Ïץs*3~!@wIxGOuiJlHN fUA:4m*VlY T}[jEmHsN3@śuȢ2yGe5l圙_94SRogPxG^সǺ>pI({Y,nY*v.p١Sf:!Θ&2\eM^OG3%}J7^ &:d.z"(.x4"o ;Ewu( b =0;API$9Q|u{)&KN|?ԉx0HFQdGy4SUGw(xB 3w,'t}ڝO(A*/Y^~THF$ u]LDR?c? 98]%Eo'zAfmm'8uE& Z6Qucj{ ;&6[ "jY#LЍDA 0pxUoTݦZ]Ulر(R+ k?7γe;i'*$8!y 8rc8p2#Orfo?ǻ0CH5~2iACi'ϾPY}6k^!h𠭰akiUG ÏϷ+meHk0Y Yfi,KPU<:@U7ɽx* 5*k1V*)2P9tα }`v*I#m6>}k[Gڗ^L>,\ Eax*bLtqD3RX10dl,F^f%Ĩ.JTƹ4Hez!Q%AUu |UW #>'FDߊ*n5ZDUR(/'Q@ƹ`J0;xU)rQ jKvA$îj!IclaoК;:J~UZ^FZԢ5Ξ_N&9|wy{r}0gE46^EJŠJPy{D$Zr|*B-h?֔]~ +dE*b^X;a5 & ܃u0oŠФz :Wk8z ֤ȖݰHՆa+ܓyxn.a}ObuXn +{"w`S#dzm Svf{6f"I0"ggY8,J;SòY0R<Hg/v=z x/٘& XPRsqZA:;RUfֻ1XŊhwE庻DUH3U1t[~Qs9R  =\qDd @< ! C A! 2]Bggܡiz6`!]Bggܡiz xxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?yǥ@pT9 P\H+21,Peln!Zpl 1pSx!'^Sn4FpS]؀a`#A*gFmCjY"P[; x|xTn@҅VԞqH4A9R8"^'ήݸB%$NHo(_9I\ J̼yfq`6+pM kFud; KXgd+(w`{tԗ1 ljnXulCo`8MЉfN|܆[hUقjN*))^5kV^5Nz8ڟ`|8F5Bk7[씨|-|/7u {cE] g9tDrcz%3Z.Ԍ77[r)MX%2)EG-Ɍ Ǣ#Q ``@dzAJXdZ%cM̋YQ* JO(徖49\fJ ߾k͌}Qh΋t!M):Ts~[MV?ݥ^i,ԐZ. ]l;}*(SiN;֜ӕw7ꚡ'ľ?_W!Dd @< " C A"!2Z`kd9>f6`!^Z`kd9>lc0 zxڕJ1gvmkEfI,EPz}ы^x?k\J4K6a_&aJ^4J%{կQ9:8Y[K1[+`>Ԉ{b3h&ʧz{EvsuqhNzYmK@qguF7H\!o+GgT]d?]V`|mQ}%1PloP1BkVI%i䒚&PCOgW*2e𴫪hq+!ڡ ])CuzOeߨ? yx'^N`7έQ2a߁5Cu>qI-ca[Z䞤%wsn R?Ʊݵ=]y0*CFIX);KJ3=sXLO3B~bgx`; .caUE:C^A/mk?Q& X, MyɒH@XHF/ VBɬꛈ#{Nyӝ&.IxN!+fXR.1V4:iH.q=Y`_8nIє3 q0fE3EA"ؖpeӒ qC/9ic8+KUo+X%_vQTa\uB2œI9XwlUCB+6삎X\h{&{"0ΰ{.O;z@wqa{̧GaU va|95 9\ ~qDd @< # C A#"2J}6[_tM.06`!J}6[_tM. xxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?$ eXr.P5;VdebX4&C $378ش\Mc221@C7O!2Mua`jh0TgFCCjY"P[; x|xTn@҅VԞqH4A9R8"^'ήݸB%$NHo(_9I\ J̼yfq`6+pM kFud; KXgd+(w`{tԗ1 ljnXulCo`8MЉfN|܆[hUقjN*))^5kV^5Nz8ڟ`|8F5Bk7[씨|-|/7u {cE] g9tDrcz%3Z.Ԍ77[r)MX%2)EG-Ɍ Ǣ#Q ``@dzAJXdZ%cM̋YQ* JO(徖49\fJ ߾k͌}Qh΋t!M):Ts~[MV?ݥ^i,ԐZ. ]l;}*(SiN;֜ӕw7ꚡ'ľ?_W!sDd L< $ C A$#2}*u1 ?6`!}*u1 ?&7 x-xcdd``$d@`,&^s0*(e112BYpL $ lʛ27 27)?)cR~x V8p( TY(26~110BbDh4% dqټ E $378nf63H.pSڋ??S22X1T&nm!έ A@Syk*ЭĺՅAXU4!*"L AKt?3Cxn 3qUP7 x$xTAO@~-. EI0x2/AnUnx/&JwtM4&ћx v%3}/PÂ6̛pN3}9LR̪nmn=0t?⦆)|۔UӰX~|1,RtBI1'_P%U_w~2%"♥PU"eS0 P"UkTD{XIVQƆvYgayliݕպcim{Wz'`!u "?=։-`WLE!g\,RNk Kߋe,aE׽X[xBHσ\ol]^hD "̈]B(X7e =1`H|<G|,u#trOPLL LҨ9^(4ptâR,yϽj|g1n`v8\zAhzV/QQ*gT*vEM(e~&ң\d\ ͌ʫ?X}\e;VWmX!.$Ȕ{5?vC9SPQT@GSL}$UP0(푆 폭%ɮ8ȣ5Ů<qcU7mz6]nlׇ1Ǝy\Sl:$+ 2/|ٱ4.riU7llёsZF:ԜgՖs*Ij+n;uw.ZADeXzY͚vg:y|0c;[zN9 zDd < % C A%$21PG2$9\YV| 6`!PG2$9\YV|:# xxcdd``ed``baV fxx,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(26C5mfM-f86m1pS*x!Apb*#s@S1Le֣ĺ5hԟxM5X0D9F 3y~5jCcMY{]^HW# xxTn@^;iKK TTKZR'nNA9R 9:1uv-&M*!U!'R#7 0<evFZ{owwhsYVۖLBr0pnXU>hЈVy}H_گѻ &FV&2X|~k@t'E]Q/%{WJ8A?()t9g\# GVY'EllkW^.ɟL[GUKHXMUr~4o8Ц͵5q!IF;Lu*w[aհ+g"%7J” ,Y,4z3xJ8U_S4ߒXHޅ\(L7y=>O '1 >:a?Bbb$51IvI(K\M KEQ^ gg4<;1"G3`|LC:7ʾHO!z9jDu'M \T*AdfNED&SwA&ltAz]E&EiO4Dn`]NtAk *Wr:eF}O$2Ywdb1p vo1Te*eQ.li^}<1opؠ5P8F=CiSjfʧȹ|(O3\۬DHJ5&2\j}J -9c̉He^cQ Ǒ(}xY0I0bhFH|4 @mT$d92h^irz1Ka^Ioe$B^rL:WNʞ7v^*, jJP0X]Y"2@awlQSl}857{yXsU  VPN~FW^R}/ϰ=ROo2Dd < ' C A'&2 v~~6`!v v~8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?șurXr.P5;VdebXմ=LA $378شL6YMc22TYLhf R?3nj4C:ONlP*4~;^ 0RxTn@]*@ԊqJ4iZA/9 ǻM]kw: C܀ok'q$#3x3+b!wkYH!_#³ڰ>9 _:cy>'r# G7_&#E EqMwk d>p^Ih0U Xj?S!~0pUHk"Wm֣`{0BXq,6ZP2I@OG =kAy9o YZ+m+Ra^B&7 5~Kx쿹o,$JeZN2#1Z7G{ND5Ģ|lBDЅX0/M1Fb@VO7/2D*N ^'WвTIaƊFYAr!!T*Cih}䰬 ׅ_5R퐵8ѶLׅEǦb~и47d<,R,C%@([G0/;{A-KLS;:5^n?wݫFozyNO2Dd < ( C A('2bfhֱ}t ~96`!vbfhֱ}t 8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȩ~x V8p( TY(t36C7 M-f86M+i`TF?[6Rf03CORQtj[3Z=` R!TxTn@B"*'4J-i~r NL]kw& 8!' x N<av$`$'7L* XJE2~&R.n8gC22xC|YU8۰wl<1##>Wr:eF}O$2Ywdb1p vo1Te*eQ.li^}<1opؠ5P8F=CiSjfʧȹ|(O3\۬DHJ5&2\j}J -9c̉He^cQ Ǒ(}xY0I0bhFH|4 @mT$d92h^irz1Ka^Ioe$B^rL:WNʞ7v^*, jJP0X]Y"2@awlQSl}857{yXsU  VPN~FW^R}/ϰ=ROoDd 0< ) C A)(2A,DAn}nj9k6`!A,DAn}nj9 `m!xcdd`` @c0qA,F^equ02Ȓcjl@?H @fM`[\Pc:@pT9 P\H+21,Pfln!ZplU~.M1pSne"֍pb*#Tm|LV-bj4U -!T3 XC" 鄩=B lr `m!xTn@^E*CU8%jKVሄ ޵7U@B p=8<evI\`wffv=*Y'7rBn\:N3;VInu{OŰ<ާ/d=˳ [ Y8鋘F(!Uu>z!&joX(W1V4f,K(aKHpڊI@֖>z'y,P`c>p6@@;w04nechbvPe쵡՜봞 &,A,_i'&4g> DRaDHK$WE}sw8%!/ F)Č񀍧 U{}hڞґdYŴHEQ|%cKr.Ŕ %T_>7R+2{|?(Ex$^F*4*cEmr4%H{܀IhC4Cc+%]aR|;9G괡ndȇA5 "df Eqi԰g [ICF\F^o$7g8 ˎ}k3pdDd 8 < * C A*)2WGS436`!+GS4v8oxڕMK@g6FozY/ׂ=`Nެx Q/?DZ/ "b|7R;;ټyؙ SPD\YS}ʡHeĊ<(Ŝ0$ jlz1&]+_=Y:Fe{;[+&zOn^~Wh2Obw/Ѭ"t7CRYbطnoMu}ć?*McWN\8NX?Y\z'ce<)@KNgRX44{9MR֚$k e5ٚe Rid()9iO]HYp-IYa-.--[jIʺl-KZ볔u^˿6.WsЊV=<7C]-Sw8:zْv8VxTݎF>v]p ^#`! -?Ze᪭c7.8& iUB/}n*x>@gf)Frr~o;Մ.TP@?iF=dpuj+E G/dz<ї.<Ƣ.܆/0>/zcӨ|z>:o`Dk!mbi'L. `F? CU6VhPʉ_;lM \DGbz^ J4J1$:.f9hbۄA"V I+q;͚ZTH1M$Jj\)7-R>DJKqfE*d/É,Xa@ORL1\Lf9c9 1TyfE(XYT*iZVT3R͛)5W03xNQ99͋9EvM۱fk(]8Dn9܎jjQ}ϙ[8ǽ0QJa6#oe 8Ĩ';k# 9hah9}|?+FOCPHIDd < , C A,+2%GLۋ1Úaٍ~  6`!%GLۋ1Úaٍ~ >:0= .xcdd``gb``baV fx,F^equ02Ȓcjl@?H @fM\Pps~x V8p( TY(16C7 M-f860+v )@2A@7FpS3x89ЭĺV^bByĺ5 h8Q!tnȸO[unUd 2 hVPh 0Lőu.T#lbsBbj4 -B Tg\C3`t^l=Z:0= xTn@;).TTNhᔈĎz@*!(gub- *!g(OAx 0v`%';|}.1`*LptL6kGq2>S~ãW*Oq @a V`Ӿ% uAYޑT/ީ`Y>P}\4jl܄O9l@Yb2p]D怦[iZ.= sb،vJI$-ckyq8(܀y^a g d鐺Xֻ xM Q\iټ~4!+ÀҮ7IHY7̣ih{i$nl͜AIjNO" u(9DYt܉;N73b41||;8.d fZYB0IocSl`pL$4SeDh4n%ߎ mY&iȃWa_ת^ {5, ZթζQ uc3 $k=gb,-] IHc,n6ng"[m&tt8}~GxST(-W+VNX~O=w@"$:&T Dd p< - C A-,2&߾dC?z+T6`!߾dC?z+, !xcdd```d``baV fxt,F^equ02Ȓcjl@?H @fM\Pc㺶@pT9 P\H+21,Pfl4M-f86m'v `71nb2awn*H[oVqbM[*HdT3 _&}Cfj,Ak\Gp, !xTMo@]'5jJ DJIUzDBP 9:YZM*TBBB ?ƅ#7:_H̼f<vTfz\UL'_i{׺9?Qٻ ڤB'gHY Ehn^%W dpoV}\7g*+Wǿ#ƚT ZlXetg*WKX 0UuK7ɭ9zW G- (GDckPAvk A&d v02 _k"w î 0H輢.@QQn5Q*$l,YP- *;V`[֞h8p.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wS˿hsvDd < / C A/.2ҟO @c02Y,P1=F (`d?3%$ lʛ! 27)?șugXr.P5;VdebX4&C $378ش\ױ?MedƟn0ƻ1֒ 6`ؠPJu<HfC6Ӕ @:3 RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g FV6`!6&I]3̨6gL/>:Kxcdd``ed``baV fx,F^equ02Ȓcjl@?H @fM\Ps~x V8p( TY(16C7 M-f86͒i`TF$[nTwr#TF @6|t+nʉTlnE[ PJu<g@C\BKP@/KxTKo1MB .TTNՈS ym\C/hMnISJHH$8s*#7oĉ2n^%'o^ۥ`lfr.itlL6>C_IJv溉9?aXG黤sx;ڬ.\핧' tAi1f_ jm(~>.4B'2wd`v*l]D eqTq7}W NTJNìvQވ xnH,5-60qjE{؞u^߷a{1_ԩ'Ti.sR$R$2Pނf:>QrHib1Jdl@8t,q$ҾG,&s}lb$>^i$zd>29̝ȋY^* JZ'ȑ6PVvTʞاf1XH ?kJѦ3Vb5fLۅAdLGMC76SNŒbg*>No8.->^6X[fMPl(+GlPh-S`]fOҨ`k9p:ML|299fxFW^w7NFTxS-m'>\lDd < 1 C A102-X .1͵R*X  6`!X .1͵R*X: xcdd``ed``baV fx,F^equ02Ȓcjl@?H @fM\P`溽@pT9 P\H+21,Pcln!ZplZ1v I@2B@7FpS3xlb|t'nzXMA9 3y#jKC){"Mg3/ xTMo1MEJzDmG%ↄzpF]otcN*TBBB 8_7(c& xf{3=v\ X- ʷqz9?~gXۤx7Y- O]a܆9ŘYq>zY:AE˽eF;q ):&K6߉\Q5+TRE\(;o65SY]rfFOCtD.EBI&2Q (x<nu.@[_ԩ'Ti.{fz*D*dP +,sـOo R ̅'<b8D!{˂a1=F"*FgAf\dp#k?Un$(F{ 8ɿ̻vnoIǰ**.mVEX%bsB^^˩USNAxBUܹ3ԒML>mыs.vyq)ꖯ:prunǻxglMDd h< 2 C A212]y4{viV6`!]y4{viV|'xڍJAK6˝`hN'X  >oacF>Ep[ ݅g#@2' ;r¶U *vb"RXkDT|TW6C+K݋ՙ4E^fTv9}c4d>-&|ߴW+zWf4S A{UWkk/,>W"D y窰PsNl Cyb##Ro|Ñ[X27ܪ ˆyͭ |]1Zѳpg.~)p2Dd < 3 C A322j n1;_y~6`!vj n1;_y8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?)~x V8p( TY(t06C7 M-f86͊6v @FT3 4GC#; l\~b 0TxTn@]*TўЈS*I(WrȉpF:u*!q |Gn7ٵ8ޙyfI}D!;dh YʎTH6`!w}6R9>T8 0!xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?ȱ~x V8p( TY(t36C7 M-f86͔ i`TFJ ݌`#A*g33Cϸ+ 2&`G@b 0!VxTn@ҒEEЈS*5i+A9D8#^'ήqJH 8r ̮đ7|3^Ϻ2kwaUX\$/oDѾBj/8.^9ĩ?9cit+3_jFR=qY% zcFu*9wq-k7zHʲu*Iuj.~u3Q!?H_1133687872FPH@PH@Ole ObjInfoEquation Native +_1133687876FPH@PH@Ole ObjInfoEquation Native }a I-"I () 2 =I b i-1 b i -1 -"I  b i-1 b i -1 () 2i=1N  =I b i-1 b i -1 -c J i i () 2i=1N  =I h -c J i ,h-b i-1 i () 2h=b i-1 b i -1  i=1N  c J i i =c J i ,0i ,K,c J i ,d i -1i ()_1133687879FPH@PH@Ole ObjInfoEquation Native _1133687882FPH@AOle ObjInfoEquation Native 7 i thB I b i-1 b i -1_1133687885FAAOle ObjInfoEquation Native ^_1133687888FAAOle ObjInfoEquation Native  J i ='argmin 1lL MSEI b i-1 b i -1 ,c li ()  N_1133687709FAAOle ObjInfoEquation Native (_1133687712FAAOle ObjInfoEquation Native (E=<  L  D I(x,y)_1133687718FAAOle ObjInfoEquation Native (_1133687715FAAOle ObjInfoEquation Native 7_1133687721FAAOle ObjInfoEquation Native (  N  N_1133687724FAAOle ObjInfoEquation Native (_1133687727  FAAOle ObjInfo  Equation Native (  N  N  N_1133687730FAAOle ObjInfo Equation Native (_1133687733FAAOle ObjInfoEquation Native (_1133687891FAAOle ObjInfoEquation Native : 0ij<D$ Dist(i,j)  L_1133687894FAAOle ObjInfoEquation Native @_1133687897 FAUCOle ObjInfoEquation Native (_1133687900"FUCUCOle ObjInfo!#Equation Native (  i  j_1133687904p&FUCUCOle ObjInfo%'Equation Native (_1133687907*FUCUCOle ObjInfo)+Equation Native @ #$%)*+5EIMTX_cjkuvwx$ Dist(i,j)  NQ b 0 =0b 1 Kb N =D_1133687910(0.FUCUCOle ObjInfo-/Equation Native (_11336879132FUCUCOle ObjInfo13Equation Native m<- 豃ۻUQYnw)^x&2qH,i/tҒ4GD7"PX`K;X*xGU-BpgRYq2>1r`Tr1hNٻX%u|lBQTIRfz}j[ɩTgS)(dǸ0|f9曺s%enZqnep܌"*jƇS(3]&!q9LR$R\og,X;]l]~2.koR}p@@&hrʅP Ds?awD[ ?Ӎһr?_NmDd < 6 C A652|Ux(6`!|Ux(X[xcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?ȱ~x V8p( TY(16bd3ǵi` А@ @7 r#TF hq*H:`eECR5;\O9X[{xTQkAޤҭ }J5mr-7E}7A.{ٻcwsMDPA>M>hݽ$WdgoffXY&fb\;/o\ᾆkj?ɭsgq,OԻ| 3E4Y Wd=2GV E{pW@ȚZX9P).L.a@ ʕnqYVH뿐61k~Lsua0C, x?RA6A4=} 0 ث@(8”_&4*: :N[J\j{[,UtC<_e.ƲDނ(yśF/G}NE՝&BlCE `|4QC@ٵs#hL]mrU)-SqQxd:$GB rRL BCK'eolO#~ZK=֊b-HWExA`{, oT͍0o8!{ʡ 0.corsTfx6_t:Xm{gUC29XMly cs3[M}A>pv/bmO~ %Y Dd << 7 C A762~ ۤH] h@Z 6`!R ۤH] h@C6` xڕKlLqiA%cSN"Q,񦞭ҢV[Qz?TSD")"B󟎑љwL̙_9|#T!?)z=LAƞ%c>9s8PHl =.?{ls+6Kpީ6N[{EoޖA1{qX#'wlNQ2UTI{ZCJL9Z ҿ2.HYDiվc=X͐BPօՑެhՏ*#?=FѭJ7PriG6Gm{Fm ߶92:KR=u^7d qIK wQ:Fr;nwwM|qor;8.Uf7zVxKU/zڽjb~so0|Vo6e&!.tb&nb q@]qOq8M7qqnJ^.v:nm_ H+`yGH;oyA|+4Zu͕GT'\n۝\\ii)5WԵMju ek=e{&گ)Wѧm~ZS&J C_Oڛ\ |@\C6` xXoU?3[2(JK#(%ZKβ#ζ@h$#WMLDg>ܙsl lw~t]o+`h'< 1w ~8!$L ]>!GxwG7ݻ6|;DY>_*p4s 9+w't;?.'l֘碸t2l4X":3 ۖË.ǓVCG85!smZ aS31ZNIQapJa ]OmOrnGun}<݌_v/Vj!q,|lY2E9nDck0YkQpiYN' Fak<]Jbr24;k''jd]+,^hTҔͷ:6tRO΂,@So"|H!Y<.wR-b$ QjI.gK )j6os'6Z} Cgn%]t [^Wڶ|lpM-2I2}Mps*cnZo$ 6ՄR|+HXZ}.>)Fyf덜P"[/DL[1m.%M4}B&!gJ3\"@&d}[/e |l#|b.iG˳\Qgo+M=u1MICiΒX': lȄx%#~Z5 0i=g]?(rq :Rd@'F4=Uie^M|md˷lyڰEdp:y% ]\ A'i/YϲK5=+(N_7QHo@[}k4(j[ٶC)ѷ LQ/__ުPޣ7L4ۡę0WW+3 ěc7 pD=!? 0_@xzyrWr3Gg'1+]]yҫN_͢ج8uӳ2\^} <tQ)Dd < 8 C A872v*?܎Hu+6`!mv*?܎H fxڕKAߌZ(MAQB@t:/֭k+̲_H]Bh{nKjofa|Y˾C7fQÚYY>7:3Gc㪪p⳿Ń*&&Rzv r2 nuT+J !|>E nficQL#s5o?Ts!Z:5UdiXSe*epY? Vny+x1V"kʚT4 滅T1N UV/Z v YOuA_館:ddaBTG8"`3{:"3Y TiTVdXT˸wNQ{!8ս2MPzIe]Ai$sΙU+RYC:O35Ժ^ɜ1]t f?#ruLU-ZՑCrO7,vPh:\f1xUݏDs]+ l4J.p (N'U^P%Ա#Is:*!*">wxA ψcvvQVdg7Y?6CW&r"OI5M#G۟r g0v2 &ج"akѐ~0S@x 2݄{|D)7'$B4WS Eķ8z>ŝ$sD ӆu2 ~|נATB\hH٬AkU&ZVT[5ˡnmEF\2*H4}X+7W*8D _q1sD sQ̮7ȏO5i{ϣY [LN bW4j̉"Tn%q~n^܏iѩ;˨|hEd'5[?aꥰWIeXTvȺn VJ7 'ς! `l`Rsq|P3J9/ z]qdA(59ѳw(/]~%tK"@4+G.hE^aP9"d6ln7={(:޴ }ƙP#3+VYkiW]Rf[e,=.DR8F{z[OT6*Uc,c ,HRԻF:i{v=%= =~cX=Zzv*Z 3=ֺNǂZG 9S%O+w8D^*.;oϯ6+HE&.B|+&ؔzt9pK-=֡!}zԑr @c02Y,P1=F (`d?3%$ lʛ! 27)?)j~x V8p( TY(26C7 M-f86:?Wv @2Bٍ؆ˍSn-` jh@ TguGFCݗ-8rl'@Ө RxxTn@'i *Uj%$4B*NM+r9:^jZލ*TDߡq L<evTB0oϸ&XO%S._q;xk-e9Oߓv_!y6/92hº>Mzji;?6.`>V7:\hق[Se Z 618eEXh9TP2?lō2MQ (؀,3Sh7e0f)I@4 |v RḲŲȈn(ogSsZBF 5Q.9oVjI޺L;9w sLEF5"Tb`AyzOjO]PS7YQ[: )s^P"9Xb cX*lxgeyl嚃bORT3c0NnGU *@`f, Q5>1Ci?apW!{ʡ h9]%8ߝJqe5ܱ۳J)16ߤFDbv2 d=޾0z QU-y|w@_ ZҲě:38K:L9\'\MDd h< : C A:92+)w[VCz6`!