{
 "metadata": {
  "name": "",
  "signature": "sha256:845f7a2188fc8f3e817af849f6088bdad0751e83b4ab047024569222e7e146b1"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*Back to the main [index](../index.ipynb)*"
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Multivariate pattern analysis with PyMVPA"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*Part of the introductory series [Python for Vision Researchers](http://gestaltrevision.be/wiki/python/python) brought to you by the [GestaltReVision](http://gestaltrevision.be) group (KU Leuven, Belgium).*\n",
      "\n",
      "In this part we discuss how to conduct multivariate pattern analysis on your fMRI data using PyMVPA, wrapped by the `psychopy_ext.fmri` module.\n",
      "\n",
      "\n",
      "**Author:** [Jonas Kubilius](http://klab.lt)  \n",
      "**Year:** 2014  \n",
      "**Copyright:** Public Domain as in [CC0](https://creativecommons.org/publicdomain/zero/1.0/)"
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Contents"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "- [Multivariate pattern analysis with PyMVPA](#Multivariate-pattern-analysis-with-PyMVPA)\n",
      "- [Contents](#Contents)\n",
      "- [Why use multivariate pattern analysis?](#Why-use-multivariate-pattern-analysis?)\n",
      "  - [Conventional (univariate) fMRI analysis](#Conventional-%28univariate%29-fMRI-analysis)\n",
      "  - [Multivariate pattern analysis (MVPA)](#Multivariate-pattern-analysis-%28MVPA%29)\n",
      "  - [Why does it work?](#Why-does-it-work?)\n",
      "- [Explore the signal](#Explore-the-signal)\n",
      "  - [Setup](#Setup)\n",
      "  - [Data preprocessing](#Data-preprocessing)\n",
      "  - [Statistical model specification](#Statistical-model-specification)\n",
      "  - [Region of interest (ROI) definition](#Region-of-interest-%28ROI%29-definition)\n",
      "  - [Detrending](#Detrending)\n",
      "  - [Timecourse of the signal](#Timecourse-of-the-signal)\n",
      "- [Correlational analyses](#Correlational-analyses)\n",
      "  - [Normalization and averaging](#Normalization-and-averaging)\n",
      "  - [Training and testing sets](#Training-and-testing-sets)\n",
      "  - [Cross-validation](#Cross-validation)\n",
      "- [Support vector machines](#Support-vector-machines)\n",
      "  - [Introduction](#Introduction)\n",
      "  - [Setup](#Setup)\n",
      "  - [Data separation problem](#Data-separation-problem)\n",
      "  - [Is that the *true* separation boundary?](#Is-that-the-true-separation-boundary?)\n",
      "  - [SVM on our data](#SVM-on-our-data)\n",
      "  - [Training and testing sets](#Training-and-testing-sets)\n",
      "  - [Cross-validation](#Cross-validation)\n",
      "  - [Nonlinear classifiers](#Nonlinear-classifiers)"
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Why use multivariate pattern analysis?"
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Conventional (univariate) fMRI analysis"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "So you all know how a typical functional magnetic resonance imaging (fMRI) analysis goes like. You first collect some data, such as brain responses to faces, houses, objects, and bodies (as in Kanwisher et al., 1997), then use one half of the data to define a region of interest (ROI) using a particular contrast -- care about faces? good, take faces>objects -- then use the other half to check what it responds mostly to. (Why split data in half like this? Avoiding the double-dip issue here, see [Kriegeskorte et al. (2009)](http://dx.doi.org/10.1038/nn.2303) for details.) If you chose the biggest blob in the fusiform gyrus as your ROI, you'll see that it responds twice as much to faces than to other stimuli. Congrats, you've got the fusiform face area (FFA) in your hands."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Multivariate pattern analysis (MVPA)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "OK, so you've got the FFA and now you want to understand the mechanisms governing responses in it. Does it respond only to full images of faces or perhaps two eyes and a nose will suffice to activate it? Or does it encode identity, gender, or race information? Great questions but it quickly becomes non-trivial to carry out such studies in practice. FFA responds robustly to faces as compared to non-faces, but differences in response to a male face versus a female face, if any, are likely to be tiny.\n",
      "\n",
      "Moreover, it makes little sense for FFA to respond more to one gender than another. It is more likely that such differences are encoded by differences in activation of units within the FFA. For example, one population of neurons in the FFA might detect the presence of a male face, while another would encode the presence of a female face. On top of that, there are probably many units that respond to certain face features or their combinations irrespective to gender.\n",
      "\n",
      "So if you carry out a conventional analysis, you will likely fail to find a significant difference even if it is actually encoded in the FFA. But notice that in a conventional analysis you are also throwing out all information conveyed in the *pattern of response* because you are averaging signal across all voxels within an ROI. "
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Why does it work?"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "What exactly are we \"decoding\" with MVPA? Remember, voxels are not neurons -- there are hundreds of thousands of neurons within a voxel, all with their unique tuning properties. So rather than observing patterns of neuronal activation, all we have in each voxel is averages of these activations. Shouldn't we therefore get into the same problem as with univariate analyses?\n",
      "\n",
      "There are two major theories here: hyperacuity and coarse scale biases (see [Carlson, 2014](http://dx.doi.org/10.1523/JNEUROSCI.0548-14.2014)).\n",
      "\n",
      "Hyperacuity was proposed as a potential mechanism behind the success of MVPA. The idea is that in each voxel the distribution of neuronal preference is not uniform (see [Boynton, 2005](http://dx.doi.org/10.1038/nn0505-541) for a nice illustration). When you take their average -- as happens in any given voxel -- some features will have a slightly higher response (because just by chance there are slightly more units preferring that feature). And that is what you end up picking up in MVPA. Given its origin in neural preferences, i.e., very small scale features, this theory is known as hyperacuity.\n",
      "\n",
      "But we don't need these very fine scale anisotropies. An equally plausible idea is that features are grouping together into maps of a scale of several millimeters. So MVPA is simply picking up these large clusters of information.\n",
      "\n",
      "To this day, there is a great deal of debate regarding the nature of signals that we are decoding. Moreover, can we really say we are decoding stimulus representations, or, similarly, can we claim that a particular area does not encode a particular feature if we fail to decode it? Of course not. When it works, MVPA informs us about the information available in that area but whether it represents anything or is merely epiphenomenal we cannot tell. Neurophychological or brain stimulation studies (TMS, optogenetics) might help to answer these questions. And when MVPA fails, then we ust don't know anything just like with any other null result studies."
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Explore the signal"
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Setup"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In this tutorial we will use the `fmri` module from the [`psychopy_ext`](http://psychopy_ext.klab.lt) package to do all the work. The `fmri` module provides a high-level interface to the [PyMVPA](http://www.pymvpa.org) package that was build specifically for MVPA analyses.\n",
      "\n",
      "In the first step, we will generate synthetic fMRI data so that you don't have to download huge fMRI files. You will need approximately 341 MB space on your hard drive to store synthetized data."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numpy as np\n",
      "np.random.seed(3)\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "fmri.GenHRF(PATHS).gen_test_data()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "FreeType import Failed: expected string or Unicode object, NoneType found\n",
        "Generating fMRI data..."
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "2 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "3 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "4 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "5 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "6 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "7 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "8\n"
       ]
      }
     ],
     "prompt_number": 1
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Data preprocessing"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The first step is to preprocess your data. In [SPM](http://www.fil.ion.ucl.ac.uk/spm/), we typicaly do the following steps:\n",
      "\n",
      "1. Correct slice timing: correct differences in image acquisition time between slices because data during each TR is acquired sequentially from slices. Produces files with a prefix 'a'.\n",
      "2. Realign images to the mean image because a participant is moving during scanning. Produces files with a prefix 'r'.\n",
      "3. Estimate parameters of functional image coregistration to the anatomical volume so that we can compare the two and reconstruct 3D models of the brain.\n",
      "4. Segment anatomical image into grey and white and cerebro-spinal fluid.\n",
      "5. Normalize functional images to the MNI template using segmentation parameters. Produces files with a prefix 'w'.\n",
      "6. Normalize anatomical image to the MNI template using segmentation parameters. Produces a file with a prefix 'w'.\n",