+)w[VCz|(xcdd`` @c0 1Y,P1=F (`d?3%$ lʛ27 27)?q:@pT9 P\H+21,Pմ=LA $378ش\"Mc22x200BF&,x !2MMbg`@L`Ȣ#<V]zŽR+ gE!YlH7$Cbv`6J=HfـPLօuKU jȯZ (ׄ-zE Qar'1%ǥiwbbӾ.m4I$"F6ah7:&{Nr e8vZsۨG hr#&z? (:HuuFv}#{B!s8M1E{X>,v,5QH4\:Lu#Hν4EY(x4[4~DG1'0+(N ԨzS>#'ː y%Pe/Yu%LUD$C¢&IqL$4I^B3 l1 ۉ?IN%iTU* Ӯ]V%74UزK-%;u j]puՉ+ n\U~I~k`usU}jsU}r] |{-p]׉jp5KkTx z hd{(En8o Cܯrsxn|PRxU[kF>3Zۉm < ^gu۵\CK~7C겫F+-vcc.-oRgF/m;Ι3ѭ߀p*\rH>~=%8_g_ӯG||? - N8Ȩ"]OK, :ށ쌓ċQD"&ӂ[p0z)f Є;@.f\E"5h'aD'_bS!k;~>۩3E~Ȇ6u4y\^Ke9kW_" ?n~#Q㲫\o WY)b3mљSCդ1U34Dd < < C A<;2@O$cK.Qg,6`!x@O$cK.Q8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ic~x V8p( TY(t36C7 M-f86m2?W~'Jqa ì;a2;7јk2@(z6k/>b78H\6P8F=CiSjfʧȹ:Qбg8>ncY] /jLtd|5Z. Ԕ77[riCG%2)Dǩ˔Ǣ#Q``ЌtizAڨH >dY{J&U,/yR'⾑j y.Pe3ko_9+{|s_`y1.$)MC;VLcuf YݱGMM![J%sabW:V/z#yyݭ>7^6XO6;2NF {ݓ)]y~V7J=aAO6Dd < = C A=<2aS806`!zaS88 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @(?׼Gb X@@ZaB7c?tӘ b`zb7Lt+FT3  ^4UCs ]^(\wǭ` h!WxTn@B"@qJ4?EVBPD8cSgݸB%$NxÑ ̮đ|x3+r p؁ ,"|n&+^8}22xM|*Mػ|6` CwƞZ+v2~>~'Jqa ì;a2Cјk 2@ʥ:K9GClc| zlQcԓ >$ZMbY?rn{1$4: Gml56+$R VPN +i|~Vh.=CO;O4Dd < > C A<=2@O$cK.Q46`!x@O$cK.Q8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ic~x V8p( TY(t36C7 M-f86m2?W~'Jqa ì;a2;7јk2@(z6k/>b78H\6P8F=CiSjfʧȹ:Qбg8>ncY] /jLtd|5Z. Ԕ77[riCG%2)Dǩ˔Ǣ#Q``ЌtizAڨH >dY{J&U,/yR'⾑j y.Pe3ko_9+{|s_`y1.$)MC;VLcuf YݱGMM![J%sabW:V/z#yyݭ>7^6XO6;2NF {ݓ)]y~V7J=aAO4Dd < ? C A<>2@O$cK.Q96`!x@O$cK.Q8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ic~x V8p( TY(t36C7 M-f86m2?W~'Jqa ì;a2;7јk2@(z6k/>b78H\6P8F=CiSjfʧȹ:Qбg8>ncY] /jLtd|5Z. Ԕ77[riCG%2)Dǩ˔Ǣ#Q``ЌtizAڨH >dY{J&U,/yR'⾑j y.Pe3ko_9+{|s_`y1.$)MC;VLcuf YݱGMM![J%sabW:V/z#yyݭ>7^6XO6;2NF {ݓ)]y~V7J=aAO5Dd < @ C A>?29ec̋9=6`!y9ec̋8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Pfln!Zpl)N:1pS\neBs7#HJu< x4LC*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<Dd < A C A?@2Iy"R+UI[؏%nA6`!y"R+UI[؏ 3xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\P䴱qm?z+8U`fҊL  B3d3Յ4F0n*#'Э܈a*Ѝ1nTV SXVM"֭ ĺUhn!֭@SEuk"n* T3 ܋C=+㍅ 3xTj@VKS -rOӺ^*݋zWl2M .kA gO^i|I6B{~3'3I UrBf:g$EkUdsJ{4}=D| wAV Url, <&dl y.zKcԲOޚX~Y^ ?LSd#&-LJnN4A}F[wT wah8KF5UIs/;Q? C+ 6l5 q bF{;Fb\Uұn{1(Sؓv]nm /,Jj415onp]r uȉE{Y9G=䵂a1e=ه&D,FAj\ϓI" %N#/RaWzƙ|ԗ?μ-jzm=qaòyˈ|1 9h4 ]uiGk(ҋ#_sEāzB4O-6Fw*j/,滶yjgcjMS? \s5Dd < B C A>A29ec̋GF6`!y9ec̋8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Pfln!Zpl)N:1pS\neBs7#HJu< x4LC*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<4Dd < C C A@B2/.;CJFmJ|J6`!x/.;CJFmJ8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?d?ׄGb X@@ZaB7c?tӘ b`spn#7V&4w3Tg4C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp4Dd < D C AAC2qsgٴyPN6`!xqsgٴyP8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Pfln!Zpl7<1pSneBs7#HJu<4zC?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp4Dd < E C ABD2x$C !]OR6`!xx$C !]O8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?d?הGb X@@ZaB7c?tӘ b`>ri`7Lt+FT3 4C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp4Dd < F C ACE26+W7%I -TDW6`!x6+W7%I -TD8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?Z@pT9 P\H+21,Pfln!ZplZ7v ~@2l$H:`3C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp3Dd < G C ADF2&(4JS`L[6`!w&(4JS`8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?d?׌Gb X@@ZaB7c?tӘ b`X?Med ݌`#A*g3@C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOpDd < H C AEG2Yy$ Yg95_6`!-y$ Yg9<?6 xcdd`` @c01؃Y,P1=F (`d?3%$ lʛ27 27)?q:@pT9 P\H+21,Pcl4M-f86k'v@.Fbx!2M] 4 T&nE[3f UX&֭Y@SUubj4UX!֭.@SM.$aT3 lpCu^|_R?6 xTn@^HʩqJPK 9T*(^J!!^'N]뤩B%JH(p=8<evq`wff<ޝ#"6#w͛urB:/Z6'w>x,i.}Q$1Za秃Y BdQ%0rޘ@ȊqM,o$1YjIwhmrl,eèUX4*5e}Fη[? ?*m(Op^  \P$<@ ;h-3eWkM X:z,<7i2)u{[10Ѷ'2_ p&@cH1 Qw0J9z|DQ#VtG!+=y-CDY7 !x"MBU }x|-'wIYY 2=,<9$nj0`|ZL WJYf77D d,yQ*Y Ϊ@G*$B+y.5c*)AT5:Dr" WE&ٿ=~uveԬR}nj!RC([lt0fttcf| L76[p DH D Ą2Ǽh/yp@Ly_ tP0RI'7,vZKS2VMNSEQꀥ4P"=HN8ŢBAYv&ܷo̾7L"/G NYĦ!Q&9Ue )=d#f Ք :̢,Ds/9d;po'4CT{W/mt:̃vg=ךKR|n(]'xic9xr|2Mv{/У/x66-{~`ϙ=\/*|h_x\cr6Dd < J C AGI2d5^%- %!()Ei6`!zd5^%- %!()8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @וGb X@@ZaB7c?tӘ b`sn#7V&4w3TgH4CJg}B>h@D` h!WxTn@BȢ"UjOh)Z&)ʵ"%'{:M*!qB;' x<av$`MoW^;P{E $du(KtjW q,Odek)R]U8۰wl<1#PqEwƞzkv2~>~'Jqa ì;a2C7јk 2@ʥ:kK9> 0£T2QO(q|v>:TO`Tk.mه}c &Pm;_{Ls'kc,"C>=AJS4!J>":a]Y4f)Nhu(MZG=lT2ΔyaKT]*V WyDcXncIVorNʝʪ;Be4Ôˑclqj!G΃mTv^YoʦuL$eJ'B*}^Ju&QE1o_y+3|sw@~1wa ]BeQF%ōLRD*P n&!5+S(p~'E^rH1ΌYa=uxUW(P^ }P ̟(ŅId$ df9G޻׏\S=TRf{< opؒ3([lt0fttcf| L76[p DH D Ą2Ǽh/yp@Ly_ tP0RI'7,vZKS2VMNSEQꀥ4P"=HN8ŢBAYv&ܷo̾7L"/G NYĦ!Q&9Ue )=d#f Ք :̢,Ds/9d;po'4CT{W/mt:̃vg=ךKR|n(]'xic9xr|2Mv{/У/x66-{~`ϙ=\/*|h_x\cr4Dd < N C AKM2>%?O}5j}z6`!x>%?O}5j}8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?qຳ@pT9 P\H+21,Pfln!Zpl9NgMc22ʄnF 3ykm5#C4``LQ` R!VxTn@B"*'4J-i~r NL]kw& 8!' x N<av$`$7+ XJE2~&R.n8gC22xC|YU8۰wl<1##^Wr:eF}O$2Ywdb1p vo1Te*eQ.li^b78Hlc d P!y)5ISbIh'tjmVW"Hy D_AրK.5> DP1DL qj$E2dñ(H¾G,$s14#lb$>]^6*C92h^*b üS)̓HqH5H΅(䲙BCYuS=o/1 y1.$)Mjg:2];ccfC9%sabW:F/z鑼ȼvnW/| ;^FqMGi ?+;aLCQmDd 8 < O C ALN2QS&x#ڧ-~6`!%S&x#ڧv8lxڕJ@g6h7=ŃEk9klRoV<A'CxE >x(\gUJ$'?;3AҘjVp LF D7P(" WhV:\+bhVڠG w{}s64d=؄ 7h4+URn%;l9ړO3*R1^D70H\]ĘqE ag͸>I456sc,pYC. Jd+,:+[nQzmsY ls{$5(fo0- ^^;\]r&ik~Q/Crz:.pv8SxTݎF>v]p ^UG\%m ^ -jUc7.8&J KwMx33v@EGrr~ojB͓spB U^ yF93>7_UDNAKx߂3ޜx1}kզH _1133687916,<6FUCUCOle ObjInfo57 Equation Native  A% F:R d AZ P={S 1 ,S 2 ,...,S L }_1133687919:FUCUCOle  ObjInfo9; Equation Native v_11336879228@>FUCUCOle ObjInfo=?Equation Native mQ b 0 =0b 1 Kb N =D a i =I b i-1  (x,y)-"I  _1133687925BFUCUCOle ObjInfoACEquation Native b i-1  (x,y)() x,y   2 b i =I b i -1 (x,y)-"I  b i -1 (x,y)() x,y   2_11336879294TFFUCUCOle ObjInfoEGEquation Native _1133687932JFUCUCOle  ObjInfoIK!Equation Native " d i =I b i-1 -1 (x,y)-"I  b i-1 -1 (x,y)() x,y   2_1133687935HPNFUCUCOle &ObjInfoMO'Equation Native ( e i =I b i  (x,y)-"I  b i  (x,y)() x,y   2 d i_1133687938RFUCUCOle ,ObjInfoQS-Equation Native .4_1133687941L\VFUCUCOle /ObjInfoUW0Equation Native 14 e i= b i-1 ,...,b i_1133687944ZFUCUCOle 2ObjInfoY[3Equation Native 4Y_1133687947X`^FUCUCOle 6ObjInfo]_7Equation Native 8: b i-1 i-1  i_1133687950bFUCUCOle 9ObjInfoac:Equation Native ;._1133687953DfFUCUCOle <ObjInfoeg=Equation Native >(_1133688580jFUC0DOle ?ObjInfoik@Equation Native A4 M i1 b i-1 +a i_1133688583pnF0D0DOle BObjInfomoCEquation Native DM_1133688587rF0D0DOle FObjInfoqsGEquation Native HS7 b i-1 +e i-1+ d i +a i_1133688590vF0D0DOle JObjInfouwKEquation Native LG_1133688593zF0D0DOle NObjInfoy{OEquation Native P4 M i+ d i +a i> b i-1 =b_1133688596x~F0D0DOle QObjInfo}REquation Native SG_1133688599F0D0DOle UObjInfoVEquation Native WZ i-1 -1 M i7 b i-1 +e i-1_1133688602|F0D0DOle YObjInfoZEquation Native [4_1133688605F0D0DOle \ObjInfo]Equation Native ^S_11336886089F0D0DOle `ObjInfoaEquation Native bZ> b i-1 =b i-1 +1 J(x,y)_1133688110F0D0DOle dObjInfoeEquation Native f7      !"#$%&'()*+,-./0123456789:;a>@?ABCFDGIHJLKMPNOQRSUTVXWY\Z[]_^bcdefghijklmnopqrstuvwxyz{|}~tĄ2|ސw`,c躃 ߋ*#M.b/gw]V'A<Ա0ɦD7櫙X@̧k*"S%QʖI_c^O}x`b~>.F<`EE|_9IzP-wJg9& /VaXv|'ʘ'ف&yyR /W]vJ1FFpSr2VOb4jOX01K@zu@@>&֭@SuR$@*gXTCVb- 喤.@6 v xxTMo@;iSJ TTNhB iOTBz ᆄ{:k뤩B%D%$$zZ@9r@&~ev|T`$';o޼ww)H3e.`buI+p(iS^)D.pqx>O "NjKLP5%kM,h;P3do/U]UEx_fhr& Տ<6;+b U0Šu FIAC*9.9= U$aB˚F9n }p7՛[3±|:duKv}P&fBQE$h_?2gχυt wax7a}ĵ]EHٺ:G'2aMC ixhFVJ͐'9I"<xDd @< Q C ANP2;Se(؉6`!;Se(V Oxcdd``eb``baV fx,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(L`ln!Zpl(1pSx!Acqb*#T[œ8Ag imaTb/zXUg1~15a."”MUDeTVjQDR:dcLLJ\ImEe^hȨ":JA##U'%=WᏅF6[եBX¡5CԱx($C܁uRv,tEQO2p,vM[3E.ݥ9yXmmqHvkS]9&T7uS{oi]_7̋}Վy>3/ZwLembym`;՟FbտedcDd 8 < R C AOQ2Qs!~Ż-}6`!%s!~Żv8lxڕJ@g6h Az*hśAڴz G(zE >Dp7R&;ؙ STD\yICH帘J<(ŜdA-4fme>գ1`x{j\˄]}~] mPuǼWN!j~yA6˳v#ݍϵ>ϠŌ]S/r(ؕW:N֯~!WXOВ'e-IsMcmYmHY$[` &[klPʺlڒؒTMZkYZ!aoYzֈVcieSiIYw kr5;WYk+Zd.pCPBgX*bD.Y v8SxTݎF>v *7-T+@c7.8&-JUK/T];&}33vjGrr~o;Մ.gTPӌ5z.Ș/\=Rr7997+\nw~33,bTK %M[putĄ2~{| &ogYx&-0zI-؄+Fє k&@X"ޛ Vݤ6mTPd‚1,5 g%7l'- 춱'۵ǝHxC]q#r,$Q!Ewvj tA Dl&3.} MXdS@h x!wJ Si犃HTmcI&i-;-򄲥iR0qki#^+3~Oؼr @g?T |z~h]YNɲ‹UMxeϓHqJuc'];}ӯmI2$i]2`)'U r?~=GuFRcc0Oƍ7̟l ]W/+Ωߔ}^FDd X< S C APR26fE=j !@v^6`! fE=j !@vZ xڕKAƟkA#- hզ~GM5H/ͫ?z*N3ɼ7$~݈f·S-ؠ4VTM+ RHwM0:.̬FډRqqs#YxɥH>i NgF{5~uμ䤝o8FÛ.8r\xIUejw/g똸kuj?qb]Cz^ukܠV4Cʺm:úfe#Ǭsu@OXZS5Ǻ uͪwkV9窱7Eus=̳9%ll@́W.9@%x[x%vYMc]ߺ'C]+>#]qv|fY|wuZe~db]XגP}CS[/F'qg]_^B3{U&B%[CGwu{'cjpl9 xV͏D$zۭ**l3{@Bm%p1MvDK%VEqY_N87vl--Xr2oMrAv, 1]l80ކ{ }Or#X> |S5x(,2 rkx[D=RI 2/SNȺ'3H40DAJ0م:#D>Jmylc/7!:yJ""P hdR>5:QF 0$ԨV2C匡A$C6.Yo*Y5`0`qRS繊kUU*/fdъe@ ^; !oYAngEdF:sup3r7pR>pMhē逇1xJdv6xV w#o10٩&7z-НD ?|~n_eG:u_, UMIeva]X{B}f,s5ŹjlMg:ǹjLbr Ϻ hWIgЁ${X!vkY݆zRˬn!Au@\Ӭ} yWA,ZÀVC̷޾ĵUJD CxVKoU>wl'1iS*@3j{EHE6vG<ԞfƮP %(*lX4;T~KX Ă?%;O75{w{̬ z.`[܄K.x$Ob\ogpYދ |s= ;wV>Xw@D]h79}$͟ )*ϽϽG340DAba| +pu+Ɗ`A^LCo+h$N!k5Q_Ju?[2 1;d>gG߂Y'~¸K!%f'Cz.F Fw55EA;( dE [ai8WE)Դz RTXuM\mЮ)L{F0Y0}$z0K[-mnGfTvoVWʠtմtji&'nNkU"JTyƮt<] FU]ŮDt$&TӦ5H]pvU {U m06m8*'g/Mh9dv.YG [C--;2APlfTd;LXͤg.$ѭtP4uM~>j<o/o>:>4n Yu|i Gm -W*[ _Y4vo5jǺfUyuGxLU@\Yaqx٧\7PyZC%o-GpH xVKoUwu3mS*TWf8Fʢx ]$d#jX3cV(B nP#X+,,)yi b$;in %<$"u#qȵ_೉>'yp,F_?>߀VCǫ(R#D]R78?/ FϽý< ~0yOH] D>J|jS7[A6ku-ҽKhdߦ5jt[Ymryg:yBe$׺GA$htU V _]z€eՋ/OxE^_Zk5.wCX :ƽ#[]Ȗ4*K"o1rp76}ZD;h98 /0Ƿ6r`0U #)j`RK{ݱ=I4 0qY`zN80f' )',; ׆T_s%;~@2Y8D$hV4\`IP O+:q2f]qXD~pA1hU"4'0M7Z 'b&yaG*%y1Ruf/b&})LyD\t(*!^;f*BZ.RM7gVى w{ dWxNნF4"ilWpKE:}y(92k3,4DM('pSEɗS#FikVa/efRZɬvsX@u\4/lL^7Tݬ ,*QK7uu5 i@{6K6+x:}yUMGBo*V(]$hBM- Z0́Vhc}h]CM # a2۴v̺lar!(Y*5ֻ%MUDX,ϒUew{x+:COCoYUCWk$FЫ p=8.ܥSpqEP;܇oF _9\™ϖpW{Z% #WUK8p#Sw+e$<7ffn% OIv[Txo\ߒgRb]Dd X< V C ASU2k>#-#S?6`!k>#-#S? )xڕ+CaƟLې֔2(QvEqCmm.$7\ ҍÍ߿{liG|gO{(4.U htD.:F5YgZiUC ÀKVAx0|JAZC`;3OR d'esZI_d:C\]'swqH~UҴ 繶X*ᇲ?S[NӸ(9&s2u-pS\\oXqq&;U5*0޲1I bQ \S|u KlkkNc=: uu uu}`]eF:ͺ>r!/\ӜF4<ۭ''Zdqҭ%6g'݊n: [)uI2뫓nm[ԛjʲa7<>k> u ˢWK9Ϻ^mT^ԼO GC>ݟxMhD֣h@ )ixV͏D;]vnUE񡧊n6vJ !h{@RibG&Z*] \{DN?#)oƎt$'͛%ّmg8r\H>~">|*^'Mv~.IbvY7QqYkHobМI{G2o5r)!C=ۖ&͉l侦(|뒖g-"4Td' bUk+^gR>fkXNW6Eb(ye[un7eoUN$[ã?c ~|z Il#PhģD=Csguͽp^V~v+[i'?S8}An9_mVG|"uQ\]Y N->.P G\>ۓ&_n7kg*?w"GnOVOoMAWy7Dd < W C ATV2 `;bԜa6`! `;bԜގ !xcdd``> @c01pY,P1=F (`d?3%$ lʛ! 27)? ս@pT9 P\H+21,Pclnٹ?MedHƟȦ@Hfnj1q{pb*#L^ 54FT3 aFC{3}І6' !xTn@' TԞЀ8$6\CO3rd7U$p<OPfO%~Vdgg]v]uج\l-u)'p$\a}O9>w${?_ux6W`{騟UF<{Ps79;kR߁,ʤTY7bj:z'ͫu'GB)½'-SyA.pag @KBfIQy>)5Np: bqdq`>`_ج)&/Jэ/r8RsR*o.S2P>Di.EGI(Z7JrA"p,R8 qP3Db23QL}pah֙ұn,bVJEQ8QDh 9VɉB̊Y*P0X­FM edEtEijf'Q骰 uTԤ2D-1p){Ρh%[qZ;=LNf4)-{oPHfq1Y7pvI>E5}jXX]6mM%,c&4uX0w!}Mv*cS]]*oy^卄M $ <nkDd < X C AUW2UK^鍵ZT6`!UK^鍵ZT hxcdd``> @c01pY,P1=F (`d?3%do/PF I9 @Gb X@@ZaBc?tӘXi`TFT[nl*d3!2Mde`C F 3y FCHKȹI hxTn@')-TԞШ4NrEBP=ȱRgnT\g(O< @];M*3&;3|xQX 2۪ "X2M3h٘]ı<ޥσ{c) غt

F"48C!awۘ$A=C,OÑ 0HUBR'V&TiXecȨ[ڊ>+yHk?GBǒ(URvEٲ9t"ӉK0 ^MRL(dħs7"&r@v\cͺtbh<˃Rq\*=V<N cN$J5ofB\c gR90΋ b_6i¸رJׅElq@MMC1s6r_rH:JᒯG8+ߝ9hJE/*uU\n}MܥCz|>5A"B۬!>Bt0qMm{H_}lUWoypӫAW ?nf,Dd 4< Y C AVX2W;PGYiexi6`!pW;PGYieL)pxcdd``^ @c01hY,P1=F (`d?3%$ lʛ27 27)?ȩg*~x V8p( TY(16C7 M-f86mWTf]Q\ TU~;b> F?2^Dcqvyfj[6qvKX6-<7bYr6Z)Xi t@tC|e87P#Q\o~(2Ƃ(Rt0b= GB^ {.0^GCt}E@rvыm8P2h7Y:r#(E1)8=Ix|̑1%AyuFk/a|r}a ː # ?%hS!kyA2Z"3 IC_IJz֚9?aXds\N=a ܇-zb̪:g^`ޱ^Ҕ Hhf 5ؼҍ\QMԇ ܻZv"[]+9u#VqfOî2'!DGblccT '\R3tu+ߋ2N=bOswݚ۬ '/F"h85)5on Rw JdfZRo(3N<p/ F \[t>3({ ^vN>]l3쑫dgi21!+dsTMWUY3wU3~:.r ƕcEc:ٺg eT+԰a&kqUA@ObM/y4n B(!K%@d"A- ۻ:N12|.c_Qu^n3a*$L-YtcZҖ-g&4 20Q*Ғ7ɕ@oQ_"r;f|uq 1C560L^} DzZܕvfE(*2=\=` N-1'S*f P{HM X"ϋT}ǚAdwnSF6TVA6;X^TI~Ϙ=F6Ds9dT8^Lo'GZ;]h)iq|v>:TO`Tk.mه}c &Pm;_{Ls'kc,"C>=AJS4!J>":a]Y4f)Nhu(MZG=lT2ΔyaKT]*V WyDcXncIVorNʝʪ;Be4Ôˑclqj!G΃mTv^YoʦuL$eJ'B*}^Ju&QE1o_y+3|sw@~1wa ]BeQF%ōLRD*P n&!5+S(p~'E^rH1ogꬰW SaaMG>1A)7FsCMO?ЖouFP{/M WNpDd @<  C A\2'͘(o6`!'͘(o xcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?ȹ@pT9 P\H+21,Pcln!ZplW2v @2A@7r#TF @`#A*gF'C.)_pł< }xTn@'ҟ-T=␊4[劄r"HܐcnT'3' fNJFdow]cە`n[6ga$ȯ0 :ӗ*ϲK EpEX/.$.0y5tr̪;~\#oQ,e:M߉عC*ut <\:(U_5:]9ڟGa(=F#la$Qz(y%.%* MARžxa]Ƭ0MNs,J!ɍd\$\cPiošPJuBba2/ӱ,c!K']R(d'CT:oIf I/ 6ti7y6cVD'"N\L&e24Pξ("_8,JѦ3c v "OCMC76s&Nh/4TT$Nwy69|X5Ytmb2P*V:vT@(v3{'7Aq{UHTW֜Nئ+sF0;W#Dd 4<  C A]26+2Y9x@Aop6`!g6+2Y9x@A<)pxcdd`` @c01hY,P1=F (`d?3%$ lʛ27 27)?Iz@pT9 P\H+21,Pclnӂ8b7LȐt+#ԍ?Mb07FpS3x80LeiĺUh*76Sn5GT,:XjU#nM*KTl:XbK -BRf0B"CneQݲ)pxTn@;iӖTTNh)VHS;IC@8#^':iP  xH\Pfvd{g,}9 :Ӏb*Oկ8yu~sQ,ws%Ygnk3; [ώ((ij e%z]aS>z!@EV^;J8 xjHwdCb:@t5G+ L&)VV)iҗu%p5JZM1+sUEU]|l)X?&-L0D1!g}մlcY;Pn:!Ș&2maӞF#^ f>?JD7^Ʒ0 c*u\J) DJ;Qӈy}hȞCv5Q!t6ct<E@i_\G7rњQJɒJ#>Qr#b  z+;r~/ːç ?+JТCplf4e*$"3$ *j2DR--Yq?p:J-8N ÂV4dq۶ϕ*촰u m3ivtHè:&v:´m[:aoDjd)dZV]* twgh¨L%:OUe=2yH}eyuDk 5ReMOzk\7~ZZkiM?qfrVHDd <  C A^2mҲ7'4QXCFM6`!mҲ7'4QXCFM.X'xڍAJ@7ikFpR*"-AF-.\w@K;60€eڮ"}H9Nel9KEÄV#;02+˜D̕MxiԞi/#pJ}dN\R2TpӲ54 YFW/B!jв~|=Ө5mJɱ ]'M4wfl¨n԰E}+!FPBjswɎ ,;{onwNv2ʒJW5|6a ]L|Jw0 ZTDd <  C A_2Qg߅CĿ!/-6`!%g߅CĿ!/!dxcdd```d``baV fx,F^equ02Ȓcjl@?H @fM\P$sr}~x V8p( TY(16C7 iX2?MedHƟȦ@Hfnj1qr#TF @61>5Xj@LX&M%*n&֭@O`#A*gD}RCM_9cA!x^!dxTn@'/uRrBSNBI􈄠z"\AN:k˻ISJHTHH3' xN@]ۉ#Xig^ Pcap4 ۺOqӺDsM_ps\gʓ]= =0`uV`kEtvMheׅ+X:Rtq̸BKS& S*0\\ͅ?`7q EoՃBFU3+ˆ 7V:(w=b\ak]ߑe=5M50Ű(3Kݽ}⎰jluZ8velsؒ~ ,61wP$ya< !wҟ.h9Duuo"T( \u0!UQ<+<_|+;60€죅>w8 mE!c4Hͽ}tNw'Y&ciQYcYjg5i[XOn,VFj,ܷв?Si^,laZA&Ps(;8~9;ZIWEqi(\2 o|؀jΧ)pUĄlDd @<  C A`2- @c01Y,P1=F (`d?3%$ lʛ! 27)?`~x V8p( TY(16C7 M-f86Ki`TF_[nT pb*#L^ D F 3ymPDCdoaTkR yxTn@'ҟ-T=␊4[劄r"Hܐcgmy7nP Rp<@];+!iofuml׮7 e3}6Mӌ 9`?}i,PtpX[t8Uؽ|~,ڬ .1z(k}h*t{qE;ÙJ_*}$+kNC~ׁurlo'.V Dd <  C Aa2Qd^ Ո'Wa vSY-(6`!%d^ Ո'Wa vSY!dxcdd```d``baV fx,F^equ02Ȓcjl@?H @fM\P$sr~x V8p( TY(16C7 i9?MedHƟȦ@Hfnj1qr#TF @61>UXj@LX&M%*nU&֭@O`#A*gK{CM_9cA!x^!dxTn@'/uRrBSNBI􈄠z"\AN:k˻ISJHTHH3' xN@]ۉ#Xig^ Pcap4 ۺOqӺDsM_ps\gʓ]= =0`uV`kEtvMheׅ+X:Rtq̸BKS& S*0\\ͅ?`7q EoՃBFU3+ˆ 7V:(w=b\ak]ߑe=5M50Ű(3Kݽ}⎰jluZ8velsؒ~ ,61wP$ya< !wҟ.h9Duuo"T( \u0!UQ<+<_|+;60€죅>w8 mE!c4Hͽ}tNw'Y&ciQYcYjg5i[XOn,VFj,ܷв?Si^,laZA&Ps(;8~9;ZIWEqi(\2 o|؀jΧ)pUĄZDd <  C Ab2is5Ʋ<,U2 6`!is5Ʋ<,U2< <xڍ?K@ƟZr-8TZtq$'(8,'?4."x> rwDѱ [}VKfìIR dHQ 19D6ay,pt~1k._棘=z?{r:./qŶ:Og[ >w78n Zr]h'WgM4*U tSRCONkk8D v&:պB]{ZWֆ*-h]}Z۪躣uiUiVi] D ]-*wvŹ(w_8Cyq sGSQMC< xn@IAJFbAHUZ$$ė91u֖I[$$$ xq>Wr3̮; x,N -)`aѼAGڣ)ic Y \ W?<k`*4tsZOxy=,KW0a I$s_HS \EXۍBQUhԇa\}[5X Þf-`Ͱ_RHG9Bhwv7A{Y&vec1a#avkX̓*r.vm3&ـpCi4<ԑ)5on R9ec Jt*: EH= r:Gr*{5a{yd Ǧ,{Z>IzTM ÜvSd#/6RaXVf ?O/273PQzXee/70 yL!.$}ՔMCbfUXd.XQ$bGMLJ}HT"1HavPa)t)[|wNŶ[ws/| mٳh7vI.t2:2ElͪvqhdBgeӶ?fn[AqaVLl[p*0q~{[}5 EU7߸vnޡ;tڋ^i^jlX N bݏnDd @<  C Ac2oPOd"{c6`!oPOd"{ xcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?Ib~x V8p( TY(16C7 M-f86-k?v @2A@72r#TF @`#A*gIEUC -5{_\ {xTn@'ҟ-T=␊4[劄r"Hܐcgmy7nP Rp<@];+!iofuml׮7 e3}6Mӌ 9`?}i,PtpX[t8Uؽ|~,ڬ .1z(k}h*t{qE;ÙJ_*}$+kNC~ׁurlӕ \2 VHDd <  C Ad2|.B( E6`!|.B( E.X)xcdd`` @c01Y,P1=F (`d?3%$ lʛ27 27)?a~x V8p( TY(16C7 iϸvc7LȐt+#ԍ?Mb07r#TF 2`wbݪ 4j@[ K@%֭pe &n5$֭@Se-[VPl)[@VPl[@'֭DgCA*gsMC谔1!yXA.XxTMo@;mSJ]TTNh)QHS;i 3rub- *P~Gnٵ +7z?@5؂ub.h@>T&GQLVeL 7wRyz/;L\zz<@ 6 -XPG^ïW9Um_o`88f66`!~mt>6Ք=@ ?xڍJ@ϙJbS7.TV7 ׮((X\|| ƅ:g2fF993d|)&N EVG1U4~m|e};Xk)BL S,&O6~ {zv;__QvE#y~5zψv6ƸR7غIfb|0&d/Lc7~T>\;~K_KRUjպACԐX jJ Zu4iU*i-҂UZU\*i҂U~SoZ-k'D ]V T`k#7QC;a& %px=@ xTn@'/uR9XmHQzR(gY[^'M*!!U P#7@];ozЂ\ҀhH6>럇qBvi?3Q*qS xAIXŧva tҹ %UY>|U@%{Gyn?IO(._ فe\8 zLPM(ԇl/@՗}%͜(ojJRœVQjEeQڪ gP=<]C/pSlN疉uDaF::tsoעvsˆnWl^1M.B'eذnM^MGGϏ͘p&WR̥Tޟ@]\(hoFEFxl8 LuHW1w.Zp <i"ɹ;nlݞ~Jɒy)+=x HFi&ṛ<73PV݁jye'Hgrti>')A*Y3K ?T]HF$jjLJ`|MYԞh?'p:J ]Nt[oXU?`qꄁmtf 6вlB/JIfOdU[GIV Q MӚ`-wIs\xX]7$9\ / 4uUX{%1k a+o4y?ڧ;]/ťU]52yƄ X-ؖ\j.CDd <  C Af2H\\[^~${6`!\\[^_1133688113FF0D0DOle gObjInfohEquation Native i c ji =c j,b i i ,K,c j,b i+1 -1i () 1iN_1133688116F0D0DOle lObjInfomEquation Native n4_1133688119F0D0DOle oObjInfopEquation Native q4 1jL  E(x,y)=E b 0 b 1 -1 (x,y),E b 1 b 2 -1 (x,y),...,E b N-1 b _1133687959F0DbFOle rObjInfosEquation Native t(N -1 (x,y)() 1iN E b i-1 b i -1 (x,y)=I b _1133687962FbFbFOle yObjInfozEquation Native {4_1133687965FbFbFOle |ObjInfo}Equation Native ~ i-1 b i -1 (x,y)-"I  b i-1 b i -1 (x,y)( J i (x,y)_1133687968FbFbFOle ObjInfoEquation Native D_1133687971FbFbFOle ObjInfoEquation Native nR E b i-1 b i -1 (x,y)( J i (x,y)_1133687974FbFbFOle ObjInfoEquation Native D_1133688149FbFbFOle ObjInfoEquation Native (  N N=16 L=256_1133687977FbFbFOle ObjInfoEquation Native 1_1133687980FbFbFOle ObjInfoEquation Native 4_1133687983FbFbFOle ObjInfoEquation Native   SQNR (dB)=10Dlog 10  s Ii2 s Ei2 + 112 () i=0D-1 _1133687986FbFbFOle ObjInfoEquation Native 8  112 i thB MAE='max _1133687989FbFbFOle ObjInfoEquation Native 7_1133687992FbFbFOle ObjInfoEquation Native ^i MAE i MAE i MAE i ='max x,y I i (x,y)_1133687995FbFbFOle ObjInfoEquation Native :_1133687998FbFbFOle ObjInfoEquation Native -"I  i (x,y)ab i thT Times{X i =0,1,K,{M i {}_1133688001FbFbFOle ObjInfoEquation Native 7_1133688004dFbFbFOle ObjInfoEquation Native p_1133688007FbFpGOle ObjInfoEquation Native  PMAE (%)=1DMAE i Times{M i 100 i=0D-1  D=1_1133688010FpGpGOle ObjInfoEquation Native 1_1133688013FpGpGOle ObjInfoEquation Native 4 MAE=1 D=10( ~"xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\Pq)o?z+8U`fҊL  B3d3V4F0n*#ЭL܈a*Ѝ*1nTv S5XVM!֭ĺUhn"֭@Sňun* T3 Ċ*Cc7}T/-s1"xT_kAVK(+2;HM!>*o%g/%MiDP?C#:wPta{|X-e^ -a(Nh]|os{=Y!}!k ZT( ؼr` 6"PEoj;s+ue ?H9X IOcR:9 ac*P2laZlش SF~kF9qsϗU2?> ~aݐ_uܮb"堃u3wX՝Vrhl[zn.3e UI Otg| p8Q;ЩCQ\d?)e`Nx!2//=x-cx7a Ct}eAj^>GOm5PuD{$sMuMe72(xnYξ*_CFXSL۵2!f;xUKoU>3vb1M#5jW誫, @'X3cQTRbQH,?5"sIڂ;w?`[؄ P G#8p4yM ­,w㯦G'h=o-` k/!{@m%Htu1įՏO'(@4YG3yM~_d i䇩Dav 7VM Pi@YNۄug*ʁyYv)C EEaDh$y-"Apfedgr&Q2"u utʊ #dPNË;ab]cN=K[ XKf#Y:$C;g=u ˌx$Ra_R'iEfUVS箍:Kv$lDhZNiѼI4w^-䕜ٔnxqi9K9ehz% Mg)s+RW ص(>HуQL: ?%lwFH/&_6t:ͤNe BdAlbaSq%J);.9/x슝Yw8zy:Rh4I@IKć6ǜ(~00m>71} о Wf^ju_0ͲQ2uP [E dݮ^u(eWv0sgp/UdU7ܷ7nT,ک& I'+lOǥ@**;*nњbM5->:Vhbx~PO&Z˘xW|2m Q8x%~aֆxަYOe)~kv-\HDd p<  C Ah2%־Ϗg0,zB: 6`!־Ϗg0,zB:, !xcdd```d``baV fxt,F^equ02Ȓcjl@?H @fM\Prm?z+8U`fҊL a1 B2sSMz4F0n*#!D1!2M] 4T&n}B[3 R*N[V?hY 3y6|Ck26=v6)&Ռ, !xTMo@])FEԞqĉ[G$ ܐcήݤB%$$CQo\8rn%dgg7.mI,]+42XDbr0q೦cOuj"Fa.~@| 0:P%+dړ^G,M]R6E F?Ws dpoV}\79*+Wǿ#ƚT2XlXeTto*zQ.8vUFMrΡA[ :,RСGϵM!i N02 ~Po!w͝  (L鼢C.PQ ٘n-q&$l,y?rɷDd p<  C Ai2'?kry"HhUh@6`!?kry"HhUh,xcdd```d``baV fxt,F^equ02Ȓcjl@?H @fM\P,~x V8p( TY(16bd36i`TFCt.7BLeh*LjO[ VqR*E[uT,Ju<_6zCBLiZQybҍg,xTkAMS[kW*ړl?({t蝑&ڴB\%[W :/)yl|0&lrCi<hjN#ȮC!t.;%l߀Y:a$fsG];P֕=h@B:~Y̺><=)ж7MCIOTr& $N 43אA.6H헭&v?mJপCV "3k"2L݅BA*ǦFsD$*Dمqd(J{!(sDhAM_'|3\:j!@̝ {1{XIhb߁ݽu"֔C73]P}|?j/Ynyu_j;@.rZDd D<  C Aj2 K$v6`! K$vt *{xڕK[QϽĚЇu,uppANBƘ<5c[[PD7'G-XCC.]:J| y!x>=Dd> |5F,Z{jgZ)<Y1knzK_7ىI'5='[8rfy&0_7A^*^QGjzЙK)oUIU4JqRAU̘XNUNV`Y5ϩque{(pj;ʚEY{9Q:sCBvPVԿFY. j(Y(#pk=AYu$(5Y#.`Ee Xf]BY]Q,j2..OQƪ/NPH<0-,[|&)%YWQA@#@_$(Ux.K#(=v|  *xVoD{7MI+'4V6)" "!?S]P"H?ތɶEg{7c{6 z^& +E[/ƫpoR7zppKa,&ON} z hE[_,ƣ8D# xz͊a1?楌LċIDDBL~ј.ZCVB{- A[HQb5P[ъ0GoEVE%ȹN< ˕ i RaG[@.CamXovڷ@Q*\jC톞ļgzv+tI7TUeym(פih}ZU#6A `ϭԞys= v'#$[kx$CNG8NS;2P!wMcsܩc^1K( æu~8Ϗ1M#7A΀<Rݷ˒ [$pxŀÈ `^ɝNpE')Vq޼4mi OEWdA7dfRQJQqxN'qE8K1:vf)<]rQwNOq~It 0;+Q)4n#y&SA#rn G)qed&S(wőE@[kgWtg/ݏ>:2لB@fCWy[#֐ ըpxlG#cU. P- *;V`[֞h8pg`fÞ*qE$S̖ެf-swTG8J37Ci)-Hg1dwѪ 0>.MMMHk̇ãqQ$qRZ~i6Jnd|K9 Nj| _Q +I|=ZR$jKSeVОOZ@j6.5/zml|)߫Zi߫bRX}WGT\Y֛R*, RXO,m^gYZBj˺̲Bj2 X5ڇ^aLu}Ԋke5&OXƄ)5ɲ>L,1 k{FV{ubL2߫# Ve5&5՘pembY ,13eY _XO&X Ǫe>4 U@ދRzƯXl5α~ݲ;WƯt6~ezƯoY~{vmem:kƯ,kƯe_7YV8jQUOC|lpINꒄ 0VxVKoTvfjjFVj$$&ľv<΢ F&n*!ǏƌcGI(#A ?0h~Kv_Æʹׯ6,+9}G; hWAP^H~B.z,}Bo=yGwzGo#u ]?YMqPG𼇚,^_/0B[ y)㯠w$9$lQ&Qo/EнQ0Rr5d%)"m0] 5!`jWD1( ACVSE]FwS5ۂDyY|JeKG~e[ ZuxU ITܼQqRڠ]VNſR6ޤ}C7w?Lgã^x wDk#up!Ncy'Ҁh&`T-.[:N;Ď}~Ln9 g ĺnfVqA4g\tb ;6J|C/:& reiڮ, Dzx*% tKv(|aedqIDGq)W"򗟱fSKɇ_Jt~Hf$Q ģ82_2k!H"!Qr8;ޥi-(]k|@MLCDG ֆۺiYMH00pXd`bR0F*Sqݪ(Mmu<>Tkui8ӧ)x@j}`Xmh5}অJ=x uɺ.W 88 @#ekp?U6H }a^)k n] qDabEzh;Bn/p+XNn*E [64QOGGɀp&wBG̥P￱߂@P (ț7F;0 n!]2KĀ{l<3C&)7эI6S:MS~ygA4J& N9hVBAYvtGwo԰No\a2iw>lg+.,RU!A$ &8T8k2NDR?SnBv![q;яN3t aXTk&a-jo0'>i5gfi#Xk#tA(8\E F0- v aF _Ԅ :y5v}2+tKUM8(-X~nc_Ls[G ot8Dd h<  C An2J=^ 6!){j{-&06`!=^ 6!){j{-2 |\xڍJ@F;i-E**Qh ;A_n/M)8Ib@;; 7(LWm|e>ٙN;<:D ׿zzE]k":WOڳJ*ۭv;Q (R~,TvP)0$TvݕPFq:"kԒjsݓ1"V+5:/;y.&mTݣ4U^&?=`SehuzMw$~cgwdGmͽQLe\iTS,ӆkp{8qIYnCXCh@YazMh "]M:`T\λS{# OrbV\-.T&[UuI*AJ9ug /6TaXfzaq?OU!B$E.VPDzef/o3QeBTuɺ~X̮ BE)@]GE-C $Q\_+;x]iȰ t-xVvrh<~o81CLޅ[E|{q똉pHŬMg4V&= Ҵ"0\TJLY-mӶfltܞTm:oǦG[K*)񸞾I2&:6ψ'\k[6U;6+M@AU1VNK2ԓTM>~L]Jfrظi f\L7&2~: Q`8q6.V9 7[^Dd <  C Ao2m~E'~QCMple4E*$#3$dCdC1(X*MY.48[g$CwmH0poE:"Ģ sjvknZ6raͽJ9BK/ÕiQT[nv Ȳpof԰EMgk!_3Iw'sy.we;ܢEkPL|~FWeS't4Dd <  C Ap2Hny;BV;6`!xHny;BV8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȱ ~x V8p( TY(t36C7 M-f864v ~@2l$H:`3C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOpDd <  C Aq2ۿe]Vk!t?6`!ۿe]Vk!t# !xcdd`` @c0q3(Y,P1=F (`d?3%$ lʛ! 27)?~x V8p( TY(t36C7 M-f86Wv ~@2,\n7h*6S5821q11PJu<.XaCwpAXU# !xTn@8E %ZRO􀄠3rubjZލ*TB"p<PfwFdow\A]%sJ6L'k_&i UU~[g1*wzgyMr_~z DLH:Hn8z+r΀%{נZ+PeTD\QC5$dR?J*kR>ʁZC^cK`6kT4ن  0Z6l89gCDuݖ͡ml ,}/fUq{A§͌0X(#G>2$@0;3o̮g\F\'5rӺ\#(d&51I⾄kMJ?ahdisBOuaͫON)Mb#]RI5FZݯ9cokbu\)e2!E&ƕ~SY6HRgtVW_-?j^_7(O4= : RІ<@- rz*6v)`i܈.3e"r%n7R'\H7f+lT@fP߂PHnBN3ћ󌢾pGBVp{.8!dB!jZ{xrܕ%&M37TAgzԓ<=1$G3`|VL L@3-<+7USu_.@(fg7D`"u a)jŢ&}P9=cn40!{!.F0Ϳccm:ЃHQAвkuhfofgkeVfo|ХT-|>O?p+Ւzg9Y:ñ  iЪDd X < ^ C A[s2ndK1 :I6`!ndK1 : !pdGxڍKQϽZ+HE = D/aE ڪ[ =t^} )zȄ3ww`8; C/-:P}Wg̱(Q {Ү C)w߼ߦz޽)G"|vy?g[ ー&OgQ7]%lnSJ_t=Ƒ}N@! ,GYUVAmuٗ؃D\Oc@J>#%LkQ#늌UaCRYF6ӊ1iRQ:g?݌2`|Zk.@~ge Sg=ӇL_( ZָE;MbRXix19*ȣKNf7nmdPY0+m 8Kk< s⮃4hs^E<^r]6H[WҠ[^$/<G |35)[ӫ m.j~ 2o[d[SC<;RR\ƀ;>F +WS^W.QnMr)xU@ Sc  C% W|c l*lԦo^˴6 w\PǑVþRa_I6<[ӄ嶐c&YKB >79/# 3.J3Gzw oȐ+=6w>;:`lt}ٞF!]a{ۿ/hK`hBaa(ao.$A9 gC,Nwϣ{MqOMB\&bîvLH<+}SM\/,ߒX*`EM %(ZKE^^'1([Z$CɔVrӨe9E RbL$Ѵ CJ0!L? 56CV";No&Oj5QTUz.z,P"?-I\pb&&A!Ue_L1tK=q"Z]D(Y/h2*3 vc$jZP2LSųLsUN5ْ\9'z6tH}m}m,FG9!1w_ZGOt1j,Xu V-g_xO(#\}utuDd < ` C A]u2i7mFطgdjV6`!i7mFطgd Rxxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?~x V8p( TY(26C7 M-f86m;W?v @2BٍXˍSn-` jh@ TgnFC6Ӕ @:3 RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g FMcrMefut "\+!2B]` 9lbsb]XtܟxCڍĺ6AȰKZMĻ!