      "7. Smooth functional images with a Gaussian kernel with full-width at half maximum (FWHM) being twice the voxel size. Produces files with a prefix 's'."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Statistical model specification"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Now we are ready to analyze data. We need to specify a statistical model for comparing brain responses to the conditions in the experiment. In SPM, it is the matter of providing the order of conditions in the experiment (produces beta-values) and specifying which conditions should be contrasted (produces t-values). It does so by fitting a hemodynamic response function (HRF) to the data. In general, beta-values *tend* to be more reliable but there's no guarrantee. These contrasts are useful for defining regions of interest (ROIs). \n",
      "\n",
      "For non-localizer data analysis this step is not required but highly recommended as it produces better estimates of brain responses. However, in the rest of this course we will only use raw data, i.e., data as it comes out right after preprocessing."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Region of interest (ROI) definition"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "ROIs can be defined in two ways at least:\n",
      "\n",
      "1. By overlaying constrasts from the statistical model on top of your functional or anatomical data, and selecting the relevant voxels (typically, voxels that respond to one condition more than another above a certain threshold, like p < .0001 and requiring contiguous regions).\n",
      "\n",
      "2. If the ROIs are small and easy to confuse with other ROIs (e.g., V1), 3D reconstruction of the brain is necessary. You can do so using [Caret](http://brainvis.wustl.edu/wiki/index.php/Caret:About), [FreeSurfer](http://surfer.nmr.mgh.harvard.edu/) or many other packages. Once you have the 3D surface, it is much easier to define precisely the necessary ROIs."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'}, rois='rh_LO')\n",
      "mvpa.plot_roi(roi='rh_LO.nii')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAADrCAYAAABkQOxeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsvXmcZldV7v+tubqqq7s6nU4nnc7cySbzCCEDJEQSgopR\n5KJeROCioKJBBdEE1B+II8pPVIiKSrzijQJqIJchDAYyCKQTQsi4M3Q6SWfoIT1X11x1/zjnOWe/\n67yna+jqIdXr+Xzq89Z73nP2dPbZZ+9nP2utlsnJSRwOh8PhcDjmE1r3dwEcDofD4XA45ho+wXE4\nHA6HwzHv4BMch8PhcDgc8w4+wXE4HA6HwzHv4BMch8PhcDgc8w4+wXE4HA6HwzHv4BMch8PhcMxb\nhBAuDSHcNwfp3BNCWDQXZTLpfjKEcM4U5xwbQtgxw3SPCyF8bopzzgwhPGuOvTGEcG/+940QwqqZ\n5HsgwSc4DofD4XBMgRjj2THG7Xsh6VcDLXsh3WOA0OyHEEJbCOHXgZuBhcnx5cB1wA/HGM8E/gP4\n671Qtn2C9v1dAIfD4XA49jIWhhBuAF4CdAO/EGO8PYSwGPg4cCYwCXwZuDbGOB5CGAZuzH97E7Aa\nWAb8BvDDebotwBnA22OMnwoh/A7w08AY8AjwKzHG9SGEbwL/DVwEHA3cBrwF+DCwAvh0COEtZKTD\nnwBdwBHA12KMP7+7ioUQrgWuyuvVC7wX+ALw98CKEMKXY4yvNZedA5wGvCGvMwB5WQ/L698OHAts\nmqpxD1Q4g+NwOByO+Y6VwEdjjGcDfwv8f/nxvwQ2xhhPB84jm8y8N/+tA/hCjPElMca782OTMcb3\n52zO2WQMyE3A9SGEtwFXAufl7Mf9wPVJGY6PMV4CnA5cBrwyxvh+4FngTTHGO4Grgd+JMb4cOBX4\nsRDC2XWVCiEck6R1JvAB4EMxxgng7cDjTSY3xBhXxxjfDqxr8tt4COG8/Lef50XM4PgEx+FwOBzz\nHY/HGFfn/98LHJb/fyX5CzzGOAL8DZBOCG6rSzCEcDXZ5OKnY4yTeVr/GGMczE/5S+CHQggdZOzQ\nTXk+O4HHgEOaJPsW4JAQwjXAJ4Aeki0kixjjk8BbgTeHEP4IeCcZiwN7sO0VY7wrxng4GRv1xZzp\netHBJzgOh8PhmO8YTf6fpHz5t9I4EWijUbqxs1liIYT/Qca2/EgyobFpteZp6dhg8tskzScgt5NN\nlB4CPkjGotROVHJx8rfJJkE3k21vzfq9HkI4IoTwGn2PMd4MbAeOn22a+xM+wXE4HA7HwYqbgXcB\nhBC6gHcAX9vdBSGES4CPAT8aY9xg0npbCKEn/3418K2cGYL6icoY0BlCWAKcC/x2jPFGsm21VWST\nrjq8AlgdY/wLMrbpJ5Lzx8i22WaCBcC/hhBOAAghvCpP76EZpnNAwEXGDofD4ZjvmKz5fjXwV7kZ\neSeZ4PYPdnNNC5mGZ5xMGKzJxOfJGJejgDtDCK3Ao2Ti5LoyCDcC/0amd/kj4Hu56faDwJfIJjlr\naq6/AfjJEML9wAvAvwI/E0LoJdMAjYcQvpNreupQpBtjXBNCeDvw7yGESWAL8LoY49Burj9g0TI5\nWdfmDofD4XA4HC9O+BaVw+FwOByOeQef4DgcDofD4Zh38AmOw+FwOByOeQcXGTscDodj1ujo6JhT\nIWdLS0vDp3Siu9OL6ty677rWHrfQeW1tmXZ4YmICgI6OzBips7MTgO7ubgC6uroA2LZtW8Pxvr4+\nABYsWNCQ3ubNmwF47rnnGq4HOOKIIwB405ve1JDnhg2ZodbatWsBeOCBBwBobc34if7+/obzBgcz\na/Tx8fGGOui4vutTbaLvgm3/4rqXZZ9tq5sbd+3uPqnMddfUXWvLYjE6Otr0xvoEx+FwOBz7DFO9\nrITZTGzsC9S+tOuu1wTEpqdJhjA2NgbA6GjmVueQQzJffe3t2av08ssvB8qJjnDOOVkszWXLlgFw\nww03APCZz3ymOOfSSy8F4MILLwTgkUceAeDQQw8F4PDDDwfKCY0mLKtWZbEwNaH5xje+AcDjjz8O\nVCd3mnTp/JGRzIpdkzedrzraCedcTECF6Ro5TTXxqYNvUTkcDofD4Zh3cAbH4XA4HHsddsU/1ap+\nd6vzui2pqbZbBDE2YjH0XczLjh07AOjpyXz2HXnkkQAcc8wxABx/fObY97DDsogPjz32GACLF2cR\nDXbt2tVw/aJFiwA4++yzG44///zzRZm2bt0KwF133QXAs88+C8Dy5cuBklERg6N20vaY6qCyipnZ\nsmVLQ13VNpa50XexVGKldL7YsQka21bHLZMzly5o6vrOVHk4g+NwOBwOh2PewRkch8PhcBwwqGNu\nmglU7TGt6Au2oYbJ0XGJfK0W57TTTgPgpS99KVBqZ/S7mBx9f+qpp4CS+RkeHm74/cknnwRKFkb6\nmIGBgaJsYoOeeeYZADZu3AiUOiCVWUyP0pYWR3mLqZFoWRoeMTT6XedL8Kw2s1oce10dLJMzlf5p\nJrBMzXRF487gOBwOh8PhmHdwBsfhcDgcc4apVtWCXZXXWec0S69OayOITRBDI7ZD6O3tBUr247jj\njgPgggsuAErLJLEZYlwWLlwIwKZNm4CqVZXy0adYmCVLlgClXkaMzsqVK4syqayp6TiUOh2VxZqs\nS2NjmRa1n/JWXaUzOvXUUwF4+umnAVi9ejVQsk8qh9JRvoNtWTp1Ohh7H3d3Tt3vcwVncBwOh8Ph\ncMw7eLBNh8PhcMwa03X0NxUzM9WqvpkGxzriE+sgCyDL3IhxEftxxhlnAPDGN74RgBNPPBEodTD3\n338/ADFGoGRJZEUlayira5G2Rj5nlK7YGB1P2RoxKcLQ0FDDp/JWnmov5anzVEcxNdLgqM7bt29v\n+F0skrQ9N910EwDr1q0DStZK7T9yTlaO7u93N9TdMkjp/VKZrEWWPvX7bFHn6M8ZHIfD4XA4HPMO\nrsFxOBwOx17HdL3R1vk6Sa/X/2JA5GdG10hvIlZDGhl5A/6Zn/kZAC677LKG4zaUwgsvvACUvmcE\npWt9y4gFEVsiT8didFQuhXJIWQ4xJSqrWA3VTb51xMBIK2MZE/spiLFR21lm6GUvexkAK1asAODT\nn/40AI8++mhDmVW+3rZMxyQ2SnW05WpWFsGGixCm2llyPzgOh8PhcDgOWjiD43A4HI59hrrVuvVh\nY89vZk1lLY6kRxH7IW3OWWedBcDb3/52AC655BKg9BYsxkbnW6++Yhqk7RETIzZFv1tfMYonZT0l\nWzYlrYsYEuVldUX6lCWXZbxsO4n5EVQWaXnEuKhO0iW9/vWvB+CLX/wiUFqSqa7W27ONv6V00zrV\nsXJ1dahjaKarHXYGx+FwOBwOx7yDW1E5HA6HY9aYrhWVYFfndd6I7Wo+PU+sho6JLRDjYrU573zn\nOwE4/fTTAdiwYQMA69evB0pWQhZH0uLod+lLVLadO3cCJZsiiAWRnka/2++KWSUdDJSeh8UeyZLL\ntocYGDE48oNjPQjbSOiC9ERiXNRW0t7I4kt1Vj6yKPvS5i8BMPytrM1t3C19pj6KVE+1W13ssNn6\nx3ErKofD4XA4HAcNXIPjcDgcjr2OOm1NnZdb68OmmSWOtZLStWJipLU59thjgdLXjJgZMT9iVqTJ\nsX50lH4drLdfWRypHGJJrB+YNF2xRtaPjfWSLIgVEoOitqjTtShPnWe9QCt/6YbE4KgcimreM5Qx\nNEOTQw3lsn5wUk2O6m+tz+pihgnTtZaqgzM4DofD4XA45h2cwXE4HA7HXsNUq/A6j8ZWo5MyOGIL\nBPmdkY7lzW9+MwAnn3wyUHrrlV8bsSJiKfRdDIz0KCqL8pOGROepTGIkli5d2lAOsSvKR0yGYmEp\nvRRij1R/awUljYsYHFlyqYxiQ/Sp86xGx7JOYmLkbVnpWgZI+qGlJ2Z1XbNmTdO6pOyU6mR1QVP5\nQLLnzZTJcQbH4XA4HA7HvIMzOA6Hw+GYc0wnKjhULWjEJNjo2alfFUGMyLJly4CSuTnnnHOAUvMh\nPzfCMcccA5SaG1ksKS+rEbFRwq1/HEFsiWV+FNFb3+XxOL1emhWrS1GeYkj0XXlYSy6121Rsh2WC\nxDJZ5sZqe3qOzK479exTG8ojKyuVu9n9mg47l6ZR59vH/eA4HA6Hw+E4aOETHIfD4XA4HPMOvkXl\ncDgcjjlH3daUtiOm+l2m1jov3fLQtoyCWf7Ij/wIAEcffTRQbpdYB33alpFQVtCWiBUFW1GsyqBt\nIG05qQ4qsxz+PfTQQwAcdthhABx//PEN5ZDYGMptNjkhlIM9G+RSIRK0ZaUtJNuudjvNioVVBtXZ\nbo2pDlaMLKitFahU+Sg4Z7qNpHayW302BIa2Buu2M2e6VeUMjsPhcDgcjnkHZ3AcDofjIEUIoRX4\ne+AkYAL4BWAcuD7/fj/wrhjjjD2t7S5IZrPj1pxZolsxDCnbIQbg3HPPBeC8884D4OGHHwbq2Q1B\nbINlJ8SS6LPOWZ5lO+TQT2yLwifoOjka1Hcb6BJK1knCabFMykNCZQmUxfDUQWVXXQW1s7BgwYKG\nOou50adYMv0+MZ6VR6b0YqUkTl67di3QGHhUZbDOCq1Je12wzamE6nVwBsfhcDgOXlwB9MYYLwY+\nBPwh8OfAtTHGVwItwFX7sXwOx6zhDI7D4XAcvBgEFocQWoDFwAhwfozx1vz3L5NNgm6cKqE6s/C6\noJpazUt/ISZBzI0CZYrB2bZtW3HtpZdeCsDll1/e8Ju0HvoU62BNlsU+2OCc0taIUVBZVDaVVemq\nTvpd2h2VR9obORwU66J85RgwPWa1M8pT3xUss46lsjoWex/UnmKZxF4pHRsqQsyZ2kQsla5bt24d\nAEcddRQAr33tawH45je/WdTNltGa49cF29zTUA0+wXE4HI6DF3cA3cDDwFLgdcArk993kk18HI4X\nHXyC43A4HAcv3gfcEWN8fwhhJXALkJrL9AFbZ5Nw3apcq3GrNdFxBXUUwyC25MgjjyzSuPDCC4GS\nXVAQTTEDYnCsBVEde6FP67jPhjGwASulZ7F6IVkYSb+i8AcKFSF26rHHHsNCWhsFv5Rex4ZaUN7W\n6aDOs2yUPq01lmXcVFdrxaa6PN2etXWMEagyRgqLcf755xd1uummmxrKZO+TymQdLFoNzkzhGhyH\nw+E4eNELbM//30K26L0nhHBJfuy1wK3NLnQ4DnQ4g+NwOBwHLz4CfCqEcBsZc3MNcDfwyRBCJ/Ag\n8LndJTDdkAxWgyMGR9fJn4qulz8YWee8+tWvLq6VDuXxxx8HSj2KLH/EVtigmGIMpIURoyANjb4r\nHTEvhQWRYaV0vhgZaX3E6IixUfl0nhik++67r6iTNCxiTlRW68NHeavMSlvskWV6BOv/RrAsl5gh\nlUPHVbeRtqxc0hmpbcQ8qY6ptZbKqjTEBoktEpukzzpfSZbJmcqqyic4DofDcZAixrgV+IkmP126\nj4vicMw5fILjcDgcjj1GndbG+owRcyOWQ59iFsQMaHUvHzOpxZFYC51rLYDE3NR5SRaToN+l/ZDF\nkmUtVBexKrYcYplUN/mq2bhxI1DqjMRuiPnRZ1p/sRhqF6WlsogZ0fc6DY7V3tg2seyItZqywTzr\nAoyqTdSGshxL/eDo3onRUtl1b3VcsGVUX5jK6srCNTgOh8PhcDjmHZzBcTgcDsecw3r9tVZTYiyk\n3RDEhmiVfsIJJzScD6VGRn5npAOx1k4piwAlG1GnTxEjYNkQq7kp9Cg5e/HMM88AJQuifFUeMRjW\nAirVqaxfv74hL6tDEZMj1khMj/VUrDpa5szW1R5XPnXxoHT/hkeyuos1s0yc2iS9X2eccQYAX/3q\nV4GSsZHO55hjjgHgueeea8hb99nW0eq56uAMjsPhcDgcjnkHZ3AcDofDMefQKtt6LLaWR/pdLIws\njHT+qaeeCpQMApQsgxgApSGdiRgWMQB1flYs02LTt9ocy/ToOmvBZNMT66J0VC6xVVCyFNKwSL9j\nWSQxJ3WWXdYjsvWHY6E6iQ2zddanMNA50FAH/a77ZqOTQ6m1OeWUU4DSh45YKPnOUTuJ4VGZLZtV\nV5dK3aZ1lsPhcDgcDseLCM7gOBwOh2PWsAyD1W5YyyStxsXg6HrpL6ThUITwVatWAVVLm2YQayG2\nR2yHPoU6RsBqcCwTYy2GBGt1pfx1/lNPPQWUDJP8wIgFSdMUM6NzrQ7F5l0Xg8paUdXpknTcMj1i\nU4RCv9Q/0VA3MTequ3zeLFu2rLhW+pzTTjsNKPVGtj2lz9Jx6Y6s/xvdl6mYHGdwHA6Hw+FwzDs4\ng+NwOByOWaOOubEsib7LcshGsRaDo/MUy0jnazWfpinWQLqP9Bwo2RDLMlktjlgHa2Fk/d6IObB1\ntf5ZxLpYLU9hiWR82KT/SwujPKylmPUAbb01Wyhd+bcRdL7YFZtuXayq9o72hrZQ2+mzmb5IuiLl\nKXbutttuA0q2SOcpDpdlbqxmaiprKmdwHA6Hw+FwzDs4g+NwOByOPYbV2EizYWMpSWeh1biNpXTR\nRRcBpWWNmIB0tS6mxOpGbCyqOsZFrIfYIfniqfPWa/Uvul51VBlVV1t3C6UnViYto9pD59i0rJ5I\nlmQ2jpb1HWPrLkZH90VtpfQV1d36GFrUlmmn+kazfMUAWQuzVOujMqmMYmpsRHQxW8pbnqrVN1SG\nurpZOIPjcDgcDodj3sEZHIfD4XDMGnWRnsWCiCGwEaS1WpdFjbz9XnXVVUDJACjyd+oZV5Y7YiGs\nBkawFkNiG2TBZeMuSUci1kJl1KfYD6WjCOiqk6KPW/2KtYwSUistyzpZ/0BiVlSWl770pQAcd9xx\nQBlZ3eqQpK1RW4g1UXuqTGKO1BZHHnlkQ36Fhmow98I8kn2KXRE7I4ZO7AuUDJeYGZVJVlWPPPII\nUOqNdF+UlmXWLFtUB2dwHA6Hw+FwzDs4g+NwOByOWUOramlFrO8YMQViPayfFq3ujz32WKBkRdau\nXdtwXqq7EFsgJkCMiYX1i6PrVCaVxbIQVnMjFsXqXeTr5ayzzgJK9uS73/0uAE8//TRQ9bfTjIGw\nkdB1jcqmMr/sZS8D4OyzzwZK1khMi428Ldj7ozqoTrIYq/PqrPRVBzFyts2a9QOxT0pLddO9/va3\nv92Qlhg2aXWsl+upmBvBGRyHw+FwOBzzDs7gOBwOh2PWsIyDtTCycZystY5lBMRYyHOxjSsFVV2K\nWCCrdZGGQ8yAjeOk88QsKM+K75e8DtYPi3Qmjz32GABHH300AJdddlnD8SeeeAIo/bso/5R5kv7H\nskcnn3wyAKeffnpDXR544AGg1CiprNYvjtX9WDZFEBslHYwipKuOOn9Hf1bmjSMbG+qiNtL9Sz0h\nW1ZHGhsd1z1W2XSvpRcS+ydM15rKGRyHw+FwOBzzDs7gOBwOh2PWENthfcCIORDjYJkerdqFo446\nCigta8QAbNiwAWj0/yJ2wfq9EQti9SE6T3nruI0hJTbCesi1EbV1vVgNWQEpAviJJ57YUJcjjjgC\nKBkdaUlSqAwPP/wwUGqSzjnnHACeffZZoPT4bOuiMoslshZlVltjfdNYfzi6X9IvSV80PJKVU+yT\n9Q2k9HQv0rKKiRFbJPbJls16SZZWR32gTnNl4QyOw+FwOByOeQdncBwOh8Mxa1hmxuorUv816Xli\nBPr7+4HSOsj6ntFqPo1tZL3xir2QRZBlYGSVYzU1shyyfnDEfojREfNgWQrr5Vf6lR/84AdAyeSc\neuqpQGkVdMsttwClrxmANWvWACVb8eM//uMA3HvvvUDJ3KgMKrtYDRtbSr+rHXVcdbQRuVX2ulhh\num5hb/b7ir4VQKmDsf6QrG4mLYuiq4vxUhmk37HWbdZztY1vVgdncBwOh8PhcMw7OIPjcDgcjllD\nq2lpNcSO2CjVNm6TtCKveMUrgNIC6e677wbgySefbMgntaKSpkUre634LVtkI5jru8qgMikWlfKw\n7JHyEZthGR+r5dFxMRViHlatWgXA8ccfD5SsDZQWXPJQrLzkS0d5iDFRHdS+YmCUt9pfTJm0Mtai\nSW2mOlrLMpuvWKvR1ka9ktK1/nygtGazEdHlhfmuu+4CSsswaajsfVMe1vdSHZzBcTgcDofDMe/g\nDI7D4XA4Zg2t2K0+wjI5+l26C1nSXHHFFUC5Sl+3bh1Q1YBIRwNVPyhiGXSurrXWUUrD6lDEVki3\nouttbCrLntT5yRGUjtiZQw89FCi1N9LqpGWVJkmxpVQmtY++Wx8yddolsVO6L2KprH7J+i2y1lTS\nN+n44K4sHbW5tYRKGRzrm0dlkjbpvPPOA0p/Qda3jvVbpLKn0dibwRkch8PhcDgc8w7O4DgcjgMW\nIYRvAn8VY/z3GVzzSeC6GOP3an7vAv4v8DdKN4SwGPgHIJAt/P4xxvjne1j8gwLSQ4ipsf5Z9Knz\n5MNEmhvpUQSxINKGaJWe+nVRGpZB0XHrY0eaDh2XBkSMgNK2lkmWJREsu2TZDsuWyBrIslupFZCs\np6Q7sRHJrWWXrtV3aXBUFsucCYU/m7xOYqN0vuqk7ytXrgRKK66J4cbYYGK5xI4p3dRXjfLQufJj\npAjyiuX19a9/veH8FSsaLbXUN6TpSb0lN4MzOA6H40DGZP43E7waaBqNL4RwAfAd4CKT7m8CO2OM\npwMvB94dQjhv5sV1OBwHCpzBcTgcBUII/wv4DWAc2AS8Jca4LoTwDuBX8+PrgV+JMT4aQrgeOAQ4\nHrgJ+CPgE8CZZBOILwPXxhjHQwhD+e+XAyuAj8UYPxZC6AWuA07M09oB/M8Y4yO7KefLgT8BuoAj\ngK/FGH8+hPAHedqfDiH8XIxxtbn0V4FrgfeZ49uBvhBCG9BDtvgbxjElrDZEzIuNhSTmQJ+KrbR8\n+XKgZCDEHIix0GdqIWUjjIsZUF7yFGz92Ii5ESuiMttYVjbiuRgXGz9LWhB9F6Ng/fHo07JZYiKg\n1N7omPKss2ISbDwm5aE6i1FRekpfrJLqJJZL6atuYmYUtVzXyQtxCAGAY445Big1VmpbKJkX5aU0\nxAqJydHnQw891FAnlUF5uhWVw+GYEUIIZwJ/DLwmxngm8AXg/SGEy8gYjktjjGcB/we4Mbm0O8Z4\nWozxGuCvgI05E3Ie2UTnvfl5nflvFwNvAP443y66EtgcY7wgxhiA1cCvTFHcq4HfiTG+HDgV+LEQ\nwtkxxvcDzwJvajK5Icb4P2OMX26S3t8Ax+bXrgVuiDHeN0UZHA7HAQxncA5ghBBWAJ+NMV4UQjgO\n+EiM8Q35b7vVGeTnHAvcF2Psm0GeDfmY31qB3wVeB/QCX4ox/oY550PAkhjjr043T8cBgx8CvhJj\nfAYgxvgxgBDCnwL/GmN8IT/+TyGEj+X9axK4PUnjSuDC/LyREMLfAL9GxrYAfD7/vIeMfemJMf57\nCOGJEMKvAquAS4H/nqKsbwF+JIRwDXAyGeuycLYVB/4VuDnGeG0I4XDg6yGEb8cY/2MP0jwoYGMW\naVWt71q9i93Qyr6wxsl1KmIcxATofDESKVMhCyzr6VZMjhgZXaO8rF8bsRtKR+yF1a3YqNZiElQX\nG5tKx8WC6DqxMdY/T/qb2lOMyvr16xvqIhbEskmqm2Vw6mKEWV2Myqo62BhUSkcM0MrOlQ3lUptI\ne5PeL2tlpjIoUrksxsTiqU5ig6QDUnvJKm0qOINzACPG+GyM8aL86zFkAkihVmewh7D5pPg14JVk\nL7AzgAtCCD8NEEJYGUL4HPAeZq6ZcBwYaOC+QwhdIYSTyPqZ7WstgOxAB5LjrebcNhoXUoMAMUb1\nkZYQwi8Bfw/sBP4FuKFJfha3k02mHgI+CKybxjW7wyuBv83L9jzwWeBVe5Cew+HYzzjoGJwQwqXA\nn5PR0KvIBty3xhgfzgfzj5OxEyuA7wM/FWMcDiEMk9HyZwJvyj/fQUa7HwL8cYzxb/LV3/8GluZZ\nfjHG+Lu70xmEEFYB/wgsAZ4jG6g/DXwLuA9YTPYCWBFC+ApwF6XO4C1kL5WKHmGKdrgWuArozuv7\nXrItCeXz5Rjja81lbwbeE2McztP4ScqX4v/Ky/tgXg/Hiw+3ANeEEA7PX/K/RPaS/yvguhDCX8QY\nN4UQ3kamz3mM6qTiZuBdwK/n20/vAL62mzxbgCuA62OMnwoh9JNpeB6ouyCEsAQ4l2wrbWsI4RKy\nZ1mOOMbInsupkJb9O8BPA3+SP6tX5vV2TBNiQbTy1wpeq3f9LkhXoVW92BNpUcRA6Hqt7tM8dEz6\nEK3slZYilFvti/VMPJU/FZ0nHZA0NmIrrDWVGBq1gfW4bK2s0nqKVRKDo/awmhlr7aS8rWbHMmVi\nTSybIljrNeu9WZ+2TZWu1fxA1WpNfnDEHilSurUEE3skxk4MjnwluR+c5jiLTOB4JvAp4J/z4z8P\nfCrGeCHZgHkc8MP5bx3AF2KMLwEezs99bYzxHLKB8U/z834BeDzGeC7wCmBVCGERu9cZ/DPwL7lu\n4WrgAhIWJMY4Abw9T/fKGOMHKHUGd1KjR6irfAjhGOAy4JV5G3wA+JDJx05uAE4CTg0hfD2EcC/Z\nC1DbFh+KMf4VMNHkOseLADHG+8m0Nl8JIXyfbOLxzhjj14H/H/ivEML9ZBPdH81ZGGvldDVwWAjh\nPuAHZAzLH+S/WWZP1/4Z8M4Qwl3Av5EtJFbtppxbyMTK3wsh3E624PhScs2NwL+FEF49RZXT8rwN\nuDiE8ADZZOf/xhj/zxTXzwuEEK4JIfx3CGF1COEtIYRVIYTbQwi3hhA+EULYG0yxw7HXcdAxODnu\njzHemv//KeDj+arwt4ArQgi/SbZNs4LGff3bAGKMAyGEHwVel7MvZ5GxIJBZjXwphHA08HXgmhjj\ndqCpziBfsb4UuDhP++EQwjealHl3g0ydHmFLs5NjjE+GEN4KvDmEcAKZWazKv7t8OoDzgdeSsUU3\nkVmlfCw5x7enXsSIMf4L2TaRPf4JMmbFHn+b+b6ZbMLRLO22mu93kE3MU3w4P6fpNlGM8XeA36n5\n7TfJJmpHFSRiAAAgAElEQVS1sOnGGJ8m05YdVMgZ7QtijBfmzNX7gNeTWb7dGkK4jozpvbEujdRa\nBsoVv1b4Wr1byxdpNhQtWxY0YgjkK8UyB1BaSckKR+yG2Ad9iiWyfmpURutfxXrz1XmWZRJ03Prl\nkZ5FdRdjoXLp/NS3j8oiFkNaF1mZiaXSeTbCub0P1teOvS+pBVdaJsskWb9E7UM5o7atsZxinFQO\na92VllX+bcTcWG/NYrikwxL7JMbHes2uw8HK4KQtL33BBJnQ8BfItq8+CnyPxhf+Tsj0JsC9wFFk\nk54P6LwY411kzM/fkVll3BlCuGA3OgP59k7vxUxZkBnpEUII5wDfJpsE3Uy2vTWdvvAsmdh0NMa4\nk0yncMEMy+pwOA4cXAHcF0K4kWzB8gXg3GQB+GUyvZ/D8aLDwcrgnB5CODPGeC+ZRuD2GOO2EMIV\nwCUxxh+EEE4hYyv+tcn15wEbYox/ABBCeH/+2Qr8IUCM8bdDCF8ATifb2mmqM4gx7ggh3EFGkf9d\nbsV0GWCtN8YoRZ363jkNPUIzvAJYHWP8i9zvxydo1C901Fz3OTLW54tkfed1VPUVTmc7HC8eLCNb\nqP0opS8ju6hb3OS6AlpVi63Q6lorfjEHluV48MEHgZJ9ueCCbK2k+ETSdFiPvVBaAImZEcuhc2V1\no98VtVssg9gPa20l9sKyIam1U3q+oLpbfYz0Mzb6uSyjUl2RGBPphsROiOUQo2Wt1MRuWP9DluUQ\nQ6bfbZwutYnSk85IbaHjHZ0dDfnpOn1Xfqn3Z+WpNMTA6fhtt90GlGyR8lYaak8xY6qb9TBtcbAy\nOBuAD4YQfkBGv/5cfvxa4D9DCP9NZg7975T7+ikXdjOwLoQQQwi3kTkEew44gUyrcFauQVgNrCHz\nG7I7ncHPAW/MdQ9/DTwBSJGnfO8HxkMI38m/35incx71eoQ6L7A3AIfmeor/IhNT9+cUtc0nxQeA\n5/Nz7icTmf6FOWc2nmcdDsf+wSbgqzHGsdyx4hCNE5o+YOt+KZnDsYc4WBmcHTHGH7cHY4zXkVk6\nVZDqB2KMg2QToxR/lvx/ZZMkanUGZJOSd8UYYx4T514ydmctsCjPc2t6vdEZfI0aPYKuN3XZQGYW\nmyKtty2nrhsC3l2Tj8754O5+PxjQ1tY2CdX9Yet108aYsRGOtXLSqkbnW0sNgBNPPBGAn/3ZnwXK\nuDYbNmwAqhYRsjaxFg9aOenzsMMOA8oVk6IfW/8jWpVrJa3VolbIti52n191SyNGKw/t1+uaJ598\nsuEaaQDEEBQrzXy/X2VT3orkrDpaD606z0aGth57tYJ+oC8z+LqyP3vspRu4/fbMPdCtt97aUD4o\n77XtEzYyt1asOk91sxoOW+bJycnpMqm3kz3TH839bvUA3wghXBJj/BaZ3q6ZJtDhOOBxsE5wDjSG\n4REyq48JsnvyRzHGh/dzmRyzhDURtS9YUbr6nk5UoHxx67jOS0WW0EjPakIjStxOnkQJ68Vona3p\nRarr9fLWBMWKLfXCVfp6IWtiY93Ea/KgF7SOaxLSTDSoMuhaTQr0KXpfZbRlUd0Ea0arOgnKW+lp\nW0ICUdVZpquayAx1Nrqf14SrznwXqn1Dn1YMq20aTcb0u8qiietUYss6xBi/GEJ4ZQjhTjJG/5fJ\nNIifDCF0krl9+Nzu0lA/VdnUbroPctam+6U+pHZWO+q73PSr/ZRe+pyo3tpy0m/a+lm1KiPHZV6s\nib7yVlnssybYiaad3Kr/qs9ZMbPqpE9drzo+9dRTQPlcptfqmMogk2i1h47bxYHdahLsM2Yd99k2\nsMFTVS71xZbOrK+qT6ovqq6bN29uyCdNS8+cxoEzzzyzaVq61m6Pqc7qEypTHQ66CU6M8ZvAKfu7\nHClijJ9jikHE4XA49gZijL/V5PCl+7ocDsdc46Cb4Dgc+wpa+dgVknXXLvZEsNtGNrCgTR/KbRcd\n0xaUdUFvV6ZaWYkV0cpIZdTKVysl6xZesNtryt8GxxOUjz61yrTbQlCu3rSasytU6xxMYkfVVdfZ\n7SB9F2tl3e3rvokl0HbdE088AZQr7ZHhLP0XRl5oOF+r+mbsiq2n6m+3H1Um3T9rct2svfY1LNNo\nV/66D2oHtbfYFa3ixbpoG1T39dBDDwXKEA7NoPZTP1SfseEIxAxYwbPtn5a5Udmtsz3Lwure63m5\n++67ATjppJOAkuHUc5HeP5Xlsccea8jTOslT3WzgStVN7W77tTWNV5vY48pPLJME0toqfnw0C6uw\ntDurixg6+6ynrJja25qc6x5fdFHmsP+RRx5pqIvaQOeJ6dH1qkMd9skEp6+vbxIaKdpm1KAd+Ht6\neipeGFtaWir70WNjY4UC/+KLLwaym3Tb2G0wCVcsugIoXxwjIyOVgb6zs7NCpS5ZsqRi26/zW1pa\nClV76sVS9dKN7OzsrGxZ9PX1FQ+3Xm6i9TZt2lSh47q7uwuaXGmkLzvrm6G1tbXi8+C4444rOke6\nbSH9gQalww8/vLgPenA0UOt6KC0gnn322eJB0u+9vb1FG6pN29raKhF/L7/8ciDrpO99bxaPUYPb\n0NBQUY40irD13aB22LlzZ9HmqbdS3Y8dO3ZMS5MQQugg8yp9DJmvnw+Tmd9fT2a+fz+ZXupA2+Z0\nOBwOR4J9MsHZ3cSmpaWlssLR97Gxscoe9cTERPFC14uso6ODE044ASi1CE888QSDbYO0tpQTqNTx\nlJ2ItLa2FjNXYdmyZcXLe+3atUC5ajvkkEOK1WdqFmdnrxMTExXX1m1tbcXK2K7yurq6iolAugLV\nS155Dg4OVswvhbGxsSI9Tcx27NhRMYGcmJgoJhuaTG3evLniDCudrNlAbIceemhRh9T9uV0ZtbS0\nFCsx1U9t0NHRURyze+Zpedva2iqrwbQP2FXY5ORkRbcyDbyJLOL1m3MT/HvJAkNO2/GZLYc9blfd\nmsSrX1umwWpwmu21S9xrmRAbNFB5WX2DZW40yda+ui2zdXlvV7hWu2NNW/Vd5dH9T++XrlVaqrcm\nucrbBki0gRUF9SPloTqnCx8oJ9Q6rk+lp1W4FgdF3Qezumtyr76s/NL7au+TLavytK79rcM52/77\nAyqLZZ1sH7JaENsXrbmxztf4FmMs8pQZuPqbztEiS+0kJsUGorRsSDqOpN813th+bR3PWSZHY6K+\ny9RbfdoGKE3T0Lm6VufakAqCZXY1dqpsaiulb0Mv2GCmul961+k81WmkJR+Pexs1V9ahYNrf7XOu\na3RPNX4prwceyIT7eg4sa2XNxetwsJqJOxx1+CyZiwDIno9R4Bx3fOZwOBwvLuyWwdkbdL1dUU9O\nTlZC2Kerc+vee3x8vKInWLhwYWVFNDQ0xFhn9nu6TQLZCsGuIiYmJooVhGaqixcvrjh8Ovroo4vr\n7G9tbW2VbbZ05ZvucVqrEZW/p6enWKFq9rpp06amKwPrnCrdvrJWGdu2bauYzUI5E5ZmYuPGjRWW\nLdUw2FXX+Ph4xax5YGCgYesPsntkVzvaa16xYgWvec1rALj//vuBzPGT7lvaL+y2md0/TsudblFN\nFzHGAYAQQh/ZZOcDNLoAmNLxmfK07KNlnfSpOlnLJvus6LjaT5oQKDU4li2wqz6Vwa6atfIVY2kd\nclkdi2WjxFroPOWjlZoNUqiVtVbe1gQ+zUtlVNq655YtVd313Ki/1VmpWUanzkpFdbM6AvXB3vHs\nGeoY72g4XwyOmEtrtZWmZetimR3b7vpumZ1ZMJZzBvU1u5JXH7JsicYcbc/rfDFkNmxCapWjcVj3\nTLoo2xd0jdXo6Lg1r7ftaC3udG8t42MDXKqfayxXSAnpVJpZv4kd0rXK01ocqj0sS6S8JZ2QOwXJ\nG4Rk2x6oWk4qfY0vek70ruieaHQIaN+lttzpb9Y9hNrn5JNPBspxQn1AfUR56bvSSeULzTDVFtUe\n0/UWdrCH6sQmbYw6KhzKzrhy5cpCgJZOGLSloRelfuvp6SkGXnWm7u7u4oamL+JUBwONdJ5+Uwfr\n6+urdPyxsbGGLSHVywrmRAu2tLQU22x6mHbt2lVcm3ZC6zUzrbvdJhkbG6uIy7Zs2VJ0HEt5qyxp\nOTo6OooOmU7kLI24ZcuWptShpddVns2bNxft+7KXvQzIJj+iL63HTFsvpW0nEGk5Z4IQwlFknqQ/\nHmO8IYTwp8nP7vjM4XA4XgSYaoLzWUrz5Tq6/gpmMMFxOA5khBCWA18FfjnGeEt++J7ZOD6zTJjd\nS7eu0uscAdrvmuilDI71GWM1LtavjSbUmuhr8qkyaYKq1Z8mr9bCSN8tG6JJtFZiSkeshtKTts0G\nS0zLZl3QWwMBu3oWrEZHEHukNpCViFaoNoCiXSzoeqXbtzgX1+dMjtpCdT/uuOOAUluR5mE/7ULO\nMrGW3bPajzr9196E1TSpr1k3+9Zfi9V9aSEqww7VTfclhFDk+fTTTwPlvVC/srpEyxZZyzqrLdPv\nekZtmS2DbQOKagFomThdb5/11HDDWoDZZ60OKovaQHVTnfUplldtYrWbVssmhkjPie7DyAtZ+pt3\nZsyb2koLYavZgaqeTmWWddlLXvKShrzkuFR+jaxfHOt/qA67neDMBV2foqWlpemK2tKBqQBLDdIs\noqyotbPOOqu4uaI7e3p66KCDzs7OYuBJnS9ZB0JLliwpbqDy2L59e5GeFVG2trZWHoKenp6KkK69\nvb0QfaWOyZSOpQ937dpVicY6OTlZiWHS399flNc6DVu4cGHF5HVoaKjoJKJJh4aGmpr9WpZNA9WG\nDRuKNkxFo8qjcHw2NNSwfZi2B5SdVJ9PP/10cf/OO+88AH7sx36sEHHqHoyOjta+zNrb2yudf5a4\nlqxP/24IQVqcdwN/OV3HZw6Hw+HY/5jSisrpesfBhBjju2kejuLSfVwUh+NFAS3qxEZoMWOtzLT1\nLubg2GOPBUpmRj5QtKASs2a9YUPJCFj/QIINwmh1WDrfbpcLVuNkWRXV1epW7KLX+nOxVkSp2wst\n0MRK2TLUWSwW2picFVGaWnAqPTFj1qLJajktC6xPMWw7d+1sSMfeC2nOUmhRrbR1jlholUFsreog\n6yp7n/TZTLaQYiqR8ZzR9dBItVkKFqqdpL29vaIbGRsbqzgy6u7urrgBX7p0Kb2DvbS2trKse1lD\nGsPDwxXdSHt7e5GvOsbw8HDxvzpyKni09GRvb2/xfxoB15pNtrS0FJ1OZUrzVEdWh0+Fv/pcvnx5\n0WH1sKVmmqLu1GG2bNlS0LoaeNL6pKyH7oPS0/ehoaGGQQZKrVNa3qVLlxbMlMqRMj2WwVm4cGFx\nntpq2bJlRVRdtc3o6GgljWZ+gZrpvPYlrIm8ZcmsENI+3DZulKB07JYNlNGSdT/rxMXWBYClfvXd\nOsuz4nlL5duBW9engv30OkvRW3Fmeq014xarax316cWp9rNm9Sq7FUSLPVQd7EvTvgiKbaHf//0s\n35dn5W0z4Wk3fOhDQFWQmqZlt6BsCADbR6yI3Pal/Wku7nAcaJiKwXG63uFwOBy1sJ6LNXlOt6yh\n3I6XgYV0FmJ25DlXCzBNCMXkaIEG5aRTE2fLStjYZ5YFsRNF6/NLdbLaGUGTX6Wj77o+dUKbpmc9\n8KaTXus/ps7/leqixaD116RFgtJWW2hib8tgmSEbP0tQucS+HNKSLbBTn2VQ3sd0gab/rWRBhiRW\nU6XFivqMFl42MK69LxZTaXDmnK5vxtxYzYQadteuXZXgc+Pj40XjqKEnJiYKU0TdtO3veQ9DL80F\nfKu/kP2Wp7/sIx8phFdq0KGhoeJYWkYr/hM6Ojoq1Gx7e3vFvDMtb+oy2zqmSl2CqyPq/MMPP7z4\nPw26J2pS5UhNfPXApdGgtfJNzdCt6/cFCxZUqMr0IbYrzZQKTt2Vq2Om4QqkH1L6YqD6+/uLB0AO\nnoaGhoo667xdu3YVHdqanKcm+nalPVOEEM4H/jjG+KoQwtnATcCj+c/XxRg/M5107ABRZ8pbZ0au\n860LBN33U08tg77rWbCmzBpo9WnDR0jfpOdArIkVY9qyWmZV11mBofJTn7a0uRUKq4+medeZTtcF\nd7QB+Kx4U9CzYZ3D6QVg3VekLOt0IG/fMmFO6XTdpzrzb8tM6tOa29qgm/vTTNzhONAwLU/GczXg\npw9fswfRvhBGR0cre6tQPtSpF9a6CMHN0NXVVaSRem5tNnBZIa9eCIsXLy6OpRS4JloaiJYsWVKx\nfBgdHa1YoaRWMJpYKPbH2WefXfEWPDo6Wrz4VSb5Wki39jSotra2FuU4/vjjgUxdb/3KHHnkkUU7\nyI9CaiZuX5BjY2MVtb/OT+u3ePHioo72hbR9+/bKfTv66KM566yzGtJbu3ZtJeZRun1gJ2SzQQjh\nfcDPkgnoAc4FPhpj/OisE3U45jHsuF1nPaWJumJOaUGmiaOYHPnHktWZNDrpuGJ9HSltWa5JL2J9\nSykvu5WncbWu7KkrEKhaKdrwQTZUjt2itb5toLrFatNOx9y0TFa+INi4WDrfflov2nZLVeO2XQiM\ntzQuOHSd9RWUpqG66D0p3ZW13KqLkG4jl0+12JiOyNgHfMfBhMeA1wP/nH8/FzgphHAV2aT+12KM\nO+suhipLaU146363JsOpJR5UB2sNYFClyNN4Y9A4SU3zti7pNRim2rS0jBpYBDsI28mrBijrMFDl\nsS/HdFC0uhO1mxYdyss6/bIaL730RKVbd/miv63GyTpEs87GpjIfVVuqrukkvi4Eg3UhYIM+puFQ\nYGqK3uE4mDEdBmePB3whjQ2UDvJ2oE9X5Rrk0oHYbmf09/cXdvRC86gdFGlZ1Xw60Fp2R/lCuQc4\nNDTUIFrWp14A6SBpPVC2tbUVA9O6desa6pwO8Bp4169fX4kbkuZrxZdLliypeH3dtGlTsarRimh4\neLhIV0LL0dHRiig23WJTemlAVBucc+vWrZXoul1dXZWtltRTrN3LHhgYqGwVpN6bbRl37txZWZnM\nZosqxvgfIYRjk0PfBf4uxnhPCOFa4PeA35xxwg7HPIXYCav5sL5K0u1yqPph0fWauAqaIIrRTq/V\nb9azudWbCBp3rZGJYKPbaxxVHZWerrN+b6zGx/pWst6cU9ZF7w61XzMDkLRsKaMF5TtEZZa350cf\nfbQhHeuLx36KWVN6Sqdgn9bn753WRq/l1jdQur1tt6pVR90HLRoEtVsafSD9bObBvhmmnOD4gO84\nyPGfMUaJOm4E/nK6F9b547HMhAY/a6FWFzTSig6h3KbUAGpZA6v/qZv82QHYap2s23YNOJaKt4EW\nbd00obaMhdJN0xYTYql41eXuu+8GSoZGW7DSgWngtW76pQlT3fS79Eyqk+qg66a7Baq2V93TNq/b\nXrB9w4phreO4OmeQDodjdtHEZzzgN3voUibH7gXqIU8p3XQv1M6+n3/++WJ10MwG32JwcLBiprp0\n6dJiVZA6jLNeaPXi2Lx5c0F9qzzbt28vyqmBPp1lpyadKbOSnr9w4cKKCfTzzz9f8Ty7ePHiyiCX\nMkTWzHfFihUV0/Tu7u7KvXnqqacaTLvTcnd3dxfpqZ0XLlxY3DfNwrdu3Vq8DNKXtPJNI6erjDYm\nUWtra+Wl2N3dXdmCSdkdu3pqaWmZC8+uXwkhXB1jXA38EHDXnibocMwnWKsdjXkaKzRptdoOPeti\nljUGWssoXaeo2FCO0WLANQZLhygWQhZbEnpbb+JWfG+9iQtWA2nZCqvhse4abLw5a96f5q2x2451\n1jWD8kidz6ZlUp30brSsh43Iba3g9K6RvyKJ5rVTsGUy23bWe0htqTZI2RW7lWo1NZbhsmW2aSve\n1lSYzQTHB3zHwQDN/H4R+HgIYRR4DnjHdBOoW4VbfyvWas1OOq0FjV3FQ8ngaABQmpbqtWkIdkvT\niiitfxZLa9vgnEpf323wTS0OBGuimraHrrFMiAZ61V1UvMyKNYhq8qx2VFlVZ/tittuwdlt0Kvfw\ngkxf1XYpE2eDawp2K1r3zwa+FeoCuzocjplNcGY94I+eWzINzUxki/3Ljjy6cmu5Pzcxma/KqT7A\nXd0ZA/Dk4ifZOpIxBktHsoGz66XjPHF089X7judvrOhs+rv6iz3FgZ1V1qUIfdCWhz4YHipk1wXt\nP1lG1j5kLBeAjibRi8ez9Ee7RxnvyC2PlmS/PT2RDYYjIyNFeosXZiuZ8R3jtA7k+qTt2cuxf1d/\nxaJow7aMol84sZDurjzi68ZsMO/q6mKgI6vX5Pq8PGMjjJLfm+3ZsS3jW+gdyV+K7TmjRr5yaO1g\nR0f2sujuyNLvX9BftNH2vmxmv3V8axl9eihnynb00TViGCFyp4mtI2wj9945mVsMDEyyczhr4C3H\nZi/JrkO7itXeYGtWr47O7OU0Ojxaux/b+b1q4M/dIca4Frgw//9e4OKZXD95vrHQaMlXYG05M9nR\nuG1UXDfauF0x0ZrvmbfngVDH8hgyi7J7cO+Ce4tri5dxW6OFRWtb/tnSqE9S2sJ4az5pahttuK69\nrb0h3eG27GXd052tMnsns74ymQ8P3SONuoFdbY2Ttq6JrA/0b+unGZq5HWgfaRRZF9tnLVn/f2Jh\ntjpff3S25XRn251ZGUd7Gs4fn8zquHVh41aTtXK0q+uJttxhYz4+dbXlFiG5g7+1jXFzC2w9NI9Y\nvTx/1luTAMP5vZxsMVtMrY3fBzuzfq77MTaau0AYyftOfhsnKa9r+e6+neRoAqmJn/UBYxlby4bY\nCaSg+yLm+fHHHy9+s/o8scfqI2JwdO3DDz8MlFujYoksq2SDDtvFiA1BYxcnNjaY1ebYiW066dUx\nTWo10baBluvikOnZUdmstZW1PFN61mGt1epo4aAt3ULnNDbQcJ5dIKR6UbujIFiNkr0f8pVko8PL\n184e+cER9nTAF9LVSkM08cnmIqqRkRFaaRyomSxfaov6sgdl6dKlpfle29RVam9vZ3gku9mTE6UL\nbdsJU3Nr67G1pbWlEomblmwSADA+Nl5JQx2oo6OjsipXedKHPKXntBpVGiMjI/Qu7G0okzr25EQp\nbNYWWEtLS9muw42BEQH6F2cDwqK+RRVfKpZShnJy2dXVVXkQW1paivulY4ODg4yMZvku6F5QtI3O\nsVTl+MR4cU/1+cCDD1TaUuN6a2srLa0tRf0PBNhVtV5CVn+Rz91pm2ik5IuJsQlkqfuRDo4LenIH\nZwPZQGsnv3Xek62VlGU7im3j8bGm12tQKyy9TN8q9C09jYNqGssMysE39WGjPtTb1rjFIShtbV2I\nFbKWRtZ/TtHv8nFH+ajfWBbMClXVz6bClq1bGsrTbLvUpmUH+OIFOd7oG8mm5Z6MHY4qZuMHZxVw\nPdna4X7gXTHG3T5VXfc0zjyhcSZrFed6oU1smqjoSyYnJ4sBR34TTuw8kcMXZP/3dOaix9WKLAEX\nrW4UaW698LwinoYGjGOXHcs42f+bRzLKu6ulq5h49PdkE4CFndkkYsfwDrYOZ7NJDaQtLS0sbM1+\nP6wj20Ps7u5mkJJFUZ7DNA6iAyPZi+KFoRdKiyKy87cObGVRWzaZW9GZWRIc1n8YxxyVLR0lplzT\ntqZoI7XNfTvuy84/7LAi3VTJrpXHkcuySLNjY2OsH8tm6pYOn5iYYNN4NmE6sjs7f+WylcWe7POD\nmYXZ+pH1FT8SvS29dLbkQtT2bPV0yMKM5RoeHi7acnR8tPjs6Wp8sWxeu5kX7s9War19vQ1tOjw8\nXJlo7bdQDd9tnNjY1dqCvsZVofXhZF9iVlu0ZFV2v89ZdU6R55Le7NgLw1n79C/I+mtnW+P+erGa\nm8j6X1dHV8PvG8ZzU+rhRlPqgV1Zn1Hf0YS70CJMNjrss0Jola+vK3u2ly/JYszoWdfE6omtTxR1\n0rOxtC+buOyg0erGrv62jOTbY/lzPDyZW3uMZf24szUr45bBxomHNbG3K9BDFhzSUCctuia/c3ND\neV5uQjV8s6vRaV+KYlzraByCVabO7ixvK/Ke3JWLkYeaW4Huj2jiaifrqFLj9MknnwyUq3EbkkTW\nUTquRZmd4GvVDlWtjCaluub73/9+Q1nUVzQJtq4C0hA8adlsPvaZtG4PbLRx/d5MJ5l+pv+rbOqX\nVoiue11ZYOewk12lYxkaa+2k/NVGYm6sta51QWG1Pap7upWre5sGvE7LYFkmvZf0HqvTPO2xFVUT\nPzgfBa6NMd4aQrgOuIpMbFyLZt6Lm718mt3IygM+OVnQixKQHX300RXB1e42JRYuXFgRMmvrIz2W\n5ltsuSTxsuxgPjExUdwg6QKWL19e6Zitra1Fee2g0NvbW3RcWYX09/dXLGm2b99eWI+ccMIJxbXQ\nKBRO6UrVWekuWbKk6Hgqz/j4ePGysWK/LVu2FGbfononJyeLVbcetPHx8UpQutTbtNpBD9zOnTsr\nUeNXrVpV0JESlG3durUiVE41IM1WtbOZ5OzphH4q1Pk8sf3EOlAUrB4GysHEegi2nnLtYGXNL9Vf\n6gZmS+UrPW0J2IFf5dKnNWW1AsWULVGftBZZgp5TpSlY81mrWVId7SRc/VzPkfq/dUgn7H5opYil\nJod16ZakFauq3+vZs1sZzYSb6XX7e1LvcByImI0fnHNijLfm/38ZuIIpJjgOx4sFczGhdzgOZmgy\nrMlbGhQZqt5+xUBbL8GaBIvRSZkKa2Fl9WzWl8yRR2aMs/XFpUWWFgY2/I0mjJpg1onzrebGLlas\n53WVK7X61Tla0Nk6Wm2MDahrmRzLBKeL2PRTdUjlD+l3LT4L6UZ3IzOptlbddL/TKAR2MSHYtK1j\n0LqwKdatQh1m4wcn3TTeydQOPRtWmpb1SEXGzVxR21gsKTuQOtibCTU7OjpacUM9Pj5e1Xck5dRN\nT31mWPPIycnJ4gaJfRgfH6+wNe3t7RVBmMzi2traig6u6w455JBKGIIFCxYUx+69NxObigIeGBio\nrB2zrcAAACAASURBVNDXrVtXoeSXLVtWdCQxWGmk9TQvyDqpyluISHftqrAFnZ2dRR7q7Fu3bi3a\nRgOWmKT29vaiLUVXd3Z2FhSp2i3VSdkYRh0dHRXWwf4/TezxhN4K6eu0I7q/YgasaNDqkpSOBgWZ\nxELZbpals67MLX1tXwR2m8YO3NbM1sb9sv52rMmr8rMelfV76rBT9/gHP/hBwzl6VlQHmQrrxak+\naj1CWy2Ndcdg495Z4arKXvj+Yfc499xzgfI+qt+nZbUvPrtNqU8r2LUDvmCtshyOgxmzMRNPn6g+\nYOsclcXh2O+Yiwm9w3EwwbIcghY3Vv+lCaPd8rPbcjbqdbpNbyfSVr9it2StBZCYGquxmUrYr4WA\nXXjr/MIIxVj9iKnRIk+MkbwEp1BZLYNit6zryi5oi1WLVxts2RoeaLGrslm2RHXZ3ttYHi1OLRmQ\nSj1sDDArzbALJm0zy9WCFhuSfkzXS/1sJjj3hBAuiTF+C3gt8I2pLrBiy2a/pUgLrz3ptKNpJaRO\numrVqsrqjN/7PRi/g+7ubk76tZ8ByjALE1u3VqjIwcHB4iakTIBWf2r41OGeOkq6wrZ74qkr8JSm\n0//yuppGLNZvGiBaW1sruoDu7m5OOeWUhvKmHVoskFb2ExMTRXoSkY2OjhZ5pZqI5cszAag+U6sq\nK6YbHx9vyrxZd+tjY2MVk3u1VV9fX6Gn0gDy5JNPlqb5iYbCPhgp26QyzXF8nhlP6K0o2GIqSxhL\no9s+1cy9udpFz4F1kmihNtLgZ9kLy8rpfA26GrD1XYOjPnUfU8/EaT76VDpyxJY+x7pWfVP1tuyg\nxgM9m5b5Ul9UXpYFtLGo6gZhq1fi938/K2fXPQCEq98AlAEED8lfVHqO0kjpaYDeNI+6bYc6U+q6\nOGcOh2N2fnDeA3wyhNAJPAh8btoJJA9fSpXbF0Eag8OKIltbWyuBANNBIbW9b6WV0dHRyp7m4sWL\nG6KIQzb5sFsznZ2dlXggaXA8q5jfuXNnQ/5K124HpPGeNFtOX+LWDX5aJk02Nm3aVLzgNElKJxVa\nBZx++ulANrha8XJqnp1uA2lAVrsqum9HR0chFtWAu2nTpopIdmJiorJKGxgYKNLVFkPqYyIVMus6\nvZBijEUdBEvZN4tUn0449wAzntA7HAcTKv6cTGwiPcdWYG7HNBs/yloZpuObTcO+CzTWWQeUdkKo\nMcX6v6nTdthFh7VEsgttG+vKMjzphFVpSS9kt0RtGfWesFvc1q+T3X62gnXrZNPqjeQ4U1qpyeHs\nfLEpSl8WTyqnfk/zsm4irFWZdY+gPmGt2myd6jAbPziPApdO5zqH40WMWU/omwWUhXp/KHWDpGUu\nLB2fMjiaxIvy1kTUskE2rpLS1gBiLYmsN1/rFMwGOVT6ts4akFQ3WePJ0uiOO+4A4Ctf+UpRJ1nq\n1bmuV1lUZ+uUzUYft4JFG8nbBgC0ruytRVgh0tyS3SctBJSP7ollmNK89PJQmXXP9TLRAO9+bxyO\nmWM2W1R7hGYuxe3eogaUwcHBii0/VN3NP/PMM8UAqgGjvb2d8YlMkLxpcFNDGj09PcWArZfEwMBA\nhYpOmSFLHQ8ODlZMtycnJ5vOou3Lobu7u8jLbsH19fU1tfm32xKHHnpo5UUoNujQQw8thJlauRxx\nxBE8+eSTAA3bUmJTNPtOGSflmW7F2X3WJUuWVPZ40714/b9kyZLCIZsVTo6OjhYvXB1btGhRcZ4G\n+YGBgUpk9lR9P5WifrrwCb3DMX1YIboVqmucslvZGiOt52ONLfq0upc0LTseWZbDyiKsGb7KoHeD\nyqI6WA+8ljnSJFgslb7rd42vYp8kGdA4pnE3PaaySFJhjRKsD586P0TWYasWFRqTNebqOhumRelb\nPzfy0SSpg2VZrBsGoGIkI9hxXGW1+iLrRsR6a67DrCc4IYTvAXI7uibG+PbZpuVwzCfUbYvV0d82\nyKx9YWjg0GCrATDVGmkw1ATfumG3HorrwhRYvyx2wFee9tMyO1YsaAWTZ599NlAOfN/5TuYlTw7a\nINPWAVx11VVAyego+J8m9HYbQEyJBmCxIdaTsbVUsnoXa+prX7w6b3goe2GsW59Zc6mt5ZhO1m5p\nOfXy0OJBLI+1oLNlssyNMzgORz1mNcEJIXQDxBhfNZ3zm5mEp7AzTw16KS1r9xuhkfrVYJjS2Tu7\nd9Le3l7Ed9GMdfPmzRW6u6+vr2KLPzY2VjE9Tl3VWx8HbW1tFR8OY2NjFZ8LUPX1oP3K/v7+4re0\nrtYp3IYNG4o20UspFQDrWmlxLr/88mKg1ew3xlgo4CVY3rx5c22ohq1bt1Yi0Q4ODhb3q3Cy2NlZ\nYd40kKftm64g7LbD8PBwoaDX58TERK1331Rvs4dm4g6HYwawW6uWwUk9vUN1ol/nyNK6EEg1eHab\n0bo1sB5yrfbF6kysxsOyzPa9VXeddf9hI3WLgdeYnY6L2uJcs2ZNQ5r6TBnvtC7NYj9BuQCw+hdb\n9rq61S1yWhbk8RtzkX9qyALle0iMEZQMl8pqDUF0jV2sCNZ9gt5JU7mHmS2DcybQE0K4OU/j2hjj\nd2eZlsNxwGMmjGUz4XN63A5U1q27dXluzW9tcDsohdiaNF500UVAvb7ECg3thNvGu1JZLKWcitPT\n8zTB1sRd6YpV0bamjQSeDvh6YX7zm98ESu3Lqaee2nCu0rJ+b1QHa1FmhaD2Zad09YKwn2ozDfAK\n0qt8tBWsBVhK1QvW5bzatS4QpUcJdzhmjtlOcAaAj8QY/yGEcCLw5RDCSTHGptOplE61A2OqY7H7\noikjYgd5KAfHtra2wpJIM+Rdu3axbem2zHx4eaNp6sDAQDFgaMBdtmxZUSalMTw8XNkfTgcay+B0\nd3cXg6euGxkZqewPt7W1VbQ6qe7FOvHq6uoqBlOdt3nz5ooeRWVbsWIFb3hDZrK6evVqIBuMVV5R\n/I8//njxIlK4h+XLlxdMz9q1a4v2UpuqbVJBp/Ut0d3dXXmxTkxMFOfpvqXxRpSHaP1Nmzbx4IMP\nAiXl39nZWVl1pA7o7Eshbec9wUwZS4fjxYYQwmHA3cAPkblGuJ5phifR2KTn1FrR6Lh16GqDmVq/\nLNYZZBobT+OrnZxqXKrzEaOy2Umn3bK1E3rBiu4tW2XZkpS1B/j85z/fkI8m6wBf+9rXAPje974H\nwAUXXACUsbxSx55pWdIQOVDVDdlt6DRETnqdYN/D1uGrAsRatw1WE5qmK0YrZf7TMtaxSXZXxRoQ\nTMXQz3aC8wiZx1dijI+GEF4AjgCe2d1Fk5OTlQKlwlVb2ZRyS8PXW7fPLS0the+JVPQ1vHCYvr6+\nitvt9vb24uZpf763t7fSmKnnUfuwtbW1VRo5FbqmHkqtaeLExERBt1qzuXS12az+Qnd3d9EO2oZK\naUB1Xq1u161bV3hWVZ5pLKo0EJvaUGVJI5JrhdmsE6cd3VqFjI6OFnVWutJUrFy5shgEb7vtNiCj\na60b9LQNrNffiYmJ2thNcwBnLB3zFiGEDuBvyRauLXh4Esc8wWwnOG8DzgDeFUJYASwCntv9JQ7H\nixazYiwFuzIVrHM3uyJLA5pC6b/I6g7S37Q9osmxVoh28VDnXdY6lEvDb0A1qq/VNFjrEk2gtb2j\nVb1E0dKAveMd7wAaJ/hy/qcyawX7mc98BiiZSE3AtUqUZYrKJEtCG+FcZVIdrVdZTdiVvrXkUFkV\nwZ1MKlacZ73qpqtT3UOVweZpY/TUxeKpc0kwQ3wEuA64Jv8+o/AkdrGjdq/zV6bfrY8xa9FkfaCl\nTiPFJljHo/q0EbKnctVgnz1rmWQXT5bpUT+3VsLWAszqIBVmBMr+fMsttzScY+tihfx2O9nqfuzW\nq+qo/qvzrD+zuvRaWxrHMy1O9bs0OalmSn3BxhuzC32r41Ld1BY63xIJdZjtBOcfgE+FEPQQvK1u\nsE+ROvWzdGX6f0pX2Y7S1dXVQFUKNqL1qlWrWLBgAccccwxLe7NtqNQMzgqzurq6Kj48Jicni3z1\nQKWaBcv4tLa2VkTJCxYsqDi42rVrV3FDrZOloaGhCj06Pj5elFcdpbe3t/hdA3i6ZaXOe8YZZwBZ\nR5BYVy/Ozs7OIt9vf/vbRTuoLZWuWK5du3YVg4zKv2PHjoYtJLWDFcjt2rWrKJ+2BVPHUPfddx/Q\nuC1mB53UQaLaNDVNbcbczFFsnlkxlg7HgY4QwluBjTHGr4YQriFjcDw8iWNeYFYTnBjjGPDmOS6L\nw3GgYkaMZR1TYyf31kzZxmOxEzar10rNze1qTaJjsRJ2hSlvqUrDxp+xkYXrvKRa/Zc1eRdjI12B\n6v7GN74RKB39aeL80EMPFXXSZFfWGJqMa9WrbenU4SGUK0hdJ42I9Qyribkmy9YXig03oTop3WIF\n/UIjA6d0rc+V1HJEiwa1n5gIa7lit8Htp+1rs2Bw3gZMhhBeDZwF/BOwLPl9yvAk6jtanVvzevVr\ntbMVg1tdi92SLszxEx2m0tJizbKkWhBaVkKwkc0tS2QdXtYJ1a1Gx1pVqRxqI4Wk0f2WgB7K/nfS\nSSc1fJfhgPq/fdZVd/VztY3az4b+sS4kbLwua2FsdUYdSzsa2qBuvEslFmo/1cG6yrB+baxVnO6D\nnn21XxqNvRlmaybeCnyCbNAfBn4+xvj4dK61JuFp49gb197eXhGNdXV1VURqS5curVh3rF+/nh0d\nO1i7di2Pdj5aXAtZ57a+MUZHRxvEtJANQjZmlejqbdu2FempjGmMqVQMVfcyACpsVEdHRyW0w9jY\nWCXOUmtra9HBrOOmsbGxBvN3naMOrDoMDw8XHUtbGwMDAwWN+uyzzwLlC2PDhg1FHSR8W758+W7d\nZqvdRkdHK04F9QL7zne+U7yQVefJycmm8XfqqOaUqUnNKOfI+mRWjKXDcaAjxniJ/g8h3AL8IvAR\nD0/imA+Y7RbVjwOdMcYLQwjnA3+eH3M45h3mirG0k3vrOC6dDEI5YdRE2uow0kmdJrKauAsKF2C9\njiov6wDQRoC2lnqaEKtM6cQ4/V1MzLe+9S2gXDFfckn2PtWiQCswlUP1gDKOmtgmG9JCFoKqo3RI\nmsBLwyOGR4J7tYXKqjpLo2CjI9vVuoT2aoPjF2UWnEvHljbU3epCUm/odsVq763ViNRZjdjvU/kF\nmQYmmWF4ErEgake1XxrnD8rFi9pBns0F61lX7a1FZsrU6R7rXgjW75YWf81i/kHZ/srDLmj1qT5n\ntWnqc6qbLbP6udpAz4nOkxYNSnZC16hf21hU1qu+ddGgtC2jZjVNdZZJtr/Wxfvqa2/U09i2Tfui\nDRmj9rLtYcdAu3i2DJBl0CxmO8G5CPgKQIzxuyGE86Z74e5W1PbhbG9vr1S4o6OjCNqoG3/ccccV\nDfb44xmRtHnzZnYt2cX27dt5ZDyjsjX4d3R0VOi6ZcuWVYRjra2tFWGXBlvlAY0Rvq2QbmxsrCh7\natpojymfbdu2FR0tfbnohmpgbGaRppfEtm3bKm62W1paKm6wFyxY0PSlqRenqPi0E8nTrF4eK1eu\nLB5mnZ+GwtC1opIB7rzzTqB8+W3atKli9pkiFcTa7Zn0haC8Usp7jjQ4Dse8h3GDcOn+KofDMVeY\n7QRnEbA9+T4eQmidDW2fbuHUOUarQ7pto9WdXrJbt25lYmKCsdExevuyF7AEt8uXLy+2XzQD37Fj\nRzFxSp2haYapFYOEuh0dHcWKQ+U44ogjKsEO073b1OpF5dWqRJOlZ555pljVakI2NDRU5KF0Ozs7\nK3vaacBDrVKFiYmJiiYgPZa6+Fe+L3nJS4BSP7F69eqi3FoNP/roo0W+0lMcddRRldXuggULirb7\n7nczC2utXlSPtM67du2qrLLSSUxaL8H6eUjbZn9gqu0xu3KyGhw74VZfsZ5hoVz92tWb9RqtbUr5\nG9I9EdT+mpDKYaBWlzLjP+ecc4BqsE7pB9Q/tL153nnZGui0004DytWpzldd1R/Teuq51DOrybUm\n+3rmxAhIHC926f777wco4rPpei1yrNO94447DiiZHukm1JetM0MtqtbevbahXHa7Ou2rupfWT4rt\nE0LdWGiNNfaH526V2fpFUT/W/VDf1Bite2192VhjD+uxPb1Gn3UMgBZOdXoSK22w2hBdb8dzGz7F\nWgfpOqvr0va8yqG+nbaTnjU9o/KHo/HYWkRq619pWuZM7WcNVexi0vYl65tGY0PB6LQ1jsVWN5Yy\nTVY/aDVTaldr1KNPyRnUN6yRSR1mO8HZTiY+E2Y1uXE4XgzYE82Zw+FwOPYPZjvBuQN4HfDZEMLL\ngR/s7uQdO3a4n/EXMV7/+tcD8OEPf3g/l2S/YUaaszom0jKVVkyulb9dBVo/F3ZP3v6fQgyNjV9j\nWQgbCVjbrmJgtLpUelo1KmimWBCxHueffz