-ARf04C7 H?K#/i&d[|xTn@-*!F\HDK'iC(C8@8q@NLuTj%$$ 8r/'|ev@7o4|JlH Ym!e ft"Ѝֶ;Cum FhA՘+RRP5prc|/0T?.󰶆a]1U\`~9r VbQS\6tz ibs)V,'܍# $$>w3J8oJ̡Rߊ]_TPSpFрQ|t:TڱIt1wZs ǡ(]EO::Q$SZM [y^A~œ4J3]q̔;:W/(z#9?@ߓeʉyğr/eeʅdd$ 7Yha}(eRKX#/+ N[əEGh/CqZ}ځW1tjJXYG 8n֚cIJjȅY W+Xs [JYzv(6Xd(i&M˪Yvn~b&0S {jНucG-hK?l4P?!\PpKC}'+*Dd l< b C A_w2ktT22j=4_6`!ktT22j=4ېpxcdd``> @c01ȃY,P1=F (`d?3%$ lʛ! 27)?5yXr.P5;VdebX4&C $3784vMcrMefut L?5+!2B]Ȑ `#A*gIFC)`{vy˖pxTMo@|@t *.S -iJ(CN d7UpDV~GnoPfG%#͛zw2lJr .`b6`(Ii^d/UyK_ɻ%XV؂vGXF:jޱ3jߖ/`޷QN}ӔKBjfp^WTeeQV:^ҩ=K2c:qrZ; %CM̋XQ* JOb:NOscH񤙹g ko^Y,*{zc3?CbfBӦ-cu?gfź=7ϴ ѡ:q(Ml9hr崥rEkÙ=x>M^LEeaT>o燕%E:I9ZDd @ < c C A`x2s6gJbr=G*Otd6`!G6gJbr=G*xڕJAxKUQAP$iaBD-ZE AE!I#+קo#bnfL_=g ~XhrVj/MrU&7J?wҶm'Q+VV#m|˗~wg[`HgbH \er/@/DatRZΧlP;/#4匯Bp> Glŗ+,\*DQf귋_uB\orrIJ.kDO-b{f[}j^f/t;uTϺڊUpϺv:kkz׀)u4aαF08z̹j-?;q=aj`,q}f]MXSt*º.Xg]XW?Χ.sDVtz]Ff-[K=_]ٶϙRCðK&dV(>PxUKT>I&.*ݠK7"!~ H Jȱ'fl'h2R+$*MĂ%ù׎0hj=|w/?&7ٞ1 Y䅩 &{8 7N X]@Cل$O8#+\eXKӔԥvW‘T*},KF9Z!qeShՅVSUQf@Y71kzW{5^^M- WRmqqb-?sRdzS:a'^!ԠƑd>!;L;:D -H'm4ķRF&52q̢g97NWz% .T\ȎVi.4fjH#h0.~X-^X)S>E<3̔lp;2֧XFAJxP)_8oqtd^O}NaMh9z񷷽v4[h,Jy]7Yz1(^e"at0ژʺS{ַ_G;[z|z.7dg"0d@َ͘G[ZfS%F݌µ !i'x 3L,k MWUgZgSW[x)Oo2[Kꦮ 1֍vʺ4L䖧Сޣ=M߰/f#\+ca=HWDZi2GQ`2 W7DިP01?6T[83}8,D1iWV͗@+w1Wղᅓ_1KIJƿ?wʮ5$rqD.ڛa<ݯ9YdmRd&+!]"۬uBv<u /-rߏG.q vDd < d C Aay2Q٧$o51wk6`!Q٧$o51 Rxxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?ugXr.P5;VdebX4&C $378ش\\Mc22d}uOd7b t]\nwkI0lPCb(H:`HrC6Ӕ @:3 RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g FvR5"ȴe]h ,J !Z=c)0Eߑc,%cʳat^GuM0д*u,2KJk r2DS˼ LI1UeuiV)w<kD8 +F_rҕd>!Z@lmEU *WuQE}fO3Sߜ5Ց}/{٧H_YTDd  < f C Ac{2zٜ_|U|CL˚$ǤȏN 5@jNQʇu#L$[E>;SڮirVصSZS*q{忢p{ڍzXހV.DsYlqm9䦝 ׬#5F+8&IeX 5JjsYVM׏n\7 bG_qi1bfC{1$EG)(!(d6xU_oD]4ąt {CDD(-ru>%QZ >}jx⁗J__@Y;f3y hOX P'Rg_U|I ov}!52w/_wAqp n^xphʤI =oSpUޑZ DR 8)Tr.v<9fY&5M7ڲ^rї5!MŬbRsx $AhyuzOҒǗlMYEk̥7L4 _19iBNce4vU|=\nHp;b/Nfmb,y~Bu:V}G4\o ,!O_9Ʋ%0Il75'G:9 ^)`Th ґ@yb'~L^cJ C`H+LY4GibizremIJJY/T3׹Z |\J"٣'u=Q6|@kBޓ*#WʓcBq>YEkZUݯWYչwW1~>Ή2IG#ƈTmuHb=#X2Yԍ*:XAWx9 hm"_B#P Bİ3w8P94w%,EFyLĞۂfcn;)06g Y`Jj4Ԝ77[r P<-E\r%.FpIpIOle ObjInfo-/Equation Native (_11382201492FJJOle CompObj13iObjInfo4  '*+,-./256789:V\_`abcdq{| FMathType 4.0 Equation MathType EFEquation.DSMT49q@6M6GDSMT4WinAllBasicCodePagesTimes New RomanSymbolCourier NewMT Extra!/ED/APG_APAPAE%B_AC_AE*_HA@AHA*_D_E_E_A  I(x,y)=Equation Native _11336881597FJJOle  ObjInfo68 =I 0 I 1 "LI D"-2 I D"-1 ()  N FMathType 4.0 Equation MathTyEquation Native  (_11382201670S;FJJOle CompObj:<ipe EFEquation.DSMT49q@6M6GDSMT4WinAllBasicCodePagesTimes New RomanSymbolCourier NewMT Extra!/ED/APG_APAPAE%B_AC_AE*_HA@AHA*_D_E_E_A  I(x,y)==I 0 I 1 "LI N"-2 IObjInfo=Equation Native _11336881655B@FJJOle  N"-1 ()  n d nObjInfo?AEquation Native (_1133688168DFJJOle ObjInfoCEEquation Native 4_1133688171,`HFJJOle ObjInfoGI Equation Native !4_1133688174LFJJOle " 0n<N  N FMathType 4.0 Equation MathType EFEquation.DSMT49qObjInfoKM#Equation Native $(_1138220211PFJJOle %CompObjOQ&iObjInfoR(Equation Native )_1138220226NUFJJ@6M6GDSMT4WinAllBasicCodePagesTimes New RomanSymbolCourier NewMT Extra!/ED/APG_APAPAE%B_AC_AE*_HA@AHA*_D_E_E_A  J(x,y)==J 0 J 1 "LJ N"-2 J N"-1 ()Ole 0CompObjTV1iObjInfoW3Equation Native 4 FMathType 4.0 Equation MathType EFEquation.DSMT49q@6M6GDSMT4WinAllBasicCodePagesTimes New RomanSymbolCourier NewMT Extra!/ED/APG_APAPAE%B_AC_AE*_HA@AHA*_D_E_E_A  E(x,y)==E 0 E 1 "LE D"-2 E D"-1 ()  D_1133688183J\ZFJJOle ;ObjInfoY[<Equation Native =(    6" !#%$&)'*,+-0./1324v5w789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuyxz}{|~B9J- !xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?Y#U@pT9 P\H+21,Pfln!ZplZ,'W9v`S|]sAÂWr%\Fkla Tl Z@SBb(H:`rYC&n9@&A- !xTn@8R#*FC"5NҪ=PB9{w-M*!UG;' yH]'qޙ+߈NnYL=F>SI#Ə0>)P^&⹖Zٕ)/[mO ~ZKUM"0 0 A5YbQ#>QE J0;0Lr 0l%o'X[mh1?nz8v4k@f{;Mgk(e`Ӛ`(_>v)x4oeT:ٞuoڵ~O\j˂Dd H<  C A~2 Dd <  C A2?su:HUU t3t6`!?su:HUU t3t4  !xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?Y#U@pT9 P\H+21,Pfln!Zpl*.1@S923:炆JP22@\kaa`hŠCA*gKZ'CﺺY^pb<4  !xTn@8R#* F\HIZ*Q(\ cݸB%J7C >2$@0;3oތ3[A\'mr˺\%(d*m1ƗQ⾄kUJ?Nh'dwyOgea,qG:!75Fjݯckbu|'7LSdC&me# A+0EHʝ[[R>w(Oh7= =BЂMRE =`vAkYґ7f]dDJ ju2?&4F! _a thF= -D O 8y}1(swĊ(d"(4!d>MUA4d} Rkk=sݞӤiFv*L9”zcCr Ƨ 2T_R3|?0Pe%PUսLB! ePMXȴ*TQRLg" [+b߉?1NzL6J7 =Tg;;5 3ڽWݦ5x2 qH춦 ]J$|Ag{7ʝgJ:Xgفf:w@.gjDd <  C A2s-! B< n:{WOE6`!G-! B< n:{W h!xcdd``````daV fx,F^equ02Ȓd`PcA`7(oހܤr GkS  ,@q]jv İ@ >F }3y="CA~ϴf*:wq&=m &9%'r Z)$gݎK͡{)ALF"|CrUfH6`!&\2>LF"|CrUf Rxxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?Y#U@pT9 P\H+21,Peln!ZplO1pS2e'1L!2Z Ѐ R?393GC6Ӕ @:3 RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g FAC;98DHF窼NSAxpDcQk:^- swb?X.IDd <  C A2f䜒Gɼ"6`!f䜒GɼL?!6 dtxڍNQgf(P`3ZP} |- XPب&"s3g~|paJymɮ8 +_JT1u-^r0*>bhҸo$؍m˾˷|ӕg/gaOL&-( jk)"(l 슐ǾΘx-3I*!Uۮ_E%ιQ#}99k>/7Di^`I>xWʣT>+~?E9 `n]l2RB&|]^)%_߁ӢʳDV!n`a2%fنfjtu{QY̪׵tp+! kUo˦ѐB_{ :ʬp{\Gv6tmFW4` sU`:8fF;{EkppX3+1wM%aG(+޷x)/#.zf̡X0HŪdULJpȚlP(J~^⬘TFR7Yu&BFoFEB<˗Yaoڋp*i"8 Q1iKQSOGęŘ%8I%6ʪȎqmgB/e{Q0Y`Ta*y"#qښA6G Dpu,?k̝ik/Yfbi&agȱN$Y[RAuvJՌ&sP9$ea1.F^*>vʸf*1_(rЙXi8$IkLVqM!F7?\K+N+Iu|="Y%1ld=>u]*̮-h.|^i*}7=wY'1e MJlC.qvn]<lyn﷈=EҾMᎧ@w`G QnLߔ|roٟxqv6?xw{߹[O&,П@xگ_OME/Wx#k?'eR߈xDd <  C A20)2B~'4"k6`!0)2B~'4" Rxxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?)~x V8p( TY(26C7 M-f86k.v @2Bٍ؋ˍSn-` jh@ Tg5G&C)quðI RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g Fwٲm$}~)Ab7U֬S&v<~w9i_$y#dj Y$z"7_&+=Q/Ml4NJ*Ku|u/YD.H(%ٔ2>$ʵ]ځgh6ѥNZkL҉aihѕ>B4z)0yߑ6YF%cɳ{nlܤGu ]Ѱ:8d+JgJV™ezO A1UentFU)<ɱL8*F_rbЕd}CL4Zac{ORfWW/}{W ꢊTWO ~s&l\GY>[=; WDd <  C A2IU Z 2N6`!IU Z 2N- xڍJAL%Q *k J$EҊւ`@[R ,lH$ 6*ٌ Y,gwVȑsAmd\kd!1#DI!ȥZ:܈:ȍgwnT˧|ԑa`I,ϯ,Q| >yxLSS(HNu IڝH~啠(`~W2Ɩ)mzx,_Ubv&oyW\#`(uZ/\(5-KkQ.|h5 ?Ǩ?os#ps#:];^rYp FI.k?+.3P7UC}J*@I6c-Q#InB^, v4k EGse5Nyyre뢝>5siMACsUg-/IY4rZ_j/RhPE%UWj]wEƀC! %UH - xVKoT>IݷDtU 0_3)M J3&3`;D*RET$vT,/ ;Ts}qn9s=w|sih!\W__8E$?z4M&^ɧov?Ith=`>Kd7` 4DkCWCէ `Yp/)uw$~M04|AU0 Z '~n qZ+5%D ĸlB< y :j,D>"JLvSV)C6djB}1%-15tq{ VnkU@γ~l/kJWy\ګ^-:ԝL܆~, C/c^i%3Y$Yz"=sf鎅7sY,qtf>CmDiaWy 4AR^DilIiO3wz6a;0-n0nppG8zRu(aVL~&L,&itQeXHjareM.=f)g0u %.gZ)]6+]-=-&]m9Ex6NpL F:87l!H.\.Hw NIMg^-Dir is=_hH]Rv#S`!> "o6`!Rv#S`!> "܆KKxcdd``> @c0 2Y,P1=F (`d?3%$ lʛ! 27)?YiXr.P5;VdebX0bOL@Hfnj1q'?`ddgf~110BL[&FA,a&~( $C F3y;8C5(3ſ8-c<KKxTn1nRRJPR9Q($mʡ'mv^mR q@p< OP|cۦ:Ffm}!16r\X:7{;|i4Y!~F|$CdƋq_T6wl?o f|wV3~<~/9*+O 6#kbUcc*1#ȃՊpvpɆ7#W'Â4 1+ {-Ej o?е=/t ZFaJ<2 ty V|t[%[kA#t3л %Q&%Pe~cQP/SƗjȢpJÎ1!{hX :ulJ&&J~r1^74)dNgbؿZHd'% l4/+$e.iD~bIF\sbRB6:DW햑CS>yz%-4dV XOF̏_+=^aFTSy܂ĺh9LJ-o@yP[m@þzwh|'{”Ev!:CrtN%̫_5IZ_ښSqw+rq1{' vDd <  C A21t}&ir=6`!1t}&ir Rxxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?j~x V8p( TY(26C7 M-f86m&!v @2Bٍ؁ˍSn-` jh@ TgFC6Ӕ @:3 RxxTn@;i *Uj%$4B*NM+r9:^ZM*TDq M|ev$`$';3o̎g F8xڧ.bR)-hZ2=Š&5MgQ_9|?1Dd <  C A2H1G vKlU6`!H1G vKlUێ hxcdd``> @c01pY,P1=F (`d?3%do/PF I9 @L>.Gb X@@ZaBc?tӘKi`TF [nl*d3 !2MdQCl$H:`FrCw&HW hxTn@'-TԞqHm'rEBP=ȱRgn\W\g(O< @];M*&;3|x<k V`v `V\\ /EѹA{u?| bfQdB} UzZY2N ugyJ*+JG"QiIr,ՉDf,TcklzvM Elt1/T;c0.O2UaA 362D͍1q)C@( zwzNf4-o6W{#`}v?ofnhDd <  C A2s9*mk O]6`!G9*mk h!xcdd``````daV fx,F^equ02Ȓd`PcA`7(oހܤr kS  ,@q]jv İ@ >F }3yi#'CA~ϴf*:wq&=m &9%'r Z)$gݎK͡{)AEy=U3{[E 0I](m3UPrBYОCY\JچC5tΣUZowښJ>hQBTE*Q# \Τ\QRQWʣ-\C_i\I`[è1-OF=ݎ~!%CO>$]e5H%tYJY)짧)XʕM:'x]Nt_{wgPC+X- :YϣXS(Ugc1.-G%k;fYu]e~.BYTWDY>ꨛ|R"+]+X.F_e 'b*_gd]1ȥZWk~-?偵{1Nhл]?y\@QjCNNwxzV{L=M>^Tl+}I/^*WX7QY7PIΘwf}X0ˣ% &rGEfsg51Q#GLXVe CKhѻE}AP;cpV5K;xWo~,YViꡤ#IB@n]4j/QPRL\JhDHѤG&"/!&rOoVrw}3ͼ+>˖g3|X+>뱥Ƹ_/+[w5ǿi{{vl/~2Mj?aG5X̡RnUDZu\ɮξ¾^*1e">YX7ȥ/?~;[ק/f //ie}=9?d߹*<dhԌt1 )_4KQ4LcXmg@]4bސLh75uH4c~[wLh1:V _u3c&y.7m~NZ7.S7Eks+\>w  6oJ>Cl(܁TX `{ȕ#Duw>vK&Q2LB9ܳKj(NT׏pwRA xRЦasm{gւgF񨲺8ۏC0.Qey;>L #=]aҢa?RˀR2ن Ĝvx7=aDNG(n0NB?eֻdL]vɕkMlnAs_ypے"mOpS,@48~Zu%s0]æܱwqwJJOs9ŵ:8}Rp%O .}_0ҵ9e`nmQ hGw G;v40KіS,&p|ZYoYd}0x8u0E$@YLfAbHAS 0$4yt=KGγnENpdH?J40-a߉x_Zu)\b#5 Kۖe+oN=NsiT ~S) sTU4Ne]nZkPVڮI \vgJmZiNVS7P,.ZcÉ(g<ē_;{ٻ&ޒ?ou|<wY0Yz6ߖ<DZySlLa<]6IY(ogvX_c>OY2c˙N|xE3bhn|\)~I#9ٜ|)#z&\j:ϕ j̓5J(Xɧe)Z/D ~_;˞F)5Dd < t C Aq2~iC/(|\6`!y~iC/(|8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?ڼ@pT9 P\H+21,Pfln!Zpl\!5Mc22ʄnF 3y:4C*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<Dd h u s *Ar? ?3"`?2j'pVs|6`!j'pVs|p hxڝTkQymͦM46ZhDIOR]54%AbċEozz?(7qfw4G޾ɼo0RXdєPJ%]L]bE@W'INI~O8Gvh~IXңs<[lM_ѾM2OΧ 7N3vއ/["mOq;0ɚ t7+uOmƱQ?U0@|T1J4T3kjs=/B  &X ְ|,k6q2'Yal$~`pO`| r>LkoJEyWlZxXS=ZߍL6p6E}w#\qތtZΦ`G0van%tH,&ĭmu&~ẀK̪tBM%,$;w0Tf4y%s.gv^+LHR|BZlJ+q:Mc! ZT $\m\'a<2K,Y򁰬Y̮zӰj̀o7翼ΫAzYZ#3$ X}{Wg8Oz5zBc5}G3Y5^xf&Zo[㧭\}I?:4Dd < v C As2Y(b+>""f6`!xY(b+>""8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ھ@pT9 P\H+21,Pfln!ZplZv ~@2l$H:`K4.C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOpDd 3 h w s *At? ?3"`?2yaH-j;[k6`!yaH-j;[k  hxڝTkQ6͇nҦjQشZhIBlRDBQHBSуPAA{ЫMh|&`773 @ 3K7 U: oY&w t(4`CM 巸9.EB8Y|/j$9xmc(j>/(bZ. T6?K5 52Y3ne-WwOVޱz:na5 WT^[< Έ\r2,_`pAL Ua/amcma T),@hzsDtɳn>l@>l:ܒzP垡syOz*פ҂;>)m~qg`OixBi_Xs>>g",9ᛩ/^LRwr4y_.]j Ea>%VCpr|:PYEt-xJ@T($?2XV*fG(&KB?deȃq\X <P8Oy< H*n؞y&\IzQ;nf7gQ fUc\_LӡL"CRwE9]Mw9r8*~>~Jqa2 C-9Y1o kO5BUzsp?{A (p($C|pz~Q MA|pB m%LjX(Uzcm&'p9g77[rk'P e2DH2ÉXTG]3HN!,]?@#uyLfӒMP66x\~BRaXVz*y+ΌTBJkB.4TTkh}쨬ОӅ_5P퐵YXh[E2 ICԤt͍_*w`)V{Ò pXw#y]g.ei=7~3;)W;:'1k<Nݝ߄%רjFgNjDd < y C Av2ٕ]ٕG{ͣ6`!ٕ]ٕG{ͣ܎ xxcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?i`R~x V8p( TY(16C7 M-f86v )@2A@7*r#TFy0@ F 3y5EC`gZh xwxTn@'-TԞ4 A9DGx׉ky7nP Rp|@];+!i7oތ׻n5`^\K2s¾o4#Akb/:rSy}L_o)s; \zy:ChǮ,ڻiq iRK,8Y\kx$4uـF Is +Z?Sn4*J\CG%AQ +|wzN tcݚ&H6kmu:>J͚rۏ;h tk M#;i+UOW$֭y@3EuAbj4UX>"֭~@S$@*g|CQ"hxd,yWG1 h !xTn@^)@Ԟq#^C{pCBNL]˻ISJH"zȑx쮓H|x<5j\xc3 ڷq³lO}-i>A|$wѳ2˰EƍG.OY Auև_s dIV$N02dc&e# EnN<2`YU].Mb1SU yC}Ճ˅/GOH#hT! ]ڋٵҠdw'NbVp>:0C bTb u(G;+J&F~䩢(38h yvdH?dBA&;#巯yf_oG_8>]YQDTDd"u a)Er@%*كIdBCD9(]a;Zo7h9 ")W]s j6" 5.Ԛ3@yDž6Nso(I?\h>yY 2Lj{f {aiŲϪ}|"Wtp4Dd < { C Ax2-ڄ^$q6`!x-ڄ^$q8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȹ%@pT9 P\H+21,Pfln!Zpl^vMc22ʄnF 3y;4C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOpDd h | s *Ay? ?3"`?2+]tAoFcզ6`!+]tAoFcզ L"hxڝTkQy4MkhQؤTд֐ҦzEMSHŀ"T<H!BMhh#od| ݔ/@$Z͐.K-vN!<{IdsA?lJ[.Ged~QOtJTWTS 'Hu:Wi{_I'kb0fVcwL"ebxkTZzbZ3I'WM O!WB'X{a58Xlw%R՚ +쐕y[;;6Pl %\;WnTۼVnz<^~+7rwjwݸd ~cDZ01 $<G0vaq*20轝Z L&]m.yO˂[9-YUu+S)m/-7#HhNW]7ϑ /IkrXJeC3ABCDX_D⁂WI>S*4R%85;(j4 ", N4@J 6o?Υxd=zTV#+$ 7}{7g8Oz5{BZxٜ6<3gW?lq'$'N9/Dd $h } s *Az? ?3"`?20Qҭ9!qNzZ6`!0Qҭ9!qNzZӒ#hxڝTkQy$Mk گ44M*T 4MI"I`H<Az-x֓8h#od7of `x)-(JtO:(۹Ūho%PbcpJ-_оp<\fٛŠbFixb)$ͻ_|p*Hsbj rr9OyۢA&F0 `ʱOeh?`* MU .M$]N%pX kG?Ony- C ku1*wK :.2Tf˳6YAgDj Q+(Fk?2&Wy=WVn:`;WGPdGa/Ʀhbh5=xD}Z>xLƎ=ay>?F]"s3Q+D<6o*y%^e2u+| ^d Ԋz³{InЌ3Fq"F䣨)hj3EB?i7H<%0:85Z ,@B"kvPlv[(J N`h{3\Ot^0N?H#ũ5CuN|=`wuG6VOTwT \!Փf腇l޾;ɼ :5Dd < ~ C A{2Ӄ@]y͇d!6`!yӃ@]y͇d!8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?