5A4WFcx3We9VytVWmqSdCqWjF65ANDbKrKZFffykPM\ni7aVxR6JcVF6ChOhT7W7mBzpN9RWYli10l7bthaAnuey+6KVrvK3DjmhupJt5vk7hf1exw7uj5AO\nYgKs5YtldqyPGrWHrrN+caz/lZQRsDGnxLDo2RFjqXtZZ5xgLRatvkXXi2mx2ind69RRavq7npM7\n7rgDKNlD2xeh7GchBKBkQbXLoGdN0HNg2U89WzbQrcqmMlstjmW7rMygCA3T2dg26rPWqitlV9J4\nhmneKSsHVe2UYL1kp1q93WG2AoX/BIZCCHeQDfa/Pst0HI4XAxo0Z8C0NWcOh8Ph2D+YrR+cSeCX\n5rgsDseBihlpzqyVk2Vs7ArHrpy0StQKTedZTYfOS88RtFJSGlrxnHnmmUAZe0z7+9ZTscoituhV\nr8r0p2IvpBmTlkS6Cq027b696q6Vm1aDEpxr9ZpqSLQqF7NifYzYdrUedK0nWO3bS4un42KZtKq2\ncedUJ7lPEOsktmn0tKxcYrtUPuktxDSlkZJVFpXZus2wVid2dSzMgdXUnMGGVtH9EuuXGmhA2QfU\nBtbCTxopa4gBJTMiKE/rb0WwjketpZaYA7Ed1heP9VNk09XvNsaVdGOq+z333AOUfVrPI5RM+cMP\nPwyUz5o+9YxYy0bV2XrQFtS+esZt31PZ9bt1gqu20HnbWrNyPL/1+YY2UttprEn1j1aHpnOsE1zl\naRlpjWcqmzSCU2ks3cTE4ZgarjlzOByOFxlmq8FxOA4mzEhzNjk5OW81Zz/5kz+5v4vgcDgc04JP\ncByOqfGfwOW55gwy9/YOh4OqmFjbPdpCkbm+HDdqe8G6Q9D2g93OtKEe0nN0jRXY1gWitNs31hmj\ntk70XVuxSl9bInXbo9ouUhsonIi2Ra+++mqAIvbeZZddVpRF21W33347UG4NaavVOh1UXtaBn8qk\n7eZzzz234bx7770XqHeapzay4mTBOvSzIR+UTrqlaO+H3U7WdqTqoPaWEL1ue65ZXMoUPsFxOKaA\na84cDofjxQef4DgcDodj1tAK34YiEfNgA7lap291AmoJ3G3IgDQty1qIdbDibCtGtR7RxULo0zri\nU7qKmSeWoy7Uhv0ukexrXvMaAK688kqg0bFlnTm4mBYb+NYGuJUoX6yIDXir9hWzkzqNhdJkXd+V\nvmVJ6kI5qFyqa4o6Ri31+A+lUYLuvaB2tKE5CtP1GuzVCc7+dJAWQugA/hE4BugCPgw8BFwPTAD3\nA+/KV+f7ojyHAXcDP5Tnv0/LEUK4hkxH0gH8NZmuZF+XoRX4e+CkPN9fAMb3dTkcDofDMf+xtxmc\n/RmU803Axhjjm0MIS4B7gXuAa2OMt4YQrgOuAm7c2wXJJ1t/CwwALcBH92U5QgiXAhfk96EXeB/w\n+n1ZhhxXAL0xxotDCK8G/pCsD+7ze+JwOOYG0kNoFW7DIEirofAcYkW04tcqXKyKtDaW/Ug1HTYg\nqj5t+BqxR1Zro7La0ADSfCj9s846q+G48pHGRvkoHIIYCbElYn7E1IiNkePLNFSHmJef+qmfAuDm\nm28GSncECkVi6yTWQ+0ttwRy4SBTdelcxIKccsopQHl/xOBYRi4NiQQwMdnoLNGGF1EbpWEvrDsE\nG6JE56rMaj/1GemNVPY03uDusLfNxPeng7TPAr+b/98KjALnxBhvzY99GXj1PirLR4DrgOfy7/u6\nHFcA94UQbgRuAr4AnLsf2mIQWBxCaAEWAyP7qRwOh8PhmOfY2wzOnAXlnClijAMAIYQ+ssnOB4A/\nS07ZSfaS3asIIbyVjEn6ar5N1JL/7ctyLAOOAn4UOJ5skrOvywDZtlg38DCwlGzL7JX7oRwOh2OO\noBW7taayGhFpTLRK12pc11k2RGyHmAFpQ6BkMSwTozQsG2T1PoLVm0j7oeDDYm4U7FFO96xDTOUj\nZ5BKT+yKmAiFatD5KZMhZ5kqg3UoqdAfylvnpe0CJTtkrZ+UlxwCWmd6lrkRrNPJuuPWOWKzcwWx\nTapDGpQayj6iTxv0VIyYtYqz2NsTnP3qIC2EcBTwH8DHY4w3hBD+NPm5D6iqoeYebwMm8y2Zs4B/\nIptw7MtybAIeijGOAY+EEIaA1K3ovmqL9wF3xBjfH0JYCdxCpgna1+XYa5hL3Vm+rfvHMcZXhRBW\nMUut0t7Qo4UQ2oBPkumpJoFfJKvvrNPM050zrVoI4XvAtvzrGuCP9jC9OdWxhRDeArw1/7oAOBO4\nGPjYbNN0OBwl9vYW1R3ADwNMx0HaXCKEsBz4KvC+GOP1+eF7QgiX5P+/Fri12bVziRjjJTHGS2OM\nrwK+D/wc8JV9XI7bgSsBQggrgB7gG/u6LYBeSkZvC9kEe5/fk72MQncG/DaZ7mzGCCG8j2wCoSWe\ndFuvJGPfrppBctKjvZKsH3w8L9ds04OMDZyIMV5Mxo7+4Z6muRut2ozTCyF0A8QYX5X/vX0P07uU\nXMcGXErGhO5RfWOM/6TyAXcBv0q2rb4n92W/Y3JysuGvvb2d9vZ21q9fz/r163n++ecbgqp2dXXR\n1dVVnNfW1kZbWxvDw8MMDw8zODjI4OAgO3fuLP527NjBjh07GB0dZXR0lI6ODjo6Ooo8x8bGGBsb\nY2JigomJCUZGRhgZGSl+1/n2+qOOOoqjjjqqyHvdunWsW7eOzs5OOjs7WbRoEYsWLSrKKCj98fHx\nBqajv7+f/v7+Ir2NGzeyceNGuru76e7uLvIfHR0t6q+81qxZw5o1a9i2bRvbtm0rrhFUltbWVlpb\nW4syKR1bR7WJznv22Wcb/lpaWhr+bHn0vburm+6ubnp6eujp6WHJkiUsWbKEhQsXsnDhwqKc+r2n\np4e+vj76+vro7e2lt7e3KJuOqy6610899RRPPfUU27dvZ/v27UUfUTurTXR/67C3Jzj7MyjntWTb\nHb8bQrglhHAL2UD8wRDCf5O9XD+3D8sjTALv2ZfliDF+kWwicSeZ/uaXgffuyzLk+Ajw8hDCbcA3\ngGuAX9kP5dibmCvd2WNkQnBtJe6JbmvO9Wgxxs8D78y/Hks2Yd1TPdVcatXOBHpCCDeHEL6RL7D2\nJL29pmMLIZwHnBJj/Pu5StPhcOzlLar96SAtxvhu4N1Nfrp0HxelQL5S2y/liDH+VpPD+7oMW4Gf\n2N/l2MuYE91ZjPE/QgjHJodmrZnaW3q0GON4COF6MtbqfwCXzzbNvaBVGwA+EmP8hxDCieSTzj1I\nb2/q2K4FPpj/vz+0cXsEG0RWegrrm0bHrV8cG3BRVlXSiOj31GJGGhelIcshXWsDdNrAkfpUniqj\n8lqzZg1Q6mJkVSVNiPKVFsT6lrFll+ZEGh3ln/p1sUFKpdvZuHFjQ12kW1GdbaBctbOOS8ukNlE+\nul6aKeWrfMREqez6rt9VN6Vnr0v956i99Jv1PK121nnS1ljrN9XFlq0OHmzT4Zhb7C3dWZrGjLVK\nuR7tv4D/HWO8YU/TE2KMbwUCmX+j7uSnmab5NrJwGLcwN1q1R4B/ycv4KPACsHwP0tsEfDXGOBZj\nfAQYonHyMas2DCH0AyfFGL+VH5qT++JwOP5fe+caa0d1nuHn+MSGUmwMiJvBdmwMC1cW1I5LCIkM\noW0aqlZUVSt+RKGgUjUKqvwjEmoRitQoaqpSIgWVVhW5OJXS/AClaaso1OKuEOJLa8tgw/IF+Qq2\nMCU2hwI2tvtj5j0z83nP2WfvfRKjfd5HQnPm9q1vrb2FZ979XVzJ2JippqfGnD2wKaV0U/kP4a0U\nP/FNilo82hdzzk8Paq+0+Xngipzz1yjS/08AG/u1mXNWHBblQ84XgAcG8PEuikDve8q4s9nA2gHs\n/YRCEf56jGPrdw1LVoX7BvpczgSx/kmsHqzz2uotXG/pqngcqwuLWK24PlYkVvnVG35804++SlUS\nUmqU/RN7KMV+S7KjuUiZ0H1SU3bu3Nmwo3o6dZvPPPMMUGVDScHZsGEDAMuWLQOqdYzZa1JyLr74\n4sac5FPsMaU6Q7KjLLaYJRX7P50cKe6XMqTxRSfFTapPrEgs1Uk1e7SNfa5i36xu+AHHmKllqhtz\n6v/qXwIeSSnNArbRW6xSPR5NsTirgYf6tEd5/ZqU0rMUWUWrKdL/+/Uxoli1fu19C/hOSknxLHdR\nqDh92cs5/yiltKqMY5tBEce2ewD/xNVAPctukDkbY2qMxJx3Y4wxZrLMmTPnFFRv23rD19u2lAK9\ndUshUPfsJUuWAJWSIEVASoS29XovMTZDW40d+zGJ2P1bqsb8+fMbvu3btw+Aq666CqhUD6lOioOJ\napVUD81Fio7q32j8O+64A4Arr7xy3Detz9NPFyLrli2F+KteUvPmzQOqWjyaq85rDJ1ftGhR43zs\nu6V91SeSiiKVReseY3M2n70ZgGVjhZIU6x3Jbj2bLCpeirnROj311FNApdzEuB59h2JPK63n0aNH\nO0p6jsExxhhjzNDhn6iMMcb0jd7Uo1Kj41I/9Latt3bFlixYsAA4XanRW7rUEikT9bH0pi8VRbcl\nVAAAE4tJREFUQVV6pbRI5ZASoOtjXJCukx1dJzvaykcpQLpPSo3u0xrE7C1lUT3+eJHUp75cUCkl\n2spmt2q94qKLLmr4qEylWFlaa6e4I62Z1lu+xr5asZKxVLK2z6J+fcxai1lvUmakHmkNtH4xA6yt\nS3zECo4xxhhjhg4rOMYYY/qmLT4iZuuIqOQo9kPdsvW2ruuk4NSr+CqDSDakNrT1WYr9kmJMiHzQ\nvsaWwqCxpUBIadD1sXaProsxQRp369atAKxfv37cR63fDTfcAJyeBSXqFZTrvirrKsbAxA7r8qmt\nn5eIGWi6X5+H/NPaa46inukWY6JkU/dq/aVWxQyumDUX47vasIJjjDHGmKHDCo4xxpi+aevUHc9L\nPZFSoFozO3bsaNwvFeTSSy8Fqrf8PXv2jNtUZpDe+GO2VIyBETFeJGZfCWVTxRieeF1UU2Q/xvwo\nnkZxRKpSrEwpgKVLlwKV4qJ56x7NRWqR9hXHE6v/SvWINXykcklVkZ2oamkNxLi9WWVs1Mg5jTnG\nta7XFpJN+aLPS7E48il+LrouxkxNNvvbCo4xxhhjhg4rOMYYY/omViCOcSxCCkBUQbZt2wZUSs7K\nlUV/WtWgUZyNVBs4vWqv0H7MtpFvscJxzBhSLIjUixgzon3FAEX1Q3FHMQZICo7Ox/o9ANdccw1Q\nxbhIAYmKl1QQ3Sub3eJSpPBIudFayL7mEOvaaE1k9+jJQm05cvxI4/qojnVSWTT27t27Adi/f39j\nLK17jGXSnEWskt2GFRxjjDHGDB1WcIwxxkwZUUlQTEkbUiDOP/98oFIwFGsSu5ND1cNJ94oYBxKz\noaKyEtWP2ENKvmsb+2nJnvalhkhhmDt3LlApFFJJFFezfPnycV/UsTzWgIlZatrXdVHRkVITVSfF\nuWg9Y92h2PFbdqTI6PM5NVaswaFDhwC4/PLLG2sZe40BHDx4EIC9e/cClRoVlRj5FhW4GB802Vgc\nKzjGGGOMGTqs4BhjzDQlpTQT+DawEDgL+CrwMrAGOAm8BNyTc55000K9VccKxjEjJsbkSD1RbyTF\nrSibqq4ISH2I9VL0hh9jOKTAxBoxMYZHKoeqAMu+7Ol6KTCKKZGiJJ91vXouyXfNUX5KtakfUzyQ\nbGi9YmdzKS6xxk7MRNLYsWO6rhOxB5jQOOPq1JFi7VSJWllcsqc5qp8XVH2yYu2dmF2l74zWX3OK\nn0OsON2GFRxjjJm+fA54I+e8Cvgs8DDwIHBfeWwEuO0M+mdM31jBMcaY6cujwGPl3zOA48CKnPNz\n5bEfA58BfjhZg1Ic9OYfFQedV+xHrGysbCopBqpwPGfOnPExYq8pKQKxH1I8L5u6Xz5JOZCvUXGR\nr1IQdH3MQIqZRLHOi+6PdXvq52Ldmlg/SNuo+Ch+JfakkiIWlTOh47Gmj/zRWox85SsAXPLx4vpP\nrCvt/PSnAGy+rXgO3rVrF9CsWxQz56LKpPWTihT7YsW4o24VjIUfcIwxZpqSc34HIKU0m+Jh537g\n72uXjAHnnQHXjBkYP+AYY8w0JqU0H/gB8HDO+fsppb+rnZ4N/LwXe7GycdvbeIzD0Fu8lIicM1Bl\nVakuDlQxGjGuJ9ajiVk3sdpyrLMSs7I0juqzyJ6UGcUHyY/YuTt2WlesjmJvNLf6+mgO9XNQxfcc\nPnwYqBQvHY+KmXzQHKKCE9UuzTF2Su+kNnVCMTevvfbaaffFOkRx7DhGrPEjFanXisaOwTHGmGlK\nSukSYC1wb855TXl4U0rppvLvW4HnOt1rzIedkcn2dDDGGDNcpJS+AfwxkGuHVwMPAbOAbcCfTZRF\nNfrJ0ca5WAlX2xg3cVpW1YyyyvBoWb/lRHH+3F8tFIrFixeP3xvrpOjeGSMzGuflS3zjP3GyuO/E\nB80u47EHkrazzipicGJNn1kzyxoyp4q5vffue83zZezOubPPbRynXLFGXZ4ypGV8/qHujdQMKSSq\n7KwYpdi9XXMce3ussRYzRss1GmmuUaxIHMedu24dALsXFs5/dE8zlmfd+XM7jgftXcFjFlTMBNN3\n4IPj7VWaRzeMcuzYsZHTTuCfqIwxZtqSc15N8UATufmX7IoxU44fcIwxxvTN6IZm9k3sYq2t3s7b\nelfFmI/Rk8XxWeeVXbNHLhkfY8GCBUCVdRNjONrifdpiN2Kdm1hDRnZUzVdxL7xP4/qjbze7Yn/k\n/WLOs5ndGFdxRqp03FiHd5sZR7Emz6LRRQBs2r8JgDlHi+yy6667rnG9tlH1iDWD2pByc+E5Fxbj\nrFvfOD+eRVWyrgxF/8h7zc8ZYORUUyWKyk38rozHMn1QXn+iVNhGXMnYGGOMMdMcKzjGGGOmjLbK\nxbE+js4ro0gKj3oZRRWlXhlX1XOlMsQ6KlGdiPElUot0n4iVj9u6iMf4Im1jrRrNWVlXsVdWnViF\nWVlNQuskFUlZVJpzrBXTNseoesinqIJJTZHa9PbddwPwv2dvBmD74iKrbcOGDQAc2bmzYa+esRYz\n6TRWzBSL1Ztj5eI4Bys4xhhjjJl2OIvKGGNM38yaNesUnP6WLfRvjN7eY5xF7FkVa9LounoGkxSc\nyy67DKh6Qyk2RkpLVJNi3ZWoGMjH6GvssC0VRSqL7Gl8qSZSM3Sd+jfJTl3R0Rw0X201V1VylnJz\n4MCBxtgaU7a1BrG6stBa6L6YcSY7MTvrZ/ysuO6F4jpldU3UJ0qKVlSVZFsqVJtyE6tgR44fP97x\ny2cFxxhjjDFDh2NwjDHG9E1bfETbddpKBYkxJFIeoqJQ74atuBCpB0uWLAFg+fLlQKWMSDGIlYxj\nllVUj3RdVHakPEhdib2opDyoh5XOK1tK2V979+49bX2iGqSeUlJ2Dh48CMChQ4caPsYsKc19vPt3\nqEETe4DFHli6XoqZ7IzH4hwpsrPGDob6OkEhqu/r3qjcxCy1WGk6fpfid6IbVnCMMcYYM3RYwTHG\nGNM3bV2qRXzrjspDzBaSohBVlLpSIdVHCsArr7wCVL2eli1bBlS1YISUBBHjS2IlXW2jwhO7lUvl\nkF/yXXPZvXs3UPWuUuxQPa5IY0nlkdKyZcuWxlxj/JDuky9x/WIWlY5rGxUfjSPftJWCJIUt9vGK\n9Y860fZZtyky8XivMcNWcIwxxhgzdFjBMcYYMzBt8RJCKomOK04lqiaxA3g8D1UMjGxKOVm/vqi2\nK+VmxYoVHW1IfYg+takQul5qh3yM6pTUjlhDRqrInj17AFi4cCEAKaXxMbQe8l3KjXpOxTGVVaU5\ntCk8QvfpuHyVHd0XM8hUw2fHjh0AvH9+WWWYZkaa1JmoXkFVw0fn9Pn12h3cMTjGGGOMmfZYwTHG\nGDNltCk5bV2r25SGGCMy0Vi6VkrLSy+9BFRqyPXXXw9U2UxSS6QktNVlEbE7ufZjD6vYX0sxQfJD\nKsv+/fsBuPDCC8fH0DVbt24Fqiq/mn+MlYnZUzGTK663iLWBtI1VmI8eLfpqbdu2DahicCiS3077\nXDSu/JBqUx8jxkDF+K22/X6xgmOMMcaYocMPOMYYY4wZOvwTlTHGmCmj7aeRtp8d2oKR4379vvhT\nhgJio22lj6u9wY033ghUxfP0M0psg6CigrERpX4Ki+0lYtCyfvpSsT61cNDPP/HnHIDXX38dgO3b\ntwNw++23A9XPZQo2lo0333wTgAsuuAA4Pcg4rmtspinki9ZCc9+8uWiq+eqrrzbm1PYzkn6W009r\n8geqn6bafOiWDh5/1pzsT1hWcIwxxhgzdFjBMcYY0zfdyul3e9vu9lbeqVlkTCuOzS0VIKt9qR9P\nPPEEUBUCXLlyJVCpD1JLpNDEooS6Tvsx0Fb7uk+BwwomloIxf/58AK699trxe9W+QenYShO/5ZZb\nGjalMqlYoFDgdNs6av00N6HU9Y0bNzbsS1HSmsTUefmj9HCdj0UY6/OOLTJEt2Djbte3YQXHGGOM\nMUOHFRxjjDFTxmRTe9tidSKxZUP9byk3siH1ICopOi9lR2qF9tX4U3EoKsSnuBMpMVFhUOyNlB35\no+NvvPEGUMXLyJ5igHbt2jU+pxdeeAGolJgXX3wRqFQeFeaTGiXFREqLlBnF6MhnrYnGlv3Dhw8D\nVfNO+SiVRbE0GjcqbbGoocaJSlKdXlsvdFMBHYNjjDHGmGnHyKCFdIwxxkxfZs6c2fEfkW7KTLdC\ndKJTDE58o4+F+mJsTCwwJ6Q6yM68efOAqoWC9qNaEVsxxCyqnTt3AlVxPMWpKJtKMTm6D2DTpk1A\nFa8SfVm8eHHD91jkUMqNYnmUQXbgwAGgUnRiVprWTHNS7Exc2/HxfqOMvXmxmJMUG9mVP52IsTST\njdfqdvzYsWMdv2xWcIwxxhgzdFjBMcYY0zeTVXC6Zb50y7aq39fNthQcqRJSIaSOSKVQ7IxUjLGx\nMaBSi6644gqgUnAUq7No0aLG/aodo/iYWH9HLRvkR6xZU/9bW81Bqocyv5YuXdrwUSpSzhmAl19+\nubE2sXlppE1Jiw1Kdf//XVuMN7Ku2Wwz2otqTf1Y23eg3+eR48ePW8ExxhhjzPTAWVTGGGOmnG5v\n41Gx6VbhuJO9WJ9GSCGJdWykikiRieqE7Gi7b98+oIrVUSyNmmXqOo2nreJdpG5IyRFSR+oqh+Jx\nonIj35TtpDkp9uWtt94Cqoywtlozoh7LVLcX1Sxt49zGm3kG0SQ2Iq2P30296/bZ94sVHGOMMcYM\nHY7BMcYY0zdtMTiiLVOmW8yNlABdN1F2Trcxo2ISe1hJNZGKEXtPRaVH6omu032xZsy42hH8iSpK\nJ187XQOVsiLb0Tft67zmKB90f+xZFVUw1c2Jyo2yqEY3NKs6n5ZtVfu82r4Dce7x+kjbd8UxOMYY\nY4yZNjgGxxhjzC+Mtrfubr8eRLVlosq4bd2mRZv6I5VCaoZUCKkZUT2KVX6FrpcSJL9iDZ+YcVRf\ng3itlBftyweNHVUiKS1SaHRfnJOOa+6yo21Ur6LCc4JmZem49hPF20yVcjPZGB0rOMYYY4wZOqzg\nGGOM+aXRrR7LZOJC2+J44pt9vC4qMSIqPMpoiplBUQ2JSE2JakmM+emkKMV6NXEdNLYyujRWN5VI\nyk5UbCaTpTaRP3Htu9W46US3a/tV/4QfcIwxxjRIKc0A/hG4FngfuDvnvGviu4z5cOEHHGOMMZE/\nAGblnG9MKX0ceLA81kovCkyn+yZ7vNMY3cZu63XULVYnKj9RHZFqEpWatho/k+mgHmNi2uKBos02\n36MPMeuqTaGJKlXb9VOZiT3VWd2OwTHGGBP5JPA4QM55HbDyzLpjTO9YwTHGGBOZAxyt7Z9IKc3I\nOXcuj1ujVyVnsqrLRNd0q6/SZqNTp/I6bapIvL9X5aFTtd82X2K9mkj0ISo00fdY4TiON9k6RW1M\ndH2vn8+gWMExxhgTOQrMru1P6uHGmA8TVnCMMcZEngd+H3g0pXQDsKXtwrYqssacafyAY4wxJvJv\nwG+nlJ4v9+86k84Y0w/uRWWMMcaYocMxOMYYY4wZOvyAY4wxxpihww84xhhjjBk6/IBjjDHGmKHD\nDzjGGGOMGTqcJm6MMaYvprIpZ9nz6m9zzp9OKS0B1gAngZeAe3LOk075TSnNBL4NLATOAr4KvNyv\nzZTSKPAIcDVwCvgCxXz79rG0ezHw38BvlnYGtfc/wJFy91Xga4PYTCn9FUU9pJnAP1DUR+rLXkrp\nT4A7y91fAa4DPgV8o1//umEFxxhjTL+MN+UE/pKiKWfPpJTupXiAOKs89HXgvpzzKmAEuK1Hk58D\n3ijv/yzwcOlbvzZ/DziZc/4UcD/wNwPa00PYPwPvlPcPNOeU0tkAOedPl//96SA2U0o3A58oP9ub\ngcUMMOec83flG7AR+Avgy/3amwx+wDHGGNMvU9WUcyfwhxT/yAGsyDk/V/79Y+C3erT3KMU/nlD8\nO3d8EJs5538H/rzc/SjwFvCxAX18APgn4PVyf9A5Xweck1L6r5TSk2UF6kFsfgZ4MaX0Q+A/gf9g\n8DmTUloJ/FrO+ZtTYW8i/IBjjDGmXzo25ezVSM75B8AHtUP19g9jwHk92nsn5zyWUppN8bBzP81/\n7/qxeSKltIbiJ5XvDeJjSulOCoVpbXloZBB7Je8AD+Scf4fiJ7TvhfO92rwI+BjwR6W9f50CHwHu\nA/66/Hsq7LXiBxxjjDH98otqylm3MRv4ea8GUkrzgaeAf8k5f38qbOac7wQS8E3g7AHs3UXRCuNp\n4NeB71I8UAzi33bKh5qc8w7gTeCSAWweBtbmnD/IOW8H3qP5ANKzjymlucDVOedny0MDfyYT4Qcc\nY4wx/fI88LsA3Zpy9simlNJN5d+3As9NdHEkpXQJsBa4N+e8ZlCbKaXPlwG3AO8CJ4CN/drLOd+U\nc765jEfZDNwBPD7InCkemh4s/Z1H8cCwdgCbP6GIX5K9c4AnB/RxFfBkbX+gz7kbzqIyxhjTL1Pd\nlFMZNF8CHkkpzQK2AY/1aOc+CrXhyyklxeKsBh7q0+ZjwJqU0rMUGUWrgVcG9LHOKQaf87eA76SU\n9JBwF4WK05fNnPOPUkqrUkrrKcSQLwK7B/TxaqCeZTfonCfEzTaNMcYYM3T4JypjjDHGDB1+wDHG\nGGPM0OEHHGOMMcYMHX7AMcYYY8zQ4QccY4wxxgwdfsAxxhhjzNDhBxxjjDHGDB3/D2gDw8MSpBFv\nAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x154d7050>"
       ]
      }
     ],
     "prompt_number": 2
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Detrending"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Once images and ROIs are ready, we can inspect brain responses in them. We extract functional data for each ROI but before we look into the signal, we need to detrend it.\n",
      "\n",
      "Throughout a scan session, signal intensity tends to increase for a variety of reasons (and that is one reason why you should always use palindromic stimuli sequences in blocked designs). This increase is not meaningful at all so we need to detrend it -- that is, find this global tendency of increasing responses and subtract it. By default, detrending in `psychopy_ext` module is done using a second order polynomial for each run separately."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Timecourse of the signal"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We are finally ready for having the first peak into the data by ploting timecourse of the signal. In `psychopy_ext`, you need to specify in the `rp` parameter that you want to visualize data which will plot the full timecourse of the first run, as shown in the example below:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'},\n",
      "                     rp={'values':'raw', 'method':None, 'visualize':True},\n",
      "                     rois='rh_LO', dur=4, offset=0)\n",
      "df, df_fname, loaded = mvpa.run()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "subj_01"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " rh_LO"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAEZCAYAAAAt5touAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGVKSpGoZUpKkav0bXRn+gwgnmscAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15598370>"
       ]
      },
      {
       "ename": "SystemExit",
       "evalue": "",
       "output_type": "pyerr",
       "traceback": [
        "An exception has occurred, use %tb to see the full traceback.\n",
        "\u001b[1;31mSystemExit\u001b[0m\n"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stderr",
       "text": [
        "To exit: use 'exit', 'quit', or Ctrl-D.\n"
       ]
      }
     ],
     "prompt_number": 3
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "You can also plot the average signal timecourse over the entire experiment for each condition:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'},\n",
      "                     rp={'values':'raw', 'method':'timecourse', 'force':'all'},\n",
      "                     rois='rh_LO', dur=8, offset=0)\n",
      "df, df_fname, loaded = mvpa.run()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " subj_01"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " rh_LO"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAEZCAYAAAAt5touAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY0lICSGEyFgSUkIIITKWhJQQQoiM9f8BMvhYXv7LzYEAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15b86070>"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Correlational analyses"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The most straightforward way to run MVPA is to compute a correlation between two patterns of response within an ROI. In particular, you can split your data in half and compute pairwise correlations between all conditions. If this correlation between the same condition is higher than the correlation between different conditions, then we say that information about conditions is available to that brain area. (For a very nice and simple study using this analysis, see [Williams et al., 2007](http://dx.doi.org/10.1038/nn1900).)\n",
      "\n",
      "Let's go over the steps implementing this analysis in practice.\n",
      "\n",
      "Let's plot responses of each voxel in an ROI (columns) for the whole run (rows) to get a feeling of what they are like:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from psychopy_ext import exp, fmri\n",
      "import matplotlib.pyplot as plt\n",
      "import seaborn as sns\n",
      "\n",
      "sns.set(style='dark')\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'},\n",
      "                     rp={'values':'raw', 'method':None, 'force':'all'},\n",
      "                     rois='rh_LO', dur=2, offset=2)\n",
      "ds = mvpa.extract_samples('subj_01', 'main', ['rh_LO.nii', 'rh_LO', '*rh_LO.nii'],\n",
      "                          values='raw', offset=2)\n",
      "ds = mvpa.detrend(ds)\n",
      "evds = mvpa.ds2evds(ds, dur=2)\n",
      "mvpa.plot_ds(ds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAFtCAYAAAAeUTF6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nXBAEIaPIBC4IgpBR/n+3z10YtbASkAAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x155825b0>"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Normalization and averaging"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Before we use these data, it is important to normalize neural responses. It is possible that for one condition responses are generally higher than to another, thus we might be leveraging these global biases. Typically, data is normalized by subtracting the mean across conditions per run per voxel.\n",
      "\n",
      "Moreover, we should average data across trials."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import mvpa2.suite\n",
      "import numpy as np\n",
      "\n",
      "# leave fixation out\n",
      "evds = evds[evds.sa.targets != mvpa.fix]\n",
      "\n",
      "# calculate the mean per target per chunk (across trials)\n",
      "run_averager = mvpa2.suite.mean_group_sample(['targets','chunks'])\n",
      "evds_avg = evds.get_mapped(run_averager)\n",
      "numt = len(evds_avg.UT)\n",
      "\n",
      "# calculate mean across conditions per chunk per voxel\n",
      "target_averager = mvpa2.suite.mean_group_sample(['chunks'])\n",
      "mean = evds_avg.get_mapped(target_averager)\n",
      "# subtract the mean chunk-wise\n",
      "evds_avg.samples -= np.repeat(mean, numt, 0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Training and testing sets"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Alright, so we're finally ready to do some real MVPA! We are now going to split data in half and correlate one half with another. Remember: if there is no information about conditions in the response patterns, then it is effectively random and everything will correlate with everything to the same extent. However, if there is some information, we should see a higher correlation (lower dissimilarity) between patterns of response corresponding to the same condition than between patterns of response corresponding to different conditions."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "targets = evds_avg.UT\n",
      "\n",
      "# split 1\n",
      "evds_split1 = evds_avg[:4]\n",
      "run_averager = mvpa2.suite.mean_group_sample(['targets'])\n",
      "evds_split1 = evds_split1.get_mapped(run_averager)\n",
      "\n",
      "# split 2\n",
      "evds_split2 = evds_avg[4:]\n",
      "run_averager = mvpa2.suite.mean_group_sample(['targets'])\n",
      "evds_split2 = evds_split2.get_mapped(run_averager)\n",
      "\n",
      "result = mvpa2.clfs.distance.one_minus_correlation(evds_split1.samples,\n",
      "    evds_split2.samples) / 2\n",
      "\n",
      "plt.imshow(result, cmap='coolwarm', interpolation='none')\n",
      "plt.colorbar()\n",
      "plt.show()\n",
      "\n",
      "# plot averages for within-condition and across-conditions\n",
      "within = result.trace() / result.shape[0]\n",
      "across = (np.sum(result) - result.trace()) / (result.size - result.shape[0])\n",
      "ax = plt.subplot(111)\n",
      "ax.bar(0, within, label='within')\n",
      "ax.bar(1, across, label='across')\n",
      "ax.legend()\n",
      "plt.show()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAZoAAAFVCAYAAADWuCkCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x157484d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFVCAYAAADVDycqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSVIyxlmSpGSMsyRJyRhnSZKS+X+XBs9I4+PpegAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15752b30>"
       ]
      }
     ],
     "prompt_number": 7
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Cross-validation"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "So there seems to be something in the pattern of response. Before we celebrate, there is an extra step recommended here. Did you notice how we split the data? We put the first four runs in `split1` and the remaining for in `split2`. Why split like that? Couldn't we have chosen some other kind of split?\n",
      "\n",