Ẳ@pT9 P\H+21,Pfln!Zpl/WObjInfo]_?Equation Native @7 J(x,y)  L  n_1133688190XhbFP}LP}LOle AObjInfoacBEquation Native C(_1133688193fFP}LP}LOle DObjInfoegEEquation Native F(_1133688196dljFP}LP}LOle GObjInfoikHEquation Native I4 0n<N "I  k I k_1133688199nFP}LP}LOle JObjInfomoKEquation Native L7_1133688202rFP}LP}LOle MObjInfoqsNEquation Native O4_1133688205vFP}LP}LOle PObjInfouwQEquation Native R7 I(x,y)A I k ="I  k +E k_1133688208t|zFP}LP}LOle SObjInfoy{TEquation Native U]_1133688211~FP}LP}LOle WObjInfo}XEquation Native Y4 0k<D FMathType 4.0 Equation MathType EFEquation.DSMT49qt@6M6GDSMT4WinAllBasicCodePagesTimes New RomanSymbolCourier NewMT Extra!/ED/AP_1138220284FP}LP}LOle ZCompObj[iObjInfo]Equation Native ^_1133688218xFP}LP}LOle eObjInfofG_APAPAE%B_AC_AE*_HA@AHA*_D_E_E_A  T==t 0 t 1 "Lt D"-2 t D"-1 () DB(I,T)Equation Native g7_1133688221FP}LP}LOle hObjInfoi  D I(x,y)Equation Native j(_1133688224FP}LNOle kObjInfolEquation Native m7_1133688227FNNOle nObjInfooEquation Native pL_1133688230FNNOle rObjInfos0 DB(I(x,y),T)<D I(x,y)  TEquation Native t7_1133688233FNNOle uObjInfovEquation Native w(_1133688237FNNOle xObjInfoy BEDB(I,T)= I k -t k () 2k=0D-1    DEquation Native z_1133688243FNNOle }ObjInfo~Equation Native (_1133688247FNNOle ObjInfo* D[0,..,50]* D[5,..,35]Equation Native F_1133688250FNNOle ObjInfoEquation Native F_1133688253FNNOle ObjInfo 4 E(x,y)Equation Native +_1133688256FNOOle ObjInfoEquation Native 7_1133688337FOOOle ObjInfoEquation Native (_1133688340FOOOle ObjInfo  D  T* DB(I(x,y),T)Equation Native (_1133688343FOOOle ObjInfoEquation Native F_1133688346FOOOle ObjInfo I(x,y)= D Bmin B(I(x,y),T)Equation Native 7_1133688350FOOOle ObjInfoEquation Native Y_1133688353FOOOle ObjInfo`,3C*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<Dd <  C A|2Hy:1t54H`~$6`!y:1t54H`~"xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(16C5mfM-f86m*n1pSneF SnFpS5cʄݭ{ukTbݪK[u ֽĺhn#֭@SeRPJu<rCП\^ssY̢"xTj@KS9nzBot+H6&3!3e-H{! ;#`=3fS(:0swDuEnk-$uɝ+,k Z#G ZTds$I,y@*MQ6>yg!˹u@iJLxȤ?u*Y!0lA:Lc_+u wa8ۏdTs_Ř>Gb[+Z; hvRE =`êt:;VSұ7n(Sȕ-h6UKp4F!32Ntz7׿.ȉHrb>h+ž?L /Ѐt<3]!Ӑ5HݫdVJO#7RAWzʙ|ԓhQE J?c4Rw!-0Ϳ촠c-{A l1(_eph;PX`Fa:tvP$aS?rn{1_'4: G=vV6k)R4R%D7AWR>K͸Os%6}Q*R4Ɍh_"3N*r8D){˂q1C3F"j"1t {їUfe(M2/fE0,*=<TW9Ʌ\4Sj(~楓g75F/fӅ_6R팵0XKy"r@ag,I>>d\pK`n4S,C@(%[pZ;={ÖەgƋ#6kQc2NS,⚮;g 'SNo+wg1O0ކOP3Dd <  C A~2r\+}P| 6`!wr\+}P|8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?șupXr.P5;VdebX4&C $378شHN._1pS,&4w3Tg44aCuѡBs` 0UxTn@]*T=TjHӦU  zڻM*!q !|Gn7ٵ86yx=kb!{E,v\/$ wM2{Wb/r\tSy>fUc\_LӡL"CRwE9]Mw9r8*~>~Jqa2 C-9Y1o kO5BUzsp?{A (p($C|pz~Q MA|pB m%LjX(Uzcm&'p9g77[rk'P e2DH2ÉXTG]3HN!,]?@#uyLfӒMP66x\~BRaXVz*y+ΌTBJkB.4TTkh}쨬ОӅ_5P퐵YXh[E2 ICԤt͍_*w`)V{Ò pXw#y]g.ei=7~3;)W;:'1k<Nݝ߄%רjFgNDd <  C A2* ?br|06`! ?br|01 h !xcdd```d``baV fx ,F^equ02Ȓcjl@?H @fM\Pp~x V8p( TY(t36bd35i`TF #nˍS.ʅa*v%֭y@3Eu=bj4UX#֭~@S$@*gЋ|CϧFaD H1 h !xTMo@])@Ԟq"^C{pCBNL]˻ISJH"?#7oPfwVdg7׈ZGW .OGLBmfxp*luv_jVyO_o:']2leqѠBqP)餷5F;Y{ګc YFLy̤?cld},뤌5XVUgt dXTUBbl_rKe#U@Bbv4öO.0CxV#7k,}2H|IՀzmn;)01gp@SG4@j_n7r *B)O Hr ='NbV>:0C bb (G;Y(J&F~䩢(38h yvdD?dBA&;#巯r̾D!Tup"|6U Y%ɩ0u a )Eq@$*TكId(J{!p-0ߝC㭷Ph9 ")W]s j6" y}Uk]5g m"H>~\h>S]Cq6`!R`> !Kxcdd`` @c0 2pY,P1=F (`d?3%$ lʛ! 27)?șutXr.P5;VdebX0@@ $378ش|\? ڡ t;LXL.7BLe L? t?Ȑ ƐTgP[C*AهP y(W !KxT_kA\bmW-*2[UmM^reݤ)iA蓠~ |M|:wI :o~37+0 \o5 ezR>6ڗhm['<82I Ȱ w`{'#"l#PIU*>~\"ݱV捓 Phf?mغ G\Q-(SޢJq:1R*xV=*waImDd <  C A2Ȱȵ1Q)ym6`!Ȱȵ1Q)y !xcdd``> @c01pY,P1=F (`d?3%$ lʛ! 27)?z@pT9 P\H+21,Pcln!Zpl ^.1pS~_B}`;?MS.t"# L.sw`V?pEUvVhPGVuƆyW^>{Aka$n0qj:@Y9XMn3db~PP9>q^٬ /"Jj\)5on R JdRN2-)Z7'{E9G=e4a#N#1 o/e(jMf^̊RaXTz.~2=IB rL:ƻ7Vʞ7fQXH_~ٔM53\cM晶 GM!c %*{y% QKwj$robӞÞǰ}. }l4@^NT+^Isg<<}?[`RKZ_s*z-Iö9?tUDd <  C A2G.1eM5#6`!.1eM5 3xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\PL:@pT9 P\H+21,Pclj& ZplZ*v @2bs#@7FpS5bʄݭjĺh*n=C[uf VubZ 4UX%֭@SRPJu<Cs4rii@RjTHY 3xTn@^U]TЈ %AC/ncήݤB% 8 FH<ev8TH7̎wwq(Y'@Wk t$/$ IzךAgh釤{H\z5tyL$<$]ƨe3/νqI,ɄGLZ1(LN4lA}F[wT wah8KF5UIs/:Q? Ck l7 q bF{;Fb\Uұn{1(Sؓ\im /,Jj415onpSr uȉE{Y9G=䵂a1e=ه&D,FAj^ϓI" %N#/RaWzΙ|ԗ?ɼ;-jzm?qaòyˈ|1 9h4 ]um(ҋ#_sEāzB4O-6>Fw+j/,yfcjO3+DsDDd tT<  C A2o:_ D$6`!o:_ D9 K'xcdd`` @c0 2XY,P1=F (`d?3%$ lʛ27 27)?y˥@pT9 P\H+21,P4&C $3784o61pS<x!'^SnTFpS؈9ЭĺV^lbsqnebPd` 2\s+/8LJĆ60T [Uu+T]Bb S|I 3yW2C_Lu9 KxTo@~vڦ 5PR;$jK~6M(4Crsb- *!!TF$ad`;; {绛 rpI&cOUli4gYj=Q0|t?? 7U؂%X[x|?= A<`Nގ]Гz_ @VndB>a (]20-XgTVaNȓ57 soAV m<9^ ǚDYcmN5 90.J,y|/2T?6sVB۵"^5ppAiP43;)8g ֺsna{emӲߕKl`gDpDd h k s *Ah? ?3"`?2-gh3/*.6`!-gh3/* hhxڝTkA~3M$Vb4$6H0[@Ҕd%dRx !7Ȃf7üypp<@l8)!Dhݶ䜭PKk4Af 5M_Kxc{y!56mEqxIgh(V>Y*ӓ2F1T|ߴ+M @xԇAT|+)O/mtlb?a,t1:60Xy%5Wˁz%pXn;Nw1&FTv=Sx@QBGj2Ihk8AI3r[r>-ċwt^0U RN/1#npjn)QW)ir (`&f}UCIROښY4+>\CA^-aOy  ?z8UP|+yF% Z"= G0yp!􎎋dD7 {h[`WIםK0wq0S-zsK킷+÷A#_Dd < l C Ai2= D%p^OT26`! D%p^O" xxcdd``.fd``baV fxx,F^equ02Ȓcjl@?H @fM\Pg~x V8p( TY(26C5mfM-f86,?v .@2bs#@7r#TF@S9t#nK[uTAbD[CV}bݪ 4U-A Tg ƍC5:~LKo`" xxTj@9N4HP\*S_Lri[V# 8K(4豷6O҇4]ɲ]X{曙ofGZL؅&҂Ֆt}|E1+7fGG7,/i y ܇{*B7u҇Gm˸tkYS7`8fNjYmrS4I,:, ^37YJr)G. J;hev͕|e*t>EIj맒Y~PBN PC6vY:>0E%\ǹ(CQr F }3yV)"CA~ϴf*:wq&=m &9%'r Z)$gݎK͡{)A.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wSU1$/ ?6`!!n5h>U1$/ âO8O8xcdd``db``baV fx,F^equ02Ȓcjl@?H @fM\Pj~x V8p( TY(16C5mfM-f86i`TF[Tb*#T DXzM%֭mUT&n$֭@Sʼnuk;n*C[uk%TEbA[4uԉMĺ5V=bIljM5#*ZC[mڠ47p>u<Ct"&Cdy:O8O xTn@^Q]TЈS,$qԕr9T ܐcSgml'M*VP!P#7ax2vG*ofv՟-rLכcIgb D&Ǥ_Q@}0{@ er={G ! < Q$}FBE}roX0,| ǿ#'YS&d{iUIA"}HR[tnH%"P u5ɰrMSlMVI%s=5ُ&/b!iu#sLA kУ}-m[/S\eZ[@8)Grך>Dd < p C Am2G) NL}#^E6`!) NL} 3xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\P<~x V8p( TY(16C5mfM-f86m/1pS.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wSqb7Lt+#FT3 \54CLESJn[g;` h!UxTn@B"*'4J-iʵA/9ax׉knT7(_M8fNJFdg͛xW4|$8:4kt6v o,le+Xf؃۰}xRuF<w߮]!}譎X4ZHDZ S-غ$ca(ԩ&1j_?i1v'Ł&=vpJS4J1$:mc]i4fNhu(MZ]W:ki3epVbD7\=EDo<1$TVs[7V }Ne՝&Be|a*ЎbPCXȡst.J4.]R7ÔTI$"J_$gRKjQL";F曗^v̡7/1].&(CMu3֊Z&."}Q(r7c㐊Cʙphu CE^rHwgF꼰vz]l]R!kclzd&\- @; ;|~i7wF\'ODd 8 < r C Ao2vjNv4? JRkN6`!JjNv4? J:v$8xڍ9KALO BA"h@mTD$B"ja6ռy3y_f""G/jyxH-S깠pH ! \x7旣7ԤW)Z;1.RRy2Ӑ)r]CJ{:qM =kg5r]pf\Kβvw\W~eY{52r+_^T]ES%i9JVSU"Jʌ\ֹ^/V%#WYe"&asCI`+Fˉ]yО v$8xVoD Im4iAZd߻ȁh R.@%Ƭ^lg(TB NH="q7?¥͌&)8xޛ,yư$],>P#["ӑN#,%ߎ~ J0wg?=!@ < Np |#=s>21 ;IE(Lf?#YZ7z)zY: թ P6 b BW#\w("6el 12{oFlY6gdm'٫w$W1Z}L꠬ܢRpLtXب0aG*yt&Y I`jD+j:%"w$x -hm^馽G?R7p2h=ԠkM0.>j& hlSM ;t>uz7M%( {Gm.eD(Nq2D|a0~ϵf*:wq&=m G&9%1$r Z)$gݎK͡{)A"zZIFiM[ZLXPzts_IMkwSsߒYb,fCAVaor6r8Md)GK<ˉBbmF:rdc]hۄ:RZt$V@r!ե͔ ! 4߼rTTNDNo'iࣺk"heD]8D0nf6w`uk6,EHq# QK²vf.A+6mNqfUyYnF 򳆛_T KDd < i C Af2Nۨ  G"K*tZ6`!"ۨ  G"K; hn xxcdd``> @c02XY,P1=F (`d?3%$ lʛ! 27)?%@pT9 P\H+21,Pel12`by/7,b}W + )G}[n]z26< A !"3[n7s\:r{ y$vB6*1".`b>(7`^3©:.z"h 2G9opf=s,s äUGG<ICEwc~]dl>uEWC'4TezYb&<>IE?f\Aivpo0NoB2|O*ΌHUd$ `Q}8RM{lOCTCa+9[`;)Zkաn^m6*v{U&ͭkZ&<[Ӑv͚;1hv~'Õx>-iZ?-U /oDd < j C Ag2NS,N]ye*R_6`!"S,N]ye츜<  xxmAƿ,BP 4HQ(]<Z ;ھ>D資nLN:r8b߉L[G˜_YLEx^CITm_ng'g2*J<  xxTn@]--Q@TNh)FiIJ EP!Tʱ׉N*TB@EP#7/n\(kqyzwW4C7Erڎ$qȝ/0^wDC8{%H erlzq2P@H*dY&+9_oK dU>Pu\Eeqx, 9IT 5kd} )ǜD_S ї&W/U*[P&ыʁ^[c P5p<;{l 0aӀ(}0Gdhv]Ďar\:E8Bf}nkՈ9aa`n QnĜD/7S1:6:ux<Baq7o)snc9mB):#%I:AxF G1&Ɩ\7$`qNjI"rĂc,+(C͡lzي7S\^h.E~VEjHr7UFb Ib -,jl4Nc)0uip7%0Jv5 &\C+utn#+hg n>}T֡3H`j 9@{fS-hf=ӠCQZ{xc=ZjAQuپzTYE>eܔ^An~! ODd @< h C Ae2?A坚᫽]0d6`!?A坚᫽]F !xcdd``^ @c0q3Y,P1=F (`d?3%$ lʛ! 27)?%@pT9 P\H+21,Pfl12llM-f86ۖkv &@S9|1Rf0_4C7|UITB !kxTn@ݤR U*qJ$JǩrEBK.nHȱ׉knD C77]'q%#3zƵs6YԍW;ߧ}׎U~{+c1,lm <@*#roCAx=Y!}A|$A er/;<"d y.z[cԲޙ@ȢνqI,o$ɘLZ1(L,vاA}ƾY2l܅ڼ1JZ93߂᨝_};|@ ~InȮ5 تA£Ġv3sXZ]lt[xnD3e"r%&43۪$̏I xGSj:uH=LߔBbv2Džh(9F|Hў#V G!+=y-CDYW!hjB&!jZxfIYY *=L> IScO0>mPZk-NVٕ(@RM :d/H5"4Sw PbSqL%* MQ92ãlуIDF }ȇfy2 97ZS]v`{G{ ϥ9cvky ӄ <۠ޮ-«=l3<ge|l]WD2Qu Dd <  C A2r8Z^#`&{TȵWu*NXm6`!F8Z^#`&{TȵWu* h!xcdd``````daV fx,F^equ02Ȓd`PcA`7(oހܤr YxXr.P5;VdebXcdpf#>u<T"C/uBD:z3ayQi8_9 /^zgyg/hݐF mrpb:R)!{PyBj=qYm1~>~ϴf*:wq&=m &9t%'r :JS0H>FA;mC]yfl>51沿%X>(SY):ͭhTm8pR8Y)c;6$ڌcu";*[h݄:RZt'VE JTfJ o^y9**G|K_C"{QP4eQ݌b42"d."r7cA;J0ܺ5 ,EHq# QKvfB+a^!b /pg 7w^bpK+5Dd <  C A2Yŋ^$J2PLZq6`!yYŋ^$J2PL8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @R^Gb X@@ZaB7c?T0Ahb`q`7Lt+#FT3 EJ4Ci6yHzD5cT` h!TxTn@B"*'4J-iA/9ax׉knT7(_M8fNHFZ{g͛d&W4lf$08:4kt6vz o,lkXd؃۰}娯RuF<w߮]#}譎X4ZHDZ S-غ$#a(ԩ&1j_?mȓb_ B?D͸R Ȳ4Nۅ9jWb Da*V IC+𸋝Rg--y N [ƒ|k\𡹈(7-'RD*QnyF*/ݩDVQH3L!1\Lj99 wVD•-uS GFX34.B%@(E qZvf. k礋-+U1>d`|2AS.Z ʝs ;|~i7wAB|rO3Dd <  C A2DO?l3[u6`!wDO?l3[>;hn 'xcdd``gb``baV fx,F^equ02Ȓcjl@?H @fM\PqIo?z+8U`fҊL  B3d36Ki`TF[Tepb*#T DՉXzM%֭UT&nu&֭@Sʼnun*C[]uk%TEb*O[4uԉMĺ5V=bݪ@ljM5C˽CA*gNC[j[yvِ;hn xTn@])-U]TЈR8NpU͡7$Y[^'M*P!p>#q\֞y3f|A| B֊ Urlz1%E4yD*2#|NBVd X%QPQRM?'[§JVr7RY6IE!D:-+5VHԨ({eRX9vK5Y%s>Ctrw˝1xBvphv( zhm1X]w a:!Θ&㡓RhY0VKGfGFJbp8_ CP￱M^Sށ(u8N# 1E}aYdvռQ !et&̣x4 X_:ݼG7fxb7i2vB-OyKAB4Jh^L,;cryf'7L'_4G(MCVsLj"e#SIbȆ5#*(TuY.8;0%q6l jnG~]&6hzǀS]xGCIWf*b4B`u.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wSĺh n#֭:@S5u/nG[s+ iXjM[ WMNhPJu<wC/?{nTʜBR)ۺH xTnFRr; )D"J$Q,raDhri!R$ `h [g= D Bmax B(I(x,y),T)* DB(I(x,y),T)Equation Native Y_1133688356FOOOle ObjInfoEquation Native F_1133688359FOOOle ObjInfo I(x,y)  T0 DB(I(x,y),Equation Native 7_1133688362FOOOle ObjInfoEquation Native (_1133688366FO0QOle ObjInfoEquation Native L_1133688369F0Q0QOle ObjInfoT)<DC D Bmin B(I(x,y),T)D I(x,y)Equation Native __1133688372F0Q0QOle ObjInfoEquation Native 7_1133688375F0Q0QOle ObjInfoS  TC D Bmax B(I(x,y),T)<DEquation Native (_1133688379F0Q0QOle ObjInfoEquation Native __1133688382F0Q0QOle ObjInfo I(x,y)  Tz D Bmin B(IEquation Native 7_1133688385F0Q0QOle ObjInfoEquation Native (_1133688388uF0Q0QOle ObjInfoEquation Native _1133688391F0Q0QOle ObjInfo(x,y),T)<DD Bmax B(I(x,y),T)0 DB(I(x,y),T)<DEquation Native L_1133688395FЗRЗROle ObjInfoEquation Native Y_1133688398FЗRЗROle ObjInfo= D Bmin B(I(x,y),T)= D Bmax B(I(x,y),T)Equation Native Y_1133688402 FЗRЗROle ObjInfoEquation Native (_1133688405 FЗRЗROle ObjInfo    k. Bmin(E k,l ). Bmax(E k,Equation Native J_1133688408 FЗRЗROle ObjInfoEquation Native J_1133688411FЗRЗROle ObjInfol ) BEDB(I(x,y),T)= I k -t k () 2k=0D-1  Equation Native _1133688415%FЗRЗROle ObjInfoEquation Native (_1133688418FЗRЗROle ObjInfo  D D 2. Bmin(E k,l )Equation Native 4_1133688421!FЗRЗROle ObjInfo Equation Native J_1133688425#FpTpTOle ObjInfo"$. Bmax(E k,l )  kEquation Native J_1133688428-'FpTpTOle ObjInfo&(Equation Native (_1133688431+FpTpTOle ObjInfo*, 0k<D  n 0n<NEquation Native 4_1133688435)1/FpTpTOle ObjInfo.0.)d%gr9W[pEPE'yfd7inIL(D}LL' Ulq4 mőW _K0I1IsCӂ6 c&45дթAO68g5aBk hʚUX%vCZE6*u͒{vټ> wP=|A?rڃ sh]# ҷJU\)lk'|qaoM zNHۮ-;B%PQ3zt+F?Cy dLWe%яC%_lhojA niX__PL_󹾰q1"஺~~zDd <  C A2 HLQ^H7~6`! HLQ^H7~I%Dxcdd``fb``baV fx<,F^equ02Ȓcjl@?H @fM\Prn?z+8U`fҊL  B3d3kv .@2BٍˍS823&V\Xjg & &Z@Su Rܪa*v ĺu%nE[Cuk%T}bݺ821hp1ˆukЭ6ĺu5 4T R?3oqDCg3fP[:I%xTn@;)R#@FlH&8PU@@!!^7{my pg' x̮ו7|3^O ~G4`9*R,'Ms]b/;zD!JA.{N'x@BƩd/Z܂*zAq4`U^WTF}ß9EF0$8吜+(a\`R:5HƙV+fLųڰ\BuejSc ;[2wUUi͒{};ޔz,Dw/pGR7m8F1C 9 ۢ)e3`{1;XW0rF7A <(0k1\J2RK"QP&VBY5ɴH)ZSFq8pFH&ƌct0-]/PyI'Oy6:M MOبJaU~ʋ(g~RO_r鲙FCeuW԰^oi(eHiu!m`R+{lAXHZCQA8}Y]cNC0w,-t"svGڃ~vZQx\)lwPI0̧^%;};k-l m]GOm|׭U %3A_UwSF閼~.ꚦkpV^$3k S?|;\_X8pI}{1Dd <  C A2}S }6`!u}S >;hn 'xcdd``gb``baV fx,F^equ02Ȓcjl@?H @fM\P ~x V8p( TY(16C5mfM-f86Ki`TF[Tqb*#T DUXzM%֭!UT&nU"֭@Sʼnuk(n*C[uk%TEbF[4uԉM*ĺ5V=bNljM5C˽CA*g_*C_MO 4 WD;hn xTn@])-U]TЈR$NHU͡7$ٵN(TBB'/ -|GٵR+=fx<˟\dկ.2_{ʒ/#&Qr Ĵ_aB˳}4{@6ٺb:%E yDJ*#}NBVTJX%1QSD<`!.S&ds ,bbVZIjN=ֲt%Mrc [4vE9%;trΘzx{MGV8a 1C0GdVcVN܁:c;LNBiCЍJ̼ !ҍ`p"@c@$Qģu8N8Z|LQ_CV4+]y oAHY?@ \u<Dn\ˣimع)%MR~)gI;x1~̀y1o_ee|3)@~R0|:6UYSɨ?TUHTA' !:X$TA3Wa!;~p܂.0lcCa-lsj8b` ꭝ\ZPkz9:L12-5p'^K+k~?5ϳ~ZϦ%. \߈UDd <  C A2G@2Ę\H\VK#ď6`!@2Ę\H\VK 3xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\PrEm?z+8U`fҊL  B3d3&4F0n*#'Э܈a*ЍѸ1nTV SՅXVM"֭1ĺUhnu%֭@SEuk,n* T3 _C+ؓrW#3%; 3xTn@^i U]TЈ ?#As­rubZM(TB!P#oPf׎J|xw yN.Q\@G2BmWp+lt_ify~H_{o:']bB$[+/ Y"[Blw9_չW1>.Ҕ2IK"&ds *bbcѝ:]-l?ΊQ}cZ﵃O@CpDGJCЂj^Cё^c`-rՋ5)8xjA]w5SW{iu0wSeOvJWHt?tQb9VαI-.+2գ~F NKf[#a.4xv]F[5߇xzh&掠0 \6mgIdطԍU"A )I 7#1osS fW! Y`П}^ /m u5gx/wxZqy'IiΟ5P8L4Y4|-+YGOIMRcpBCiuG1_EVoSa҄;yS1dM-iB0RIl`aSs̄$,Si5b7#pJvphl@E~[[=4f5w;j c{K:s4/S0a@ AjP |ϖeZ&vv̹tIr¸i~/ٜX/ir>|)Z]}fayUQ/vu@ߋz^V/&9eG[@TE황Dd 0<  C A2G[[%%֍쒴846`!G[[%%֍쒴8\Sxcdd``eb``baV fxb,F^equ02Ȓcjl@?H @fM\P|`@pT9 P\H+21,Pclj& Zpl1#1pS\ne'1L .7BLed̐h2cscb]4S0+!L*E[Vy SnB[Vĺ;nG[V&M.bbt5n}NljMu"I[7MDh% }3yCsɽC*x׮3(\8xTn@^;iKK]qEJM#P!AT :1ז J 8 x&2vWg%3|f=:kU|"ӵ/#L\u'hWY[D!KEY$7柌Ã( Bn u,zUrģ+~QO{[Bh I(Kg&ɖ%rm뇔ckc-;PbLv b]򌪲ИCTUi V`WZtslA1uWR?.j;)О.l Zք$ ` M!XVĥ#/a&6そRh[2gVOG2̋J?#;C`\OɮSn@.Gq)ya4+Ž;!t4um&>Ix/㜭YE(x4ځ҃~B4Jƙ!^2`ѴRCYupW2vo#a Cpw8du,ν,Sv!$K(>R! =Un‘B{9UrJ>Q~2C[ C]ۊoxm0;$jfv hvZS4 Qh3ߵ BIŸ3io¯Yi7%Snj#0Qyl\Hʦʚ1HvS)Pׂjrh^wQeaIQO+WrYٚW'{ygce{O{X(M) 'brQ:Aѐ Nj!6aݩS3l܅ڢ1,W3&Q;X8 / AK~Į4;Մ1Qg-s3pvK@*L=&bORsa5Fʂ 澐C 竨pvS3cj  JxR31Z7E{Y5G=䵂Q1e}9D,FAj^ϓI2$N[E0,*=L>RKNr#Ə0>kP^LM~Qe) qw!˦UaP@rC6q~|8܄i(CQ+vbsNۅ^[Ox@ش쎃cރV*zLCmcpwۥy8Qcn3q`oLx^{R~uqw_;_xW }<ͳyZ7&)u Ss5Dd <  C A2ozA'+ PC6`!yozA'+ PC8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @1Gb X@@ZaB7c?T0Ahb`칹?Med݌`#A*g4CLESJn[g;` h!UxTn@B"*'4J-iʵA/9ax׉knT7(_M8fNJFdg͛xW4|$8:4kt6v o,le+Xf؃۰}xRuF<w߮]!}譎X4ZHDZ S-غ$ca(ԩ&1j_?i1v'Ł&=vpJS4J1$:mc]i4fNhu(MZ]W:ki3epVbD7\=EDo<1$TVs[7V }Ne՝&Be|a*ЎbPCXȡst.J4.]R7ÔTI$"J_$gRKjQL";F曗^v̡7/1].&(CMu3֊Z&."}Q(r7c㐊Cʙphu CE^rHwgF꼰vz]l]R!kclzd&\- @; ;|~i7wF\'ODd D<  C A2)eSvZe"gԕ|9H]ָ%6)9hJ#ߡ1E͝wFG}W vi?`g-eB!.0GЬNsG` ݁b:NFmC:fʼ$pc~F(7d^W1S¡؁LQSzJa*BIT(1Z#i8dpJuPW X0gy4`}Iϝɤ`kvJ&MGN|t?ffR7I.},5Si(\FQw C|q^04w#l㡊KtҚ3e;HdMi&3erC|`YG-#6`!>rC|`YG- 3xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\Psq}~x V8p( TY(16C5mfM-f86Mkv @2bs#@7~FpS5bʄݭukT.bX ֭ uk%TbXjMBKeCA*gCiHXv: 3xTn@^i U]TЈ &uK84^ *!^'ήݸB% 8 FH<ev8TH7̎wwq(Y#@nt$/$ kIz תAwhds%Y)+Mr2A,y@j-Q6>ygce{O{X(M) 'brQ:Aѐ Nj!6aݩS3l܅ڢ1,W3&Q;X8 / AK~Į4;Մ1Qg-s3pvK@*L=&bORsa5Fʂ 澐C 竨pvS3cj  JxR31Z7E{Y5G=䵂Q1e}9D,FAj^ϓI2$N[E0,*=L>RKNr#Ə0>kP^LM~Qe) qw!˦UaP@rC6q~|8܄i(CQ+vbsNۅ^[Ox@ش쎃cރV*zLCmcpwۥy8Qcn3q`oLx^{R~uqw_;_xW }<ͳyZ7&)u Ss6Dd <  C A2%f? r{;{+n6`!z%f? r{;{+8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @N?S  ,@q]jv İ@i{ 4@Hfnj1qi?عnc7Lt+#FT3 <4CLESJn[g;` h!UxTn@B"*'4J-iʵA/9ax׉knT7(_M8fNJFdg͛xW4|$8:4kt6v o,le+Xf؃۰}xRuF<w߮]!}譎X4ZHDZ S-غ$ca(ԩ&1j_?i1v'Ł&=vpJS4J1$:mc]i4fNhu(MZ]W:ki3epVbD7\=EDo<1$TVs[7V }Ne՝&Be|a*ЎbPCXȡst.J4.]R7ÔTI$"J_$gRKjQL";F曗^v̡7/1].&(CMu3֊Z&."}Q(r7c㐊Cʙphu CE^rHwgF꼰vz]l]R!kclzd&\- @; ;|~i7wF\'ODd <  C A2'R7t]4N6`!'R7t]4Nü!xڕKPϽiզVAD}@-n:tUAA`vtpqN 룎.]_"$M7sB!V\DF<|ze}T*bخvi!aq'~ u)lwҥZ~ zzgZ &V)~ǐ[iP~ӞecL6MuvǜiZTA*lǏTv˱AcQ5e ֏Ƙ9DMj#k@5Aήרka5&G]oPSksO4 aYu8SE]瘚jVw2)1hJ)ZȣVy; WI=uRuy*)qFTRVURu1UR uD|2* Coq7C %:!k\!xUAoD;i-녕V,B8"Kⱝ8R/Tl@!!7fe;لR eE+$7,"q&H7cǞŽd{7q=Woh;^ O5~B*aޅ.ǔ?[ǟ D|>{.RaSOY#A<(ᗮ7sPO'Pi")+4b!<{(Ty B:PPN1Ttgtj@٪ }W=eVc=ŨV#4@3Gj][[m-j>38ޒbq h!"g:NO hׅ@M?1xBsf`ӵ& !9m91 i ~&L20$81[?XZqܞXm>#h8(5w( |2`@a;r']:tAjIő/ k2R+`f+ %ՏDB"m4!e"ap(ԇaN ݴ釳'zy]A)ǛDDꆡx'*ܟM_}V7\Hv^M5WbDd ,<  C A2`2K?2$6`!`2K?2ˢO8O7xcdd``db``baV fx,F^equ02Ȓcjl@?H @fM\P$sl?z+8U`fҊL  B3d3&ui`TF[T~.W\n7Uh*'n}@[=V7R*a*v>$֭@Sʼnu;n*C[JĺՃ821hp1[u' 4Ռ*XMAinh% }3yCnE6.=yaj><O8O xTn@^Q]TЈ-$q~JPEJnHȱ׉6& P+g(O0<@];#X7|3^O9&wMrױ$3I]gc/r{L Bq>|D| DYEv֞N=#du(z[pJ~u{_\(( )eI^CamRQIjN I% R'ُ*0YX +e4dK]2ӒhkR "Q72ďMWvzﲥ@ vѨY Cntb YGdgu _)f FNr 10o}SF1fW  ciCL_={.˅-u{GY?.d Q/Hi,ӔԳP04i86=%-8iG>nH O0L:X/fouy)Pܝ_$R$St$K64Ir}9\"rfNµ3I w u(rlp𘛠AͨdnWj]ߛ9'dBṪިg9oz40a j@ٸJ'6=e&4sM57R&-T4j ~s"xgnah GzQ/|ّb"䋩|qˍw--&Gr'ԁDd <  C A2&2å6`!&2åіH Dxcdd``fb``baV fx<,F^equ02Ȓcjl@?H @fM\Pqsm~x V8p( TY(16C5mfM-f86Y+i`TF[nF Sn܁ˍS823f\Gk5f1@$Z@Suk<)n0 [w *ĺ5XM$֭uk%T=bݚH[4uLM{ukЭĺ5pV*@ TgAC/?{nTʜBR)ۺH xTnFRr; )D"J$Q,raDhri!R$ `h [g.)d%gr9W[pEPE'yfd7inIL(D}LL' Ulq4 mőW _K0I1IsCӂ6 c&45дթAO68g5aBk hʚUX%vCZE6*u͒{vټ> wP=|A?rڃ sh]# ҷJU\)lk'|qaoM zNHۮ-;B%PQ3zt+F?Cy dLWe%яC%_lhojA niX__PL_󹾰q1"஺~~yDd <  C A2d|9Up6`!d|9UpI%Cxcdd``fb``baV fx<,F^equ02Ȓcjl@?H @fM\P$sn?z+8U`fҊL  B3d3&$ȕ4F0n*# ЭP7Dv#@7r#TF y&c;6fZ Ąе\ 3h4n$ŭ 2awknJ[uT-bH[+lĠ`Jl"֭!@bUh Z1Rf08Cg3fP[:I%xTn@;)R#@FlH&8PU@@!!^7{my pg' x̮ו7|3^O ~G4`9*R,'Ms]b/;zD!JA.{N'x@BƩd/Z܂*zAq4`U^WTF}ß9EF0$8吜+(a\`R:5HƙV+fLųڰ\BuejSc ;[2wUUi͒{};ޔz,Dw/pGR7m8F1C 9 ۢ)e3`{1;XW0rF7A <(0k1\J2RK"QP&VBY5ɴH)ZSFq8pFH&ƌct0-]/PyI'Oy6:M MOبJaU~ʋ(g~RO_r鲙FCeuW԰^oi(eHiu!m`R+{lAXHZCQA8}Y]cNC0w,-t"svGڃ~vZQx\)lwPI0̧^%;};k-l m]GOm|׭U %3A_UwSF閼~.ꚦkpV^$3k S?|;\_X8pI}{1Dd <  C A2W C3Ɵ}6`!uW C3Ɵ8 0!xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?:@pT9 P\H+21,Pfln!ZplZ v @2l$H:`3CsA@7"3/` 0!SxTn@^EEԞЈS*)ʵrȉp^&ήqJH 8r :#!if&]R#+:Yl{fDS7>9._Zey&?dn'#{ 'xoO"%d (L W)w Dd ,<  C A2yHYgVf>Z)'U6`!MHYgVf>Z)'1h xcdd``.ad``baV fx ,F^equ02Ȓcjl@?H @fM\Pqs~x V8p( TY(36bd3vi`PT .F W2a1ʣ\ 1ZFW`HCMTkk fj6/2 zXMU#֭uĺ5h1tqbݪ Wmt1Rf0GCP=X|6,21h xTn@;)-TTFl54V7r­rubb- PzBpD|@Frfޛxw,&lkpI33Hn_qB%z$ۇWO5dN V`sLPMЩcw}˰RQP%QQjETU;$w[1daG#ُ/t5wcx/>wxڮ"M|˒{uNZk)M%#;ЊRWTz0'\dc<5Sj(HufFQN7S=9/M.4ΣyS*7YV^̺CI,q}0)l)R{!"VrJvhzvBgj:#ݠYnS͖5,\hz`!$< l7؝S|ND3ڱIM3-#Qn;t+˫iNmu:Vz_YW iM= `VթA9i|Zx\Dd @ <  C A2AП5`"aZ6`!AП5`"aZ $xڍK/QϽgDD,XBDHIYH,l,$> $;MHso3QR*x(Al˴wUa:UR6*͞84I֢*Yt7sOfg`MLlVW$݇BE>ޜyLt %bB-͇jˇfRjaM]uepПgԀ#NHT?J$$uBPP j:̺vºFXUK{GH:8c]pe]Gٺcs5j`]]hzu]=*15 5_/YuP uu5ضԗ|=pź_}`{Rvَ;",S~~d]cNS ,kupͰQPgb]A*u}zTQ2i׼Ȭuڔ7C85)IKl] $8xVIoF~C9qYZɥd9-q-|h!=JO͊"U% pr9-zoo)o.^R8͐sʻ/(|F\?~MR_"ӑ?G\Y>K_L K8~1o&@<-XNpv7O{koPݝ4|q2h#.f` zY&|q߄- HG ΐԝ%+R@NVH"D8Z>9I,>> #nͰ,kSFV+%Kq+//20h Qv19cn—5&g]`0[yW6g1$a Fѣ:ڽE_\,7ae_l8):B7Vi#A$hIc 5B)5Y3g6g0R0΢aSfKYm7I2'tAn'_mVOY :]zN{a#ZhLjiଌoL}(OGa\ηCwR`Lhl٦F !)SVw",n{WՑ=PXժݬQv"U.=GAՏs2ħ9pes`5Q[ZLæ0[@Md &bSl[Brp S:%eTnL(DjG#,On~'zǦqյi[R2 CmC\<ȏpe"+'2V*:G;& Kt:d$ >Mq(Nvc'E0wN3 4y7?ajh_MEnPl,H(T)C j& #Qγglн@@wɑ W[3m'B˺_ՇxPڡm5* TfAEO/)j&lh12@tOAl(t9f^ 3GiUn2X9=|}K4mgjsjbN:7Cm.kI+[&#^]Oo*6?)+\ـ HBYl4.\%pU9*Խj*4h\p(M)u63sT-3Bc;Dd <  C A2s3:+YȷO6`!G3:+Yȷ h!xcdd``````daV fx,F^equ02Ȓd`PcA`7(oހܤr *7WS  ,@q]jv İ@ >F }3yug#FCD('* F h!@xSn1v҈#8'4J@$"qBJ/9h{e;DxI67ߌ3sUPZ>3ayQi8_9 /^zgyg/hݐF mrpb:R)!{PyBj=qYm1~>~ϴf*:wq&=m &9t%'r :JS0H>FA;mC]yfl>51沿%X>(SY):ͭhTm8pR8Y)c;6$ڌcu";*[h݄:RZt'VE JTfJ o^y9**G|K_C"{QP4eQ݌b42"d."r7cA;J0ܺ5 ,EHq# QKvfB+a^!b /pg 7w^BLjDV1K bDd <  C A2H^,QN)|N_o 6`!H^,QN)|N_oF xxxcdd``^ @c02Y,P1=F (`d?3%$ lʛ! 27)?@pT9 P\H+21,Pel120@}' Zpl6f3p;8|1Rf0i4CfOBV|ʐ xxyxTn@]'-KT=J4rE\ \AM]kw D%>|Gn/|ev$`,;o޼ovxm76ncIHnޛs~ES!^wI>XTx@ngH4.b!YEXZ﷍K d޳Q} 4HɸE55wf8-Tm8}Nvn )k[%kt/ۃ 'dx&'Z8H&#l^ʦAla).Uk]1MLؒ>"?[dSZu?,B"'\rbL:֛Wʾ̬!~Q]ES 7՜v+VQi0A|ljZtNŴyΕ;0PQExbUi VͣIU]{ ,*T&lxq`6ӢwL㯷%,{)Q?pqx '¿SNf*KC&:ԃV}nukS;JZijJZ%u0՝ZꖾV:2L\a$xº>6qK0;qj lRX:v+&e`' w[l}s78$w3֟''HC/eBB pbȟ^ >/t wax7a}Ե]e&s}l5 Ey)+= '~qPZ_IE` 5WѥwDd ,<  C A2x ;<IT6`!L ;<I1h xcdd``.ad``baV fx ,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(36bd35i`PT .F W2a1\u-#+0 k*еSfj6/2 ݚH[u֩ĺ5h1tkn6Z: R?3MC؊"*)Ds21h xTn@;)-TTFl5mb'"n [%^[^'M*pD|@Vdͼ7,kLP/̖y \e't^&1'>̐Li.&=P`uaڋӰtv%ȭ+U/`9P3L8yI;K`6:~ejP=*qW9ʽ +eUVxUNrC|u=bV]Iz>R6q$ .r#dh~#lX6v,;`W"S-lZ ['܍#$$!ws_'Ȩ#/EJBq([F#F;yؤ{i-3v]i^$>:Q|ZDk+M@+Ry^I~œ4JNsN8hVL<;:#W2vo<~iytϋTyꎗrfV dMECƀII,3e'^Ns SE'Ż$Gֳ!y6Gtcݠ^nS͖5,\hz`@\{ځdzPDmͩz@EQ،nl1 ze {psH_ݕ;4U4T:S]=?+pCQ괦i|Xw1Dd <  C A20ҵCfFF9vE}6`!u0ҵCfFF9vE8 0!xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?ȩR~x V8p( TY(t36C7 M-f86Wv @2l$H:` 4rCsA@7"3/` 0!SxTn@^EEԞЈS*)ʵrȉp^&ήqJH 8r :#!if&]R#+:Yl{fDS7>9._Zey&?dn'#{ 'xoO"%d@  I k  l 0l<LEquation Native  (_1133688451CFUUOle  ObjInfoBD Equation Native 4_1133688455=MGFUUOle ObjInfoFHK Bmin"I  k +Bmin(E k,l )()-t k () 2 ,"I  k +Bmax(E k,l )()-t k () 2 {}I k -t k () 2Equation Native g_1133688458KFUUOle ObjInfoJLK Bmax"I  k +Bmin(E k,l )()-t k () 2 ,"I  k +Bmax(E k,l )()-t k () 2 {}I k -t k () 2Equation Native g_1133688461IQOFUUOle ObjInfoNP   nA D Bmin B(C l,n ,T)Equation Native !(_1133688465SFUUOle "ObjInfoRT#Equation Native $]_1133688468EeWFUUOle &ObjInfoVX'Equation Native (]_1133688471[FUUOle *ObjInfoZ\+A D Bmax B(C l,n ,T) 2LN  NEquation Native ,4_1133688475Ya_F+W+WOle -ObjInfo^`.Equation Native /(_1133688478cF+W+WOle 0ObjInfobd1  L N=16Equation Native 2(_1133688481]mgF+W+WOle 3ObjInfofh4Equation Native 51_1133688484kF+W+WOle 6ObjInfojl7Equation Native 84_1133688488iqoF+W+WOle 9ObjInfonp: L=256A D Bmin B(C l,n ,T)A D Bmax B(CEquation Native ;]_1133688491sF+W+WOle =ObjInfort>Equation Native ?]_11336884945wF+W+WOle AObjInfovxB l,n ,T)  n  lEquation Native C(_1133688497{F+W+WOle DObjInfoz|EEquation Native F(_1133688500yFPXPXOle GObjInfo~H= D Bmin B(I(x,y),T)= D Bmax B(I(x,y),T)Equation Native IY_1133688503FPXPXOle KObjInfoLEquation Native MY_1133688506}FPX8ZOle OObjInfoP 2N 2(N-1)C D Bmin B(IEquation Native Q._1133688509F8Z8ZOle RObjInfoSEquation Native T:_1133688512F8Z8ZOle UObjInfoVCE      !"#$%&'()*+,-./0123456789:;<=>?@ABcDFIGHJLKMONPTQRUWVX[YZ\]^`_adefghijklmnopqrstuvwxyz{|}~Equation Native W__1133688515F8Z8ZOle YObjInfoZ(x,y),T)DC D Bmax B(I(x,y),T)<D I(x,y):/tD^5ppAiP43;)8g ֺsna{emӲߕKl`gDp3Dd <  C A2Cr1Sbay6`!wCr1Sba8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?ȩ~x V8p( TY(t06C7 M-f86͑3v y@FT3 4ECuѡBs` 0UxTn@]*T=TjHӦU  zڻM*!q !|Gn7ٵ86yx=kb!{E,v\/$ wM2{Wb/r\tSy>fUc\_LӡL"CRwE9]Mw9r8*~>~Jqa2 C-9Y1o kO5BUzsp?{A (p($C|pz~Q MA|pB m%LjX(Uzcm&'p9g77[rk'P e2DH2ÉXTG]3HN!,]?@#uyLfӒMP66x\~BRaXVz*y+ΌTBJkB.4TTkh}쨬ОӅ_5P퐵YXh[E2 ICԤt͍_*w`)V{Ò pXw#y]g.ei=7~3;)W;:'1k<Nݝ߄%רjFgNDd <  C A2)ql$*}Hy6`!ql$*}Hy1 h !xcdd```d``baV fx ,F^equ02Ȓcjl@?H @fM\P4p~x V8p( TY(t36bd3ui`TF #n\ˍS.ʅa*v> ֭y@3EuBbj4UX>$֭~@S$@*g|xCo≜:T1 h !xTMo@])@Ԟq"^C{pCBNL]˻ISJH"?#7oPfwVr3ޛxv׈ZGW .OGLBmfYU>'4{vߠuN>Ye"dƋA', RIokzև_s dIsiRu\'fe21@Y EnN<2`YU].Mb1SU uC}Ճ˅/GOH#hT! ]ڋٵҠdFgyNb[Pki ZB9w\h4F8,Bᜲgˌ 3ouk{{ht{aiŲϪ}d?7>tppDd  <  C A2YP0٦JU# 5e 6`!-P0٦JU# PxڕM/Q{!m~X ȰLU}t!XJKji,;K[;¶ib؉ Όjj$3{y3g|H A:;0#s oGR;5viH#5P!J k'b %g=~^ѩ@niuZI+v eM?&*ER}W1Z>ZbVHusYöKwƉ'l9ɋ\}!1nNsUwRub-X%. 