      "We could and, in fact, we should. Because if we don't, how can we know that the result is not spurious, that is, that the observed difference is not due the particular split that we performed? Maybe another split would not yield the same result? And, on the other hand, we might completely miss the effect because just by accident it was not present in the particular split that we chose. Of course, in practice we have several participants so if the effect is stable across participants, we are doing fine. But chance are higher to observe the effect if we average outcomes of several splitting procedures. In `psychopy_ext`, by default random splitting is performed 100 times. We will refer to this procedure as *cross-validation* though the true meaning of this term will become clear only when we consider support vector machines."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'},\n",
      "                     rp={'values':'raw', 'method':'corr', 'plot':True},\n",
      "                     rois='rh_LO', dur=2, offset=2)\n",
      "df, df_fname, loaded = mvpa.run()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "subj_01"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Could not load or find the results file ./analysis/corr_raw_subj_01.pkl\n",
        "Will proceed to do corr analysis from scratch\n",
        "rh_LO"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stderr",
       "text": [
        "C:\\Python27\\lib\\site-packages\\numpy\\core\\_methods.py:79: RuntimeWarning: Degrees of freedom <= 0 for slice\n",
        "  warnings.warn(\"Degrees of freedom <= 0 for slice\", RuntimeWarning)\n",
        "C:\\Python27\\lib\\site-packages\\matplotlib\\figure.py:1595: UserWarning: This figure includes Axes that are not compatible with tight_layout, so its results might be incorrect.\n",
        "  warnings.warn(\"This figure includes Axes that are not \"\n"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAEYCAYAAADmugmLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhpQkKVv/A9QuURodBXhdAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x158308b0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAVAAAAEYCAYAAAAK467YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15571c90>"
       ]
      }
     ],
     "prompt_number": 8
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Support vector machines"
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Introduction"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Correlational analyses are simple and fast but they might have issues dealing with noisy fMRI data. In the analysis we just did, it is really not so clear it there is anything or not. This is the major problem of interest in machine learning so here we will employ one of their most robust and powerful techniques, known as support vector machines (SVM). *(Note that in practice SVMs are not guarranteed to yield a better result that correlations.)*"
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Setup"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Modified from PyMVPA example http://www.pymvpa.org/examples/pylab_2d.html\n",
      "and http://www.pymvpa.org/examples/pylab_2d.html\n",
      "\"\"\"\n",
      "import mvpa2.suite\n",
      "import numpy as np\n",
      "import matplotlib.pyplot as plt\n",
      "import seaborn\n",
      "%matplotlib inline\n",
      "\n",
      "np.random.seed(3)\n",
      "\n",
      "##############\n",
      "# Parameters #\n",
      "##############\n",
      "npoints = 200\n",
      "xymax = 9  # Absolute max value allowed. Just to assure proper plots\n",
      "ngrid = 101  # grid for evaluting performance\n",
      "\n",
      "# set grid\n",
      "x1 = np.linspace(0, xymax, ngrid)\n",
      "x2 = np.linspace(0, xymax, ngrid)\n",
      "x,y = np.meshgrid(x1, x2)\n",
      "feat_grid = np.array((np.ravel(x), np.ravel(y)))\n",
      "\n",
      "def train_svm(ds, cl= mvpa2.suite.LinearNuSVMC):\n",
      "    # define classifier\n",
      "    clf = cl(probability=1, enable_ca=['probabilities'])\n",
      "    # enable saving of the estimates used for the prediction\n",
      "    clf.ca.enable('estimates')\n",
      "    # train with the known points\n",
      "    clf.train(ds)\n",
      "    # run the predictions on the test values\n",
      "    pre = clf.predict(feat_grid.T)\n",
      "    # get probabilities from the svm\n",
      "    res = np.asarray([(q[1][1] - q[1][0] + 1) / 2\n",
      "                for q in clf.ca.probabilities])\n",
      "    # get support vectors\n",
      "    supvecs = clf.model.get_sv()\n",
      "    \n",
      "    return res, supvecs\n",
      "\n",
      "def plot_data(ds, supvecs=None, contour=None):\n",
      "    ax = plt.subplot(111)\n",
      "    ax.set_aspect('equal')\n",
      "    \n",
      "    # plot samples\n",
      "    ax.plot(ds.samples[ds.targets == 1, 0], ds.samples[ds.targets == 1, 1], '.',\n",
      "            color='indianred', label='condition 1')\n",
      "    ax.plot(ds.samples[ds.targets == 0, 0], ds.samples[ds.targets == 0, 1], '.',\n",
      "            color='steelblue', label='condition 2')    \n",
      "    \n",
      "    # plot support vectors as larger dots\n",
      "    if supvecs is not None:\n",
      "        feat_pos = ds.samples[ds.targets == 1]\n",
      "        feat_neg = ds.samples[ds.targets == 0]\n",
      "        pos_test = np.array([f for f in feat_pos if f in supvecs]).T\n",
      "        neg_test = np.array([f for f in feat_neg if f in supvecs]).T\n",
      "        ax.plot(pos_test[0], pos_test[1],\n",
      "                'o', color='indianred')\n",
      "        ax.plot(neg_test[0], neg_test[1],\n",
      "                'o', color='steelblue')\n",
      "    \n",
      "    # plot probability contour\n",
      "    if contour is not None:\n",
      "        # plot decision surfaces at few levels to emphasize the topology\n",
      "        z = np.asarray(contour).reshape((ngrid, ngrid))\n",
      "        \n",
      "        ax.contour(x, y, z, [0.1, 0.4, 0.5, 0.6, 0.9],\n",
      "                   linestyles=['dotted', 'dashed', 'solid', 'dashed', 'dotted'],\n",
      "                   linewidths=1, colors='black', hold=True)\n",
      "        \n",
      "    ax.set_xlabel('response in voxel 1')\n",
      "    ax.set_ylabel('response in voxel 2')\n",
      "    ax.legend()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 9
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Data separation problem"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The basic idea here is that for each condition we have the data as *n*-dimentional vectors where *n* is the number of voxels and we want to know if those vectors are somehow separable. In other words, we want to find a particular partitioning rule that would allow us to predict with confidence from which condition new data points would be. If we can do so, we can claim that in that there is information about those conditions in the given ROI. \n",
      "\n",
      "To illustrate this, let's consider responses from just two voxels (so that's a 2-dimentional space that we can plot). The responses are, of course, all over the place but it is not so difficult to decide which response belongs to which category. What do you think the *optimal* separation would be?"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#################\n",
      "# Generate data #\n",
      "#################\n",
      "dist = 1  # distance between the two distributions\n",
      "\n",
      "def gen_data(dist):\n",
      "    feat = mvpa2.suite.pure_multivariate_signal(30, 2)\n",
      "    feat = np.random.randn(2, npoints) + dist + xymax / 2.\n",
      "    feat = feat.clip(0, xymax)\n",
      "    return feat\n",
      "\n",
      "# two categories of samples\n",
      "feat_pos = gen_data(dist)\n",
      "feat_neg = gen_data(-dist)\n",
      "\n",
      "# create the pymvpa dataset from the labeled features\n",
      "pat_pos = mvpa2.suite.dataset_wizard(samples=feat_pos.T, targets=1)\n",
      "pat_neg = mvpa2.suite.dataset_wizard(samples=feat_neg.T, targets=0)\n",
      "ds = mvpa2.suite.vstack((pat_pos, pat_neg))\n",
      "\n",
      "# plot data\n",
      "plot_data(ds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAERCAYAAACD2umRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x163ea270>"
       ]
      }
     ],
     "prompt_number": 10
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Here is what SVM thinks an optimal separation is (look at the black solid line; other lines indicate decreasing probabilities of where the boundary could be):"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# train SVM and plot\n",
      "res, supvecs = train_svm(ds)\n",
      "plot_data(ds, supvecs=supvecs, contour=res)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAERCAYAAACD2umRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCsdxcvI/hCIyKJxwR38AAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x154d7350>"
       ]
      }
     ],
     "prompt_number": 11
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Well, that looks pretty good. I mean, it could have provided many other solutions where this separation line could be, such as slightly rotated lines. But in fact, SVM is guarranteed to find the optimal separation boundary and this is why this technique is so great.\n",
      "\n",
      "It is also fast. One of the reasons it is fast is that it detects the so-called *support vectors*, i.e., the data points that are close to the separation boundary, and uses only them to make it decision. Everything else doesn't matter so we can use even very large datasets.\n",
      "\n",
      "It is also important to understand that SVMs are not magic that can do everything by themselves. They have certain parameters that a user can manipulate in order to yield better results. By default, in `PyMVPA` these parameters are set to some hopefully reasonable values and we do not manipulate them. It might be better to perform a grid-search for best parameters but for some reason this is not done in practice."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Is that the *true* separation boundary?"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Of course not. This is the *optimal* boundary given the data. The *true* boundary, in this case, should be a line at a 45 deg angle. But SVM cannot know the underlying distribution absolutely from just several hundred of samples. Need more precision? Get more data."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "SVM on our data"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "OK, that's all you realy need to know about SVMs. Time to test them on our data!\n",