2܊HX%ZK܊8"p}$_W \Rql']aV}n\8"Xh&"<%WgrK:O|'.Wlէv.]"rǬCi.y ^C-Qr\̾ЌC޶ZT Y΃PxUKoT>ݚ&m]lM{lT"A.R1ccl4Q(Rl++; V~ s='W=|@[p[Mq,9%@ 8a%d>oEVY} @p'i8 \_ѧg]>G~Y ,s≠ 7d0rMngaPĵmr9M9#lj IbiD&VUW׸]'V;2"q&w;^݆w)] mZMQvj]3jAL}~ʵ9(fA\I|@Tƞ<#Yq9ϘOJoALIө e (3}ay 8y8i*wS62;)0D+R6n3E$cA= nЧ$s?CViPP # Jې9-.$ fî^DKE{QTv]1HR+Zm6ID5 5UU'q&Ctİ5慢(sf՞5Өdƻ Sh9|( MEfۯm,Mx]$4A<%D7*󡄝Jq.5n\kִJYD}5 ʤɆ:S,h ml5dӶa5h][)jd ^ V5Lfk&"Y1al;S? %/Oݝonp-/z[_ 7U{)/%h_Tbųu/tm#mDd <  C A2# e~cU߽jyN6`!# e~cU߽jy !xcdd``> @c01pY,P1=F (`d?3%$ lʛ! 27)?iZ@pT9 P\H+21,Pcln!Zpl!.A1pS~_B}`;?MS.t"# L.sw`V?pEUvVhPGVuƆyW^>{Aka$n0qj:@Y9XMn3db~PP9>q^٬ /"Jj\)5on R JdRN2-)Z7'{E9G=e4a#N#1 o/e(jMf^̊RaXTz.~2=IB rL:ƻ7Vʞ7fQXH_~ٔM53\cM晶 GM!c %*{y% QKwj$robӞÞǰ}. }l4@^NT+^Isg<<}?[`RKZ_s*z-Iö9?tU3Dd x<  C A2Ooxu /f K6`!wOoxu /f K8 *!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȹU@pT9 P\H+21,Pfln!ZplZW=v 9@2l$H:`G 4C +4ir QZQ"豇J4!J>":a[Y8nZH#Q{ݕZF28}(+ p>f<迹ߒX>(Ui):ɬhDeUXrX$e4EȎlaQD<ȝ{ UzQYJn,(EJ'U"'9\TfJ 1 ޾[TN$vŜb˦ ]vk8]8Dn9܌MjjgQֽy^rHd1΋Όywe Dba-Kv8MN0^O9=Ж7+fR}/M NDd p<  C A2*mV\reB26`!mV\reB2, !xcdd```d``baV fxt,F^equ02Ȓcjl@?H @fM\P~x V8p( TY(t36bd3V i`TF #nFpSM0 ["ĺUXM"֭?uT,Ju<EI|ZC'= U ׼, !xTMo@]iKK]T=K)ns A rubZM*TB*@(#Gn:#dg7.e7Iܶ5XIg}&Y6JR\[lVyOz_;л ̬N+ \zvIԹO -QO},k]U߱+4L&Xe˪uI@lBU+몀:G5/öO@ChT | ڍ\kP*RrHHdW Jxc6χY:ft\P  Ĕuej& d9?G''iLS@4i:b'+YGɇQJ}#r1'2 (~*{zc>(Tr BO8T~hVNYP H&{ltA(WaҞj00J~=gN[ (c=a)V\Z-D9:֨PkN=vwZsx#DzqkVBswF9pZiM9gv,S*p4ԭuA{[< ˫b_g}gd'rp+ Dd pD<  C A2/`5^Vw!6`!o/`5^V,X":xڕkQϹIMIHMR(v AP"bZm}՝ Ap %(n7QpSϝL w&sp绿|sFh%2Ch3nVF&I(o 0l<ϣR-rM,l g. \!|Ǯx])ν-/jAt]nfKg/SW;u6VcKF4xskL)4EG\eUu(.*]ثLieQo`L}P@1CQgGQB{uG5Ԙ&ZՓ(keS1W|=~[5)A @J^sP ^ޝFYm] T6s]ΐ48J_F#hJQ-X/P3߿RF3 ߯2ǫG_8n5QYLx5fsZ|*:~@uvB/fO%(M@*@rL욭?znO2f}7N#7SݪdiwzIJws,H]ۉ;9_mVxu{O4NfЂj K@[M>ȕ:(kj^ӣof[0ñv:4a)ǘ죗n֒^z;tH*BA$^ %9v܈i%G \Dvv // L*![|%-A)FCu!a>> l~* r]rAa(y59ݢޥ[ɍl.WқI׍iJ-x{i[)LUTన`U3eK)Luǣ̬x!1TMR08trLj2cLojkfnzU8rZ]LT^mno9[*5CE=6‘23dL67:oj΄k;v]vTM.z-BmX41m;݉A=tRylUVmŪ``2l r04&gjVe];r2`"5s LinW|tƺ)&;*eVθe-gK3yfڧA31,k9b+c ,\*7(2 tSl:Um: 0|c+%wb|O 'dq7Sx n/]C>KFcfGOP+ Dd D<  C A2N~qÚ(w*6`!oN~qÚ(.0:xڕOaǿﵠ=A͵#J[&81:TIqa07cb upprEQU`P罞5T}zsy~MN!&w:G\{6+ҳjա`f)<1u߾}rC]֭ ݳE/\2pqy^Ղ O'jbϏƌj۳Ք_v1=*徻mV]̵]T\Us3 @ocUAy9ͬlfXB{If/lSEcn5/Y֯,먨^KX]5'y.X[e"I\e뷇%/u΄;K>&c UY9Poy>)tD]{Y2`]b%KyC1xۅ&匉?fD{x36}&^If߰ooY֊g8kX/^1u^ߜbxċgY/>fs0:/^`kyċ+ O’ۆa&#Q=j[цYɳHT媉+# &RdGtDZQ47P]iVWp2mEv 6rk&W+ϋ&5>Cݖ#nP8N .0:sxWKoTδiҺψ"p)BXtDc'n7~7&p>gz=(o&7]}9za4WvXDwoYFmH{Tg\:u1_:W{@" 2CRr>2ów~x4qMaa1gj<ʿO57F5UL8 iN#oNO&2ұ(RM6=8nK3f8" C7JF)S,CO ., UqZͮ 8C1T7$oV#d%s-SblԫYBmȭUSok݈ը747QoRm4\tmq W}}r}T[$ʖdV,TX [^- P8^ZmPz/V+`2 rhv#X$A0(uLʚ2kMM=kl)$ɏ4#X4DŽ*q2S/\UiP1lYfWi~,Q׫l ;)R@IӨlS7ʳc67? @D7dx_{u*-xh&2qd܂ӯ+uӘn hYDd!+eLnȓ&""w;.Dv:P/&(7Yfn'r# Ǹn#EEsއj7@ȶ}PI((Ņd, f9߿5S>iV6q{0~=Bq,6ZP2I@OF 5@>=;PEOQЉo8kжA&5 ,*E 6^Ԝ3L -9(2Uin$FRsDTI,*a. &$\M]EKFbHvm$٬d*M 'WвTJaĊ3#լJf* Ձo_9;*+zs8E톆u!WMi65-Kn%AscK"j4tX:yH^imwYZύ m#()oSwoykNoz,OqDd ,<  C A2;6_88{086`!;6_88{@=@ :xڍJ@3iz*""VP͓  <owG-xE }^uv-j#df/_6  eQq5U캪DmӢWbc VɵV%teȯ~vovyz|A'0YO" 2d2!O vzr貘F.Lq%q*ÿs*aQ\EWZۆ0kМ~g; ui ZA5~j]7Z׺h]BmޖպԺPw~9Ԯ\'c7,LC)%XS~b+n=@ xTOO@."HE"̋6.[X]o$SNNw$f9_?Godi$D'^33䘅uӯf ǺB<8Ii>Aόin{(釤u-EeL­W'~T` y UU>y_ >RY)Y<3dq!95  bCU0Mۆ*q }x =xcdd``V`b``baV fx,F^equ02Ȓcjl@?H @fM\Pssl?z+8U`fҊL  B3d3vk v .@2BٍhˍS823&V\Xjg & 5#Z@S1€ [^bC֜XMU'֭ˉukTMb V Š@d]A[CVKbݪ 4-B TgCn5PsQRq>x xTMo@;))*qEJb7iK%ڊC/nkS{my4U¡BpEqSlyx޹_ "l~~ itlL1Ҍstݒ17=}"I>=xM3|{BU7`ڋ j<*zGqej?#WQ 724 ynCrpzaUY`R uF#Ձ̭hYumW,-R5kJBJR+2ּf5Y5z2T7nz9Ð_z\U$P C\ {& ύec!+O(y==3o:&j ~[[শ=OF S?)H4w\&쬧Ot{#BDd <  C A2!iṭ͜-p@qaC6`!iṭ͜-p@qa& ]!xcdd```d``baV fx,F^equ02Ȓcjl@?H @fM\P䈱qo?z+8U`fҊL a1 B2sSMui`TF#[u\n7:TN S v_c& ]!xTQkAK[kO*'DJZ\4h%gvM O>%]vfVVd*]+42M1QgEŬd㰳LGʣ}}9HncfyZa\'מ&]R&F{dA]T^΁EͽJ;qAleɔGL*1f,J'J*k=X+:K%P"ĩX1BHvސ~4/K{k ]1Aadn!4v ߋ鼢<&bORh5| &,A ^ag 4uH}#ށ\(i! %l_‡Y:ftBP)Ĕd\X@G 2XOԳ;?G9nNS@4i6b;/yQF}ɳS#r '2 z+{zc>(TBO8Tj~hVvMP H&KN6Mc'nW]F3ۜHkwbTK/|M~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOp6Dd <  C A2-%;55Ox( @K6`!z-%;55Ox( @8 h!xcdd`` @c0q3pY,P1=F (`d?3%do/PF I9 @Gb X@@ZaB7c?tӘ b`ӂc7Lt+FT3 >4"Csb$c?ɻM` h!WxTn@BȢ"UjOh)ZZ A NL]kw 8!p<G0vGB0&7kýsXHEr :I: q,Ote|/+mع*`vC[cO 6]経ߨR\TF0N\,fn~pMUVhCrf7NZ}eOCb|5zAc >$aS?rn{1_'4: G=vV6k)R4R%D7AWR>K͸Os%6}Q*R4Ɍh_"3N*r8D){˂q1C3F"j"1t {їUfe(M2/fE0,*=<TW9Ʌ\4Sj(~楓g75F/fӅ_6R팵0XKy"r@ag,I>>d\pK`n4S,C@(%[pZ;={ÖەgƋ#6kQc2NS,⚮;g 'SNo+wg1O0ކOPDd <  C A2JDYrMQ?0P6`!JDYrMQ?# !xcdd`` @c0q3(Y,P1=F (`d?3%$ lʛ! 27)?9ɵaXr.P5;VdebX4&C $378ش6J1pSne"֍qb*#T[|L*`h CA*g.Y'C7_jimnX3'# !xTn@8R# H-4$ cήݤB%$*$>|Gn'-|evI\`Mv޼y3^]5/9LmD}}K5V : bTeoygnSdiVaǃHʰ .܃)zpz!ɽkPn02U*b\#5YEXԉLR8U)훪ՖK;xm]ב YM @ab~lↇH.r#dй#ln5 \6*L|Űfc8X$ yh xN8O`RG,ߒX*\(i!:)AѼ1I ']70S}l`C6~TY{s Dz,&F~楢(Pp XDvlE8ŴBC:#ύʾDO0yzFtgMI:T}Adwn]F6t!%^rDɔ*{8l֞i('pJA}NőE-1x$0~zmlo2hN 5T'bsC5loFw/I:D_;{;-Y^-/,WΛsN/fDd <  C A2 "l&́Qv#vT6`!"l&́Qv#v( 8!xcdd`` @c0q3hY,P1=F (`d?3%$ lʛ! 27)?c㚹@pT9 P\H+21,Pfln!ZplZ5W#v >@2Y1n-Tnlbsk!rMef025 cAjgXCX${f 7OU<( 8!xTn@8P"UjOh8$RK'r A 3rubjZލ*TBB"p >#G>2$@0;3o̮g\F\'5rӺ\#(d&51I⾄kMJ?ahdisBOuaͫON)Mb#]RI5FZݯ9cokbu\)e2!E&ƕ~SY6HRgtVW_-?j^_7(O4= : RІ<@- rz*6v)`i܈.3e"r%n7R'\H7f+lT@fP߂PHnBN3ћ󌢾pGBVp{.8!dB!jZ{xrܕ%&M37TAgzԓ<=1$G3`|VL L@3-<+7USu_.@(fg7D`"u a)jŢ&}P9=cn40!{!.F0Ϳccm:ЃHQAвkuhfofgkeVfo|ХT-|>O?p+Ւzg9Y:ñ  isDd ,<  C A2̛w^FA"WY6`!̛w^FA"@=@ <xcdd``V`b``baV fxl,F^equ02Ȓcjl@?H @fM\Pں@pT9 P\H+21,Pglj& Zpli^ Mc22uOd7b t6\n6!73?/d,~ڋĺVh1atvB 4U # naeP&6dwVjĺ2nATlݺxr1) n j@[wVMhb(H:`wC)%XS~b+n=@ xTOO@."HE"̋6.[X]o$SNNw$f9_?Godi$D'^33䘅uӯf ǺB<8Ii>Aόin{(釤u-EeL­W'~T` y UU>y_ >RY)Y<3dq!95  bCU0Mۆ*q }x =xcdd``V`b``baV fx,F^equ02Ȓcjl@?H @fM\P䈱q~x V8p( TY(36C5mfM-f86.}1pS\ne'1Lq>.7BLed̐Xh2csb]4S0v!Lu*LX02 uTubXM$*n]D[tLukTCbݺXjM@˿CA*gB0Cn5PsQRq>x xTMo@;))*qEJb7iK%ڊC/nkS{my4U¡BpEqSlyx޹_ "l~~ itlL1Ҍstݒ17=}"I>=xM3|{BU7`ڋ j<*zGqej?#WQ 724 ynCrpzaUY`R uF#Ձ̭hYumW,-R5kJBJR+2ּf5Y5z2T7nz9Ð_z\U$P C\ {& ύec!+O(y==3o:&j ~[[শ=OF S?)H4w\&쬧Ot{#B1Dd <  C A28ɀ 9Si}>d6`!u8ɀ 9Si8 0xcdd`` @c0qUX,P1=F (`d?3%$ lʛ! 27)?9ucXr.P5;VdebX4&C $378شo\Mc22YLhf R?35CБ@^}>C4A` 0SxTn@]*T=TjHU  zڻM*!q |Gn7ٵ8ޙyfI5r쐽ڢA(d);.Rߦ³ڰ>9 /^:cy>fn'r# Ǩ_ґL""CRwE9զ r8e}$Md2Zb1Lj߿5S>i"W6q{0~=Bq,6ZP2I@OF =kCS9f~7 N|-ԕ2a^BR䛈 5~K9࿹ߒX>(Y%Fb/9G{ND5Ģf>`ADЅX|4 mT,dl"f%Vlb:°T $VfɥW\6Si,|QY75ġ/j4į jJ!k8ѶLׅEYcSb~и47d<,R,C%@(#U=향I Ц8^prrcx G רNozdN3Dd x<  C A2؋S4uښ.oh6`!w؋S4uښ.8 *!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȹ|Xr.P5;VdebX4&C $378ش\?Med ݌`#A*g5C +4ir QZQ"豇J4!J>":a[Y8nZH#Q{ݕZF28}(+ p>f<迹ߒX>(Ui):ɬhDeUXrX$e4EȎlaQD<ȝ{ UzQYJn,(EJ'U"'9\TfJ 1 ޾[TN$vŜb˦ ]vk8]8Dn9܌MjjgQֽy^rHd1΋Όywe Dba-Kv8MN0^O9=Ж7+fR}/M NDd <  C A2K/oDH l6`!K/oDH H Dxcdd``fb``baV fx<,F^equ02Ȓcjl@?H @fM\P庳@pT9 P\H+21,Pclj& Zplv .@2Bٍx!22pMef 4߱VXjc & Gȵ0S=JV!R*a*v'֭@SUu0nI[JzĺU821hp1[u( 4 T R?3v'CHJ'c8_ RH xTnFR;SnS1DHY>Ph/z+RbM. .Hp(/豷Oԙ]R Kμyfv܍@=8}ahjH1~M/F?(D|N; v -؆e4 yMUő~[*KJſ{0i1IsCӂ'p1 c&45д3թAϷ8L3jjbAԕ5V-JlSZE6*-͒{v}1z,@~x!Q$PL/1!k{rg8ͽFb\Dnΰ`^F;~. brЫF@ΘGGsKCSvK4Ikx'-#x=\5i|6g>< X|x$]léB4d̍TIx~f˓lQ\5G5TTGo&VY?w-7r i. ~ՔMeEPd.$p$ybLJ+&%Q\W]]i$p:J^.$ 7p~̣$1EG{0 a-<>+wdaw;Y] |.W [h($_\揦ZZ'=[mDN ! ?(T9euj=/w 2&諲Βw9ݒ_hlhYOe/C}avN~;y3ח-f/-i܎X[xxœozDd <  C A2o^VD ,"r6`!o^VD ,I%Dxcdd``fb``baV fx<,F^equ02Ȓcjl@?H @fM\Pz@pT9 P\H+21,Pclj& ZplZ(v .@2Bٍ!22pMefM 4߱VXjg & }Kȵ0S=JVqRܪa*v#֭@SUunE[JĺU821hp1>[mu 4T R?3tWC'{T5I%xTn@;)R# 8"1M8pzpCBnLuB¡B#7x2v\WB3|z+?A 0:ӀH9>Cβ6uiS(ley=\ 8g̓4hsڪ%ő>~BU{CR YwCr:p rI}hڎ ݛZědlC3jkr A֕5N+nuUmTag5KُHxSM "=SK޲1Od9'0^or躃b{\^pŞsݜY*4OHnƒRaF'r*u|J -9DAٛX eiֈ&"h_N'7qa#]#d3_lt @QW$9G?pP6)h7Y>b*U)/vEKe3OaUz~Q(ːBM TyȺ~XZFWeB2JH,Yy|9`R+{laXHZCQa8}Y]cNC0w,-t"svGڃ~vZqx\)lwPI0g^%};k-l m]GOm|׭U %3A_UwSG薼~.ꚦ.jp^^$3 k S?|;\_X8Hp“Dd <  C A2n-VSTyT $Cw6`!n-VSTyT $C !xcdd`` @c0q3HY,P1=F (`d?3%$ lʛ! 27)?9tXr.P5;VdebXi ZplFMc22Hr#TFׁrbʄݭ?uT>0H:`<[XC~J㊹>B}P\  !xTn1)- ԞDjII\C{pCB]ot^6Uďw(O=x$`$'oώ]A Vj\$$73{ײ9?aੱ}}9H6г4ɰIkON.O#kEj^5ïWs d1V n02)Wۿ#ƊlE:߉S*1*a 㙌EV;GsoOH#hmC Є7R/2|#?8rƍt'EIlFٝې4UhU zRAjJT1"'2* G)(>?f sSg5}FkmϚ~<5āA; +I_{|Kn-TXXr*ʛk=}"W#gYDd <  C A2i'n:ix6E'|6`!='n:ix6/  xxcdd``bd``baV fxx,F^equ02Ȓcjl@?H @fM\P@pT9 P\H+21,Pel4M-f86m//Q1pSne$֍_qb*#@S9t1bݪ4T&,XM$*n=N[u@~'֭@SU^x 2v6f<^o9 rG*+\d**O' %~ۭW*T&zY#YTX'7ꍗ6*FK)FnS;V?BJů1~P<9D?` o^ez#iAy0|UdQAT]HD H!0;>S) l((R{!(y3(whY/xe؆X88&" Ӷk(]л5gǙ̲cZXy\CzQ)֎vuxo9_5ԍ"oۏZ[xu駎~r1^ #ro!Dd 0<  C A2F5萅,f˳ 6`!F5萅,f˳\Sxcdd``eb``baV fxb,F^equ02Equation Native [__1133688518F8Z8ZOle ]ObjInfo^Equation Native _7_1133688521F8Z8ZOle `ObjInfoa  N E(x,y)Equation Native b(_1133688524F8Z[Ole cObjInfodEquation Native e7_1133688528F[[Ole fObjInfogEquation Native h7_1133688531F[[Ole iObjInfoj I(x,y)E I(x,y)=J(x,y)+E(x,y)0 DB(I(x,y),Equation Native ka_1133688534F[[Ole mObjInfonEquation Native oL_1133688537F[[Ole qObjInforT)<D D-1  DEquation Native s._1133688540F[[Ole tObjInfouEquation Native v(_1133688543tF[[Ole wObjInfox 2D-1  D  DEquation Native y4_1133688547F[[Ole zObjInfo{Equation Native |(_1133688550F@m]@m]Ole }ObjInfo~Equation Native (_1133688553F@m]@m]Ole ObjInfo D=224 BPrD Bmin B(I(x,y),T)<DD Bmax B(I(x,y),T)()Equation Native 4_1133688556F@m]@m]Ole ObjInfoEquation Native _1133688559F@m]@m]Ole ObjInfo t sub t addEquation Native :_1133688562F@m]@m]Ole ObjInfoEquation Native :_1133688566F@m]@m]Ole ObjInfoEquation Native :_1133688569lF@m]@m]Ole ObjInfo t mul= 2(N-1)t add +D(t sub +t mult )+(2D-1)t add ()BPrD Bmin B(I(x,y),T)<DD Bmax B(I(x,Equation Native Y_1133688572F@m]^Ole ObjInfoy),T)() BPrD Bmin B(I(x,y),T)<DD Bmax B(I(x,y),T)()Equation Native _1133688575hF^^Ole ObjInfoEquation Native F_1133688061F^^Ole ObjInfo* D[5,..,35] R DOh+'0(4D \h   Equation Native 41TablelvSummaryInformation(DocumentSummaryInformation8tȒcjl@?H @fM\Pz@pT9 P\H+21,Pclj& ZplZ '1pS\ne'1L5.7BLed̐h2cs]b]4S0 !L*E[Vy S-nB[VĺnG[V&M.b{bt5n}HljMu"@[7MDh% }3yCs*݃](\7xTn@^;iKK]qEJM#P!AT :1ז J 8 x&2vWg%3|f=:kU|"ӵ/#L\u'hWY[D!KEY$7柌Ã( Bn u,zUrģ+~QO{[Bh I(Kg&ɖ%rm뇔ckc-;PbLv b]򌪲ИCTUi V`WZtslA1uWR?.j;)О.l Zք$ ` M!XVĥ#/a&6そRh[2gVOG2̋J?#;C`\OɮSn@.Gq)ya4+Ž;!t4um&>Ix/㜭YE(x4ځ҃~B4Jƙ!^2`ѴRCYupW2vo#a Cpw8du,ν,Sv!$K(>R! =Un‘B{9UrJ>Q~2C[ C]ۊoxm0;$jfv hvZS4 Qh3ߵ BIŸ3io¯Yi7%Snj#0Qyl\Hʦʚ1HvS)Pׂjrh^wQeaIQO+WrYٚW'{u<p:CCWhC3Es&]6xTn@^;-*q@vLJP*h:1זIJ(*$p3ٵ?+9of7뙻AĺD%"9Ef뙌ȥ/$EE)?՝Hdyz=FXVhde$ڍCBjdhse+#]z ߰DsiJY4GNA^QUI]!D>eSvZe"gԕ|9H]ָ%6)9hJ#ߡ1E͝wFG}W vi?`g-eB!.0GЬNsG` ݁b:NFmC:fʼ$pc~F(7d^W1S¡؁LQSzJa*BIT(1Z#i8dpJuPW X0gy4`}Iϝɤ`kvJ&MGN|t?ffR7I.},5Si(\FQw C|q^04w#l㡊KtҚ3e;HdMi&3eygce{O{X(M) 'brQ:Aѐ Nj!6aݩS3l܅ڢ1,W3&Q;X8 / AK~Į4;Մ1Qg-s3pvK@*L=&bORsa5Fʂ 澐C 竨pvS3cj  JxR31Z7E{Y5G=䵂Q1e}9D,FAj^ϓI2$N[E0,*=L>RKNr#Ə0>kP^LM~Qe) qw!˦UaP@rC6q~|8܄i(CQ+vbsNۅ^[Ox@ش쎃cރV*zLCmcpwۥy8Qcn3q`oLx^{R~uqw_;_xW }<ͳyZ7&)u Ss4Dd <  C A2 ݛ/rŷL6`!x ݛ/rŷ8 R!xcdd`` @c0q3Y,P1=F (`d?3%$ lʛ! 27)?ȹcXr.P5;VdebX4&C $378ش\/?Med ݌`#A*g5C?6`eeY` R!VxTn@B"*'4J-iZ A9${:M*!qB;O@xI HoW^;PE$dr :I~Jqa ì;a2C7zшk 2@(6i/nuAA$6豉G'dQyl(5ibIh't[lmVW"HY X_Aր+.5> DP1DL Qj$EF2dñ(H¾G,'s10Cl`$>Y^6*92hZ*b üS)̓HqH5H.(䲙BCYuS=oկ0 y1.$)Mjg:2];c#dC% ,r8t_gCyyݮ=7^Zzmd&'_5w;gs*No1ϰ=YOpDd <  C A2HhJbbhR&/$6`!hJbbhR&/#xcdd``.fd``baV fx,F^equ02Ȓcjl@?H @fM\Pq%m?z+8U`fҊL  B3d3%4F0n*#+Э܈a*Ѝɸ1nTN SUXVM#֭)ĺUh0nU$֭@S%uk*n T3 {`CN И hp9#xTn@^U]TЈmzQ(^ *!^'ήuDj x}n<ev8TH7̎ww~HY%~BPT6U$/#FQz׊ia(g #Y.*T-r<9N yH*m>y_!*J%1aSD<`!?TY*h؇N%]kl?U 8-haG8)Q:x@v!hVb ]`h427nt[NH3&D$vlh5gQqxtCe|%5NTjJ]Lݔ":DJAp xJўCV+]y oAHY/Ct<D@rm_ˣq6" &S'4RWzY$x3~€i3̷Xye'7Dr^T w#hJP!~5g Iygce{O{X(M) 'brQ:Aѐ Nj!6aݩS3l܅ڢ1,W3&Q;X8 / AK~Į4;Մ1Qg-s3pvK@*L=&bORsa5Fʂ 澐C 竨pvS3cj  JxR31Z7E{Y5G=䵂Q1e}9D,FAj^ϓI2$N[E0,*=L>RKNr#Ə0>kP^LM~Qe) qw!˦UaP@rC6q~|8܄i(CQ+vbsNۅ^[Ox@ش쎃cރV*zLCmcpwۥy8Qcn3q`oLx^{R~uqw_;_xW }<ͳyZ7&)u SsDd L <  C A2Q$~:9k-/6`!%$~:9kwpxڕK@߻4FF""i.."tQPut)-b777/ .R_$9ܻ Sx3-nIM\tJL*L/urlU?]*׹S\eR;S6Jqi}npoDA~4ǀ sإ{0 Y"5ٮv6t%}/y1{9UZ7cAp8,0h7dAטF5X4qFi6Q`o(u:ftFW )Ĕ} bOC/Y$I<+'dkNJ9&F^°t'a(~dd% X2mPQ'ree/ߡ?(BWMqTqȺ~X̴. B0IlaS`O%p侀 a!ޕKã7bއ@mׁ|MJBSko=\Z{KӶ*ZӸ xLZuX]XWL7]ǚXjoFaVJ+k{q>肉hCϨbf\s{ͼmmǘ-Y\: ^2k9W2'o%,,*zV_գruxpY= $~KN~eDd ,<  C A2;S73sMNV6`!;S73sMNVO8O8xcdd``db``baV fx,F^equ02Ȓcjl@?H @fM\Pų@pT9 P\H+21,Pclj& ZplvMc22ʈ͍ȋˍSjMd ҭ׈u'T^bG[0Leĺh8n'֭:@Seu bZ 4UX V&M.ubMbtn$6\5SnE[mڠ47p>u<!>C3^i||ubsO8O xTn@^Q]TЈS,$qԕr9T ܐcSgml'M*VP!P$A'(kq+9ofYI |Nq:d}&66B sLEt D<]H1P&ɠ{,-REo JҧoK dU(W}\!eq,V892Y#]w@#I !%CڢSuM*G],q3lb+ZNrW.~~ N F# ǦU]+=rF5}σhF 9.[}GE84Y1um+ف4?ܐى^tS:͵oJ(*B䢃Qc4o(aeeb(;l:mC.;$4IV, cS. LS'IڙFWQ黋I *=RjxM JBݨfnW`2𼤂г74-*@AlPk2c&* )`*;qZm7 S^9_UM6 !xTn@^R*P8%RCVA9R9:1wݍ*TB ! 8ry2뤎@0&;3x&"5rFfe2>:\k:$wLĨ<ї Gyrl^zxB$'St`:~rρeýgZqA`,e*1SG5!+dZ?N*R>\gVW[udmigXz8K?`~BAg8P0Ø-m7A$9C\n{N O袠>(wskf\".RꕶL?4@j_n7rK](2i8fRS +ҁ^8 lFЂt2s0D̆9<;-^g OOTqzԊ3~€Y3lur ՗ύՊʾT yyzEtMI3f fW8A%!\~IG8K˫羭pXO\e2Dd <  C A2藵=ZfZj~6`!v藵=ZfZj8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?KtXr.P5;VdebX4&C $378ش.>)Mc22ʄnF 3y|3Czpuq` !!UxTn@BEEԞЈS*)ʵR NL]kw& 8!p< fNHZ{goW^]UXRve3p4Q/ѪZ[/ű~[AmY-70&0ڿ&9;,P)Ԩrk뽾RۧDFb[x@%dɲ6YBS5O3kߋ&Qбg8>ncY] -jLtd|m\p))4ol ҆O1'JdRS#ɛ7)'}EG=e$a#RQ:]g}e2Ѭr%C*b <)L7R7R2K! ,PPTcK'Gyfon_nXHo~U3Vl:"]){5͎)R> Y^q0:J~A#NUfmuXwrxqc`t8|`Ϝ'4UY)uI?@O;Dd <  C A20d1 I: -6`!d1 I:& ]!xcdd```d``baV fx,F^equ02Ȓcjl@?H @fM\Pc~x V8p( TY(t36bd361v F@2F\n7:TN S nuG[u.T ")n5 T3 m|C!aTvJ & ]!xTn@^)-F UjOhPKN)r 9:1w-&M*!!UE; Wܠo(Ο@&;3xw7&m_+d<6uČ$"7 U~ *Oyk5Z#I:Y8Ny@*:-Qj}>B4*Xq~QP&s0(Ę02Y[l'eTuFwٺ$ *6!2UUP]jDǐ_,|E0"Cs $С݄͕ yP4]L{;@c۟an)WEDH ;MhS۩,ʹżP"7LMQ"$y>c/4ppsH)fE d=ǐ%iNBy_b7i1RLeǜVRP؈2~Āq13T߼-3z#>$Vr‰qw1'E l:d06+.bU(1e54C$APbCQ G)EF=~d&=Y bwCS\x;15\ ~w]_saߚÛ Oe&f\PCtL>@ Ҋee3C'rF:pm4Dd <  C A2bm[}-6`!xbm[}-8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?KrXr.P5;VdebX4&C $378شK\۰?Medp ݌`#A*g4C*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<5Dd <  C A2A SFN!6`!yA SFN8 !!xcdd`` @c0q3pY,P1=F (`d?3%$ lʛ! 27)?șΥ@pT9 P\H+21,Pfln!ZplZ~b1pS\neBs7#HJu<r4.C*ǩAʱ.C` !!WxTn@BEEԞЈS*MRk!'{:M*!qB;' x ̮đ|x=+K Xn|}$~awyS>~&kEvV.~>e T\=W]C経ߨR\DF0NL,^4⚪C P>ʥߺw{K9[Clb| zl*Ǩ}|@5}|[^7 =q J8|T# +pc_ܧ࿹ߒH>Ŝ(I!:JhHu8pBc.f D'K FEb@vo/ifU(MR/fy0+=t"}#4#Zf eO5޼trW̿fV(ŬBz4mf0t]XDf:쌍g8 ˽a4J~CNum[Xwg:Gk& 49Q\awI ?+twJw'XGO<Dd H<  C A2 QkDxcV6`!QkDxc* $!xcdd`` @c0q3hY,P1=F (`d?3%$ lʛ! 27)?Qb~x V8p( TY(t36C7 M-f86m61pS\ne"֍̸1n-T^lbs^Bn`ddA -!ƂԪ3 XC!Ӓ 3"* $!xTn@8"UjOh8$UKǩz)r9:1w-M*UB  8r>7.<@uW y㝙o̸(Y&uB.QT6Gi e$ڢYfos{r=K [d_ryLHyN1j+~Y@Ȃ qAleɔGL1Ff kzQBfY#5:aU:k q>X:N{Q}< /AKAĮmnAĨ27حblSұ?f{1(3ؓv]h;snf,H=RJ wnpܛp8܈+K-٬r+݌i3IܔB k#6:#oJülBi6FYh/Nfz8(a]ş8"`'U>΋,g}$]lŮiI)e (+}bf+l!D4Yiձ7˱+޴[?aKa%L尨u@nL1D$C`j. 9)JgHaڕF Gk{O6WgIRzi=l'aðm8Ё$hh 8`Mö:$0QYkޢ0M\QcAᣣak8ƼaNf.6 XfUѭ78MUܨxXQ`pE=h¸c|.*.Ι6mMUP *A|cv ]pژ%6ײ`/{z̩;rMU~3{H[?}<5ڕ2k 5 X,/CyI=X O#?YKuǬt_f$?}ۓB_e)] "Dd |<  C A2dVm)_ j=/Y6`!dVm)_ j=/Y0xcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?Qb~x V8p( TY(16C7 in\\?Med(ƟȦ@Hfnj1q9pb*#TfҤ@;PÃl(H:`xFCc 4`~0xTn@'-TTqJ4JHʡ'9:1uv-M*! ?Px܀'!xI\|}~Y+e8ӵf#|')k,ntYeyI{/: ,Xxz< gjVŘr6ƯTw`(MЉf02ZNI2_FCHqBF=U<*/hb{H\]l0q3wfnwJP>֭|/ u {^wݹ͚pBn$65> DJSDLJa%E2dñ(H¾G,%st1OM/P4} 2Ͻ}er\Y{JF̋Y!#)(徖qNr(@!Ŕ _>kP&fQh̆v!gE)j;Tai0 qzT82Dŵ2p){ơ h8)ȣ)ki Eϕ*ic3?ACv:\ 6h/| [Nf [i)PF8)Wc6}Q6k7©..9׺źNTUNG[弃BrDd <  C A2_ҩs!ʝ|6`!_ҩs!ʝ|xcdd``> @c01Y,P1=F (`d?3%$ lʛ! 27)?)皿@pT9 P\H+21,Pclns`7LPt+#ԍ?Mb07.F `SSR@ F 3ysHCf&ZxTn@ǡ-TTNՈS iPrEBP=ȱ׉ky7iP \y p< 2vWg;;7w0 uX9 ̤n=|V?OҌUzV7OŢ<^O礝 w /=9d:0ن%ncpzY9Cͽi X02.t*c\LL ^yꗺ+bc n/<*ѾBa{8X\]ja&ըɲ0=:vS ``*~PgP9q[謙0 -ِF"Jxpb>y@s?%ҔP*Ӓw8֒E}C9/܉(X܁O,p0!TT:E>2=.Y{JGɳ* Rq-P#BΊ)c0VxfM"fƚAXSEyD OEM!HT\r Qb**RPީ @c01Y,P1=F (`d?3%$ lʛ! 27)?q:@pT9 P\H+21,PclnR8?Med(ƟȦ@Hfnj1qr#TF 9yv#PZu<3GC{.%OT.J0xTn@;V*j)4n("!(g^[޵*TB*$~^OCٵ?+9ٙm*pN kD|efoҳ}Oyf-8A?0p:%Q2CloGvqKύEBBF]gnv&48)}AD m9DI-Gs?%TQh\U}qRpHQ(ja%^)F\  S}* Ѐsb$=Ь; np#VIA(aqO%qIr(#"S+TGzܬJU %z^PX[eB#J@ڡg,vq9>drM+?e$(S4C@(y5[pR};9J45ێxߧet4,j\einFoSr Dd L"T<  C A2Z ӌlcC|A6 6`!. ӌlcC|AUKxڕ_HSalj[]P,M?]x7"%e$meeVZEBEtQHЅxS]]FADNٖ{`;;<|~#|7v9J{*h͕{1^Y/[mG/A9soe}wGF<{, S<_z5"ʛQ+y~T>]&B <񻀚9 %GiQ_OVc5NbcVsEG kemTZj䀰CYZ 9 !MTrP(B.JD JWNX7cz`A4Qf톒&PgquNvQSkh j5)8ѱ`uNq@5LAkfYNieEK(&V'(yڞኦ3@Y(Y^._EY6/k(ywhMi*ZyX9kTGWf&G| M5ܢ޳AUdEYM+z];Cgd b(x 7Ltj5ڢae͠iQ (kDTc(,ʚ1S c( P'(kva- >=Q.ѡ.qiIʛD2(> Zy.1t{(낦K,4];.et tjv:gOZϤK{>0^Y)/7Z?i藃/zJ#l R#0 _eLmYN]i"PlxFjL,_1 FqK^&X76Sn"4`kPz :umy8ˌb2Wc-WKv7fz'IX)fJ&/L,<۪ l6*eʭ̭:gmǷYUaqB8%2/ AbfrӬ2bhMj0򠃹K~;7ERnq} I=R߲JE`+-i pwbxZ~'-p4=*?0xB89;8fZ1/ ɽmqx oY; H4˾CWV13 -bZ.N22]AAO@BٿI:|261WPg u~S}>fLGiB(꼭λ?P6Շ[5xe^Uu>V9L E`SC-L*#;BғD'؝Rgq&^ݝ Y_ȿ ̏ńl[#3 D7PY5ΦN& BM:Dd T<  C A2e~tR=6gᐖA6`!9~tR=6gᐖv (g%Kxڕ;KA浛y(G BAF)l XEXDRHJF \l`J νΙ˝ 2 R#3*"[&ɍ"ˉ1/Bx!8dp4I?gBq^Oeu%O*Ń2UW9isnTi:t_>jj$ u\%ՄR$[O#1vU*hlJxX0ֈjPi014T5 ` jJZcE5fYu e 'L֦5B(kY7QZ nA¬auUwnp58+.1ǨKD`حKUܼ CL +Y:.q&u\eDeMD[%PQxT_D eUe邫}q 7fC\b=+SQRj|*(g%KxVMFϮ)L :>{baai[hZ4lYJCzKhoͩnߙ%i}g>k7ktCCЪ#|4<{ F}im:k^g6nUvЭI!B:wQ.\JX~p)>@72B4 i0w8nnØP&R$vMA j!NCIt(X@-޼:(MY-i\b[&f }wYm}.;۳Wx۬}u}X߿C=~acvƹ;#$+ 9 V` 1pDΧ'S@j@ΰc .{u"si=UbQ?Mr|& bR_5aMxfSgăTU~R(Mh<+-Ɍ@\G91칠Gld ]y㬯P:-`5I6s#,eOZŒxE-}<&+3 C:f9V~śVVv;cG8 z)0]0vbL>a8'R6Y Rv%Qq>g@Yg,t[p~j0:`Nt 7  >03XӰms7 LԱz֚3L4WTiXP@Dh+1$G~=CٯaYbtoM8Nts7*nz&2Făal0n05~smM[n`sS꺁it\|dGwp v屼Zp;HuMBP^R{w|~(ORc1<ݗO_'rd$P<>G!Dd <  C A2N:Ckĺ*6`!":Ckĺ<  xxcdd``> @c02؀Y,P1=F (`d?3%$ lʛ! 27)?fXr.P5;VdebXcdpx! 6 r5cL 4U3$< OcfŠw)6\m,P9Z@Qf0A9CH^z_Q:TVJ<  xxTn@]; -Q@TNh)FiIJ EP!T*Y[^'M*J"(_ 7.ٵ8yzwψh!"FfmGz8I8(0'})4ޠuA>D!y2M6n8_l Z$"ޕKw*cJT(~:.f8 <k N sJ`5zC1:ה}BR\"l^~TS/Ш1hcKe D>" @3"v ҉=0-.+ƑŸoZ 0 [F shr#$z 11o`1F  0i}`L^}u˸m u5gOY?@<̴xy/I7㔭5pjhlZuLOwq'",8bY1`9޾MO3[zS1>륉\ϋ⸨bUm7iU& H $ @lEM!i,3en"M (AvJH4Z:e<σ$k_v}F)uhgL<:ZcUkݨahCL >נCQZ{xc=ZjAQwWzTOiE>e̔^$TqDd @<  C A2k,{fb6`!k,{f xxcdd``> @c02Y,P1=F (`d?3%$ lʛ! 27)?Y)%@pT9 P\H+21,Peln!ZplZW v A <А@7Jr#TF. l0@ F 3ycECܱ F$Ͻ{ x|xTn@'҅VqH4A9R8"^'ήݸB%$NHo(_9I\ J̼yfq`6*kpM kFud; KXgde+(w|<1@c٢cvӹruТF?IS.t"#'_aV?sEU6VhRsBJuW`q_yG= "_#qnSǨ&|H`f趻%*# +ߋMAzBŞuu>k"HY2D^Ɍ 65> DJSBLJq%E2/ñ(H¾G,$s1#`$>^i$dser^%M̋YQ* JOн(徖y.r*@!͔ʫ)l}mvQ֛М3 B/Rt掵0XKy0rH0wlQSsʍ$*\0O1 AW/jݩ@T npTitolo"$$&#$./a$ CJ0KHaJ0BO2B np References$ & Fa$aJLOL np IndexTerms$`a$5CJ\aJB @RB np Pi di pagina  !<Ob< npText$d1$`a$FOrF npFigure Caption$a$CJaJ`O` npFigure Caption CarattereCJ_HaJmH sH tH DOD np Table Title$a$ :CJaJ@O@ npReference Head  & F@@@ np Intestazione  !DOD npEquation$ d1$a$VU@V npCollegamento ipertestuale >*B*phNV@N npCollegamento visitato >*B* phZC@Z npRientro corpo del testov^v`>O> npIndent hd7$8$^haJ6O6 npPP5!$d7$8$a$aJNO"N np HangBullet"@(d7$8$^@`(aJ4O24 npPP#$7$8$a$aJ8OB8 npPP24$$7$8$a$aJbR@Rb npRientro corpo del testo 2%$7$8$`a$HZ@bH np Testo normale&7$8$ OJQJ^JHOrH npHang"'$d7$8$^`a$aJ4O4 npai1( x 7$8$aJ`Oab` npStile Text + Prima riga: 0 cm )`BOB np Reference*$d1$H$a$aJ @UV'). P 2!!+""#$%N%&'*"***+.3)7W8==G>>?t@@@APAQA`ABD#DRDoDFGmJLNnPP?UWY [[A\^\t]%^B^^_cj>t@@@APAQA`ABD#DRDoDFGmJLNnPP?UW [[A\^\t]%^B^^_jŵտadefghjklmnopqrstuvwxz{|}+*tH_~biy~cEk}{136OQt2KM ? X Z g !!2!K!M!a!z!|!!!!!!!!"" "&"("+"D"F""""""""###0#2#W#p#r####$$$0$I$K$$$$$$$$$$%%%0%I%K%N%g%i%%%%%%%&'&)&&&&&' ' '9';'V'o'q'''''''((!(4(M(O(i(((((((((()) )#)%)))))))** *(*A*C*\*u*w*******T.m.o.......5/N/P/b/{/}////[0t0v000001 111122222233344!4+4D4F4444444555,5E5G5555888H9a9c9:::======G>`>b>g>>>>>>>>>8?Q?S?c?|?~????????@@@5@7@9@R@T@V@o@q@x@@@@@@@@@@@@@AA3ALANABBB C$C&C.CGCICOChCjCDD!D4DMDODRDkDmD=EVEXEE FFUFnFpF7OPOROUUU V$V&VA\Z\\\t\\\]]]]]]^^^%^>^@^^^^^^^^__2fKfMfUfnfpfffffffrjjjjjkk-k/kWY*CEXqs24:SUʎ̎!#>WYfÏɏ -FHNgi2KMёӑّG`bϒђג6OQjΓ F_ad}ʔ̔ה:SUWpru .06OQǖ24Kdflԗ֗$=?Unpv˜И*,2KMh֙*,B[]x(*[tvśǛ+DF$=?B[]c|~$=? 9;%>@8kԷXX::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::XXX !PointTmp _Hlt6027018@k k ktĖ k k k\ k kJ%kK%k6k 6kT 6k 6k 6k!6 6 6IN>66=HH     66NND ;HH$$PP  B*urn:schemas-microsoft-com:office:smarttagscountry-region8*urn:schemas-microsoft-com:office:smarttagsCity9 *urn:schemas-microsoft-com:office:smarttagsState9*urn:schemas-microsoft-com:office:smarttagsplaceC *urn:schemas-microsoft-com:office:smarttagsmetricconverter 1. In13, in ProductID    L Q f k  b l nx}4 > ##%$/$c$l$& &''Q(Z(#+,+++,,,,u--q.{../2)223V4_4v44 6666;8@8B8F8999999;<)<2<wAABBC CCCDDNNibnbuppssFwPwwwyyotƁȁ΁˅Ѕ '0eo19itAK=FʢӢ_h%agMRW["/9<pv{ڭ&. °Ȱʰ̰Ͱհڰ^h %)Xa=GU^GN2?$+Ӹ> G -24 ? !!H"J"##%$0$k%m%'''())"*'*..[/b/33==c>f>@@t@v@@@C C$D)DOFUF{ZZe!effhjrjr#rhsjss tQwTwxxyy4z8zz{j|p| }} 7> **2=G<F1:26 $ә֙)+z}Ӹ333333333333333333333333333333333333333333333333333333333333Ӹ DW. KX$N6U l y#/d}:]eE@xzQzR$|Kc@xzm-m6D;)mŽ:'Fz@.@.@)@0^`0)@P0^P`0()@ 0^ `0()@ 0^ `0()@0^`0()@0^`0()h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hHh^`OJQJo(hHhll^l`OJQJ^Jo(hHoh< < ^< `OJQJo(hHh  ^ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh||^|`OJQJo(hHhLL^L`OJQJ^Jo(hHoh^`OJQJo(hHh ^`o(hH)hll^l`OJQJo(hHh< < ^< `OJQJo(hHh  ^ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh||^|`OJQJo(hHhLL^L`OJQJ^Jo(hHoh^`OJQJo(hH@hh^h`.hh^h`[]h^`OJQJo(hHhll^l`OJQJ^Jo(hHoh< < ^< `OJQJo(hHh  ^ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh||^|`OJQJo(hHhLL^L`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHhll^l`OJQJ^Jo(hHoh< < ^< `OJQJo(hHh  ^ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh||^|`OJQJo(hHhLL^L`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHhll^l`OJQJ^Jo(hHoh< < ^< `OJQJo(hHh  ^ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh||^|`OJQJo(hHhLL^L`OJQJ^Jo(hHoh^`OJQJo(hHh  ^ `OJQJo(hHh^`OJQJo(hHoh^`OJQJo(hHh| | ^| `OJQJo(hHhLL^L`OJQJo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJo(hHoh^`OJQJo(hHhh^h`[]h^`OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHhl l ^l `OJQJo(hHh<<^<`OJQJ^Jo(hHoh  ^ `OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHoh||^|`OJQJo(hH d}:;)mm-mKQzR'Fz#/$|KceEl N               WGnptLUi*?mJZp|rStQӸ޸)j9]09]0Y1 6ttEh@dO,"@UnknownGz Times New Roman5Symbol3& z ArialG  MS Mincho-3 fg?5 z Courier New;Wingdings"q36NN!242qHX)?np2TCompression of 3D Scientific Data for Archival, Distribution and Fast ClassificationFrancesco RizzoFrancesco Rizzo<         CompObjurd MSWordDocWord.Document.89q