      "\n",
      "Procedure:\n",
      "\n",
      "- **Normalize** data by subtracting the mean across voxels per chunk per condition (target).\n",
      "- **Split data** into a training set (about 75% of all values) and a testing set (about 25% of values), unless there are only two runs, in which case it is 50% training and 50% testing.\n",
      "- For each pair of conditions, train the classifier (this is a binary classification; multiclass SVM is also available but I don't use them).\n",
      "- Then test on the average of the testing set, i.e., only on two samples. This trick usually boosts the performance (credit: Hans P. Op de Beeck)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "import numpy as np\n",
      "np.random.seed(3)\n",
      "from psychopy_ext import exp, fmri\n",
      "PATHS = exp.set_paths(fmri_rel='%s')\n",
      "\n",
      "mvpa = fmri.Analysis(PATHS, 2, info={'subjid':'subj_01', 'runtype':'main'},\n",
      "                     rp={'values':'raw', 'method':'svm', 'plot':True},\n",
      "                     rois='rh_LO', dur=2, offset=2)\n",
      "df, df_fname, loaded = mvpa.run()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "subj_01"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Could not load or find the results file ./analysis/svm_raw_subj_01.pkl\n",
        "Will proceed to do svm analysis from scratch\n",
        "rh_LO"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 0 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 1 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 2 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 3 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 4 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 5 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 6 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 7 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 8 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 9 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 10 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 11 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 12 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 13 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 14 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 15 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 16 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 17 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 18 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 19 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 20 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 21 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 22 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 23 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 24 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 25 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 26 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 27 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 28 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 29 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 30 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 31 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 32 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 33 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 34 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 35 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 36 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 37 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 38 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 39 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 40 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 41 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 42 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 43 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 44 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 45 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 46 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 47 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 48 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 49 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 50 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 51 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 52 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 53 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 54 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 55 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 56 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 57 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 58 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 59 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 60 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 61 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 62 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 63 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 64 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 65 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 66 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 67 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 68 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 69 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 70 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 71 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 72 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 73 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 74 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 75 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 76 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 77 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 78 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 79 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 80 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 81 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 82 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 83 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 84 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 85 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 86 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 87 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 88 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 89 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 90 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 91 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 92 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 93 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 94 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 95 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 96 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 97 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 98 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\r",
        " 99 %"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAEYCAYAAADmugmLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAFwBJREFUeJzt3XuU1XW9//HnDBdlABFkFKSQS/ZWF+rSsDwgqCEaXjmm\nx8yyJLIyPZX3bqZietSl2fGnp7Qs0czyklkk9itNUzucqEzt8tZQvIB4VAwIFGOY88fe6HYchq3x\nnflO83ysxdr7e9mf73u7Fr74fL6f/fk2tLa2IklSGTV2dQGSJK2PISVJKi1DSpJUWoaUJKm0DClJ\nUmkZUpKk0urd1QVIXS0idgfOBbag8g+3J4GTM/OPEXEncHtm/kebz5wETAb+HXgM+GVm7tnmnG8B\nHwKGZubSDdTwN2CHzHxiI32tDq8DbAmclpmHF3k96R9lT0o9WkRsAvwYODEzd87MHYHvALdFRCPw\n/4Bj2vnoTODS6vuXgG0jYmRNu/2BPYB6f4jYWT9YbAXIzPkGlLoDe1Lq6ZqAQcDAdTsy8zsRsQzo\nBfwQ+GpE7JGZ9wBExJ5AQ2b+LCJGAS3A94CjgPOqzRwK3AKc1N5FI2ISlZBbC8yn5h+MEXEQ8Hmg\nL7CKSq/uvyOiN3ABcACwBrgPOK76sYuBd1drmQd8JjP/1uY6v1l3nYjYC7g0M3eMiG8Dy4AdgbcC\nfwbel5krI2J/4D+q7d4P7ANMLLrHJ61jT0o9Wma+AJwKzI2IBRExOyKOAX6emX/PzDXAFcBHaj52\nLHBZm6auAT5Qs3008O32rhkRfYEbqATJrsBPgX7VY9sCXwamVY99DLg5IpqoBNKuwE7AOCrBegTw\nBWBYdf/OVP5eXxgRfdpc5/Z112nHrsB+wPbA1sDhEbEFMBs4KjN3Ae4ERqzn81IhDCn1eJn5FSr3\naP4deBo4DfhdRGxWPeUK4JCI6B8RQ4B9aRNAmflbYG1E7BoRbwUGZuYf1nPJHYGXM/PO6mdvBF6o\nHpsKDAfuiIjfAddS6cW8DZgCzM7M1ZnZmpnvy8xrgfcAX8vMlsxspdJzmraB69RqBebWhPKDwBAq\n99z+mJkPVj8/G1i+wf+g0kbkcJ96tIiYCEzIzAuBOcCciPgclf9R7wPcnJlLIuL/A0cCA4AbMnNF\nO82t6009S6UHsj6tQEObfWuq+xqp9OLeV1PjSOCp6jm1tQ+l8ne4sU17vYA+HVynPS+1U9/f2/n8\n2vV8XiqEPSn1dM8Cn4+IyTX7RgD9qQTVOpdTCaCjef1Q3zrXAv9GZQjuug6u+SDQEBHTAKr3fZqp\nhMMdwL4REdVj76FyL2hT4GfA+yOib3VSx9epBOftwMcjond1/yepDCGu7zpttQ0iqrXcC7w9Inas\nfv69wOZ03iQPyZBSz5aZDwPTgVkR8VhE/AG4HvhoZj5Sc95dVIbAlrUzjLduxtxi4I/Aw5n519pj\nba7595pr/g44DHimeuyPVO55XR8R9wOzgIMycxWVUPpN9c8DwGLgq8A5wBIqYfZHKj2pT1WH7tq9\nTpvaWtdT5wtUQnB2RPyGyjDnGiqTOaRO0eCjOiS1JyIGUpmUcWZmvhgRuwI/ykwnT6jTeE9KUrsy\nc0VEvAz8OiL+TuUe1b91cVnqYexJSZJKy3tSkqTS6jbDfc8+u8IunyT9k2puHtjeLFN7UpKk8jKk\nJEmlZUhJkkrLkJIklZYhJUkqLUNKklRahpQkqbQMKUlSaRlSkqTSMqQkSaVlSEmSSqvbrN0nSXq9\n1atfYtasL/H888+x5ZZbcf/9v2XkyG0YPHgIK1Ys54ILLuG8887m6acX0dKyliOOOIopU6Zy/PHH\ncuqpn2fkyG245ZYbWbp0KfvvfxDnnPMl+vXrx/PPP8eECZOYOfPjXfr9DClJ6sZ++MMfMGLEWzjn\nnPN54omFfPCDR7DNNqOYOnU/Jk3ai5tu+h6DBw/hjDNmsWrVKmbM+ADjx+9GQ0Pteq6vvn/mmSVc\nc8336dOnD8cdN5PJk/fi7W/frvO/WJXDfZLUjT3xxELGjdsRgJEjRzFo0OavvAd4/PGF7LzzLgA0\nNTUxevRoFi166jVt1D5XcIcdxrHpppvSq1cvdthhHE8++UQnfIv1M6QkqRsbPXosDz30IACLFj3F\nsmV/pbW1lXUdpW22Gc3vf/87AFatWsmCBX9h+PAR9O27Cc899ywADz/851faW7DgEdasWUNLSwt/\n+tMfGDPmbZ37hdpwuE+SurEDDzyEc889k+OPP5atthpG376bVIfyKil1yCGHcv7553DccTNZvXo1\nM2Ycy+DBgznssCO4+OLz2XLLYTQ3N9cM/zVw6qmfYfnyZeyzz76MHj2my74bdKPHx/vQQ0l6vYce\neoAXX1zFbrvtzpNPPsEpp3yK66//wZtq6+mnF/OVr1zIBRd8ZSNXuWHre+ihPSlJ6sa23noEZ575\nea666krWrFnDiSee9qbbamhooKHdqOg69qQkSV3Ox8dLkrodQ0qSVFrek5KkLtLS0sLChY9u1DZH\njRpDr169NmqbXcmQkqQusnDho3zqwltpGrTlRmlv1bL/5aunHMzYsdtulPbKwJCSpC7UNGhLBgwe\n0dVlsHTp83zrW9/gpJNO4/77f8vAgZsxduzbOPjg/bj11ttfc+5tt/2YgQM3Y489Jhdel/ekJEkM\nGbIFJ51Umb4+Z86tr6xG0d6U9GnTDuyUgAJ7UpLUo3zkIx/koosuZcCAAey//xQuu+wKtt02mDbt\n3QwfPpzTTvsi8+b9ikceeZhRo0bz8st/56yzvsAzzyxh0KBBzJp1Pldf/U222GIo22wzimuvvZq+\nffuwePEipkzZl6OPnrFR6zWkJKkHmTRpT+bNu4/m5i3ZeusR/PrX8+jTpy/vfOfuPPPMEiK2Y/fd\nJ7DPPvux1VbDePHFVXzsY8czbNgwTjjhYzzySL5mBfVnnlnC7NnX8/LLLzN9+ns2ekg53CdJPcjk\nyXtz3333MG/erzj22OOYP/9/uOeeu4jYnvYWd9hss0EMGzYMqAwJvvTSS685PnbsWBobG9l0003Z\nZJNNNnq99qQkqQutWva/ndrWmDFjWbx4EX/96wt8/OPHM3v2Vdxzz92cdtoXuOuuO4DK8kgtLS3V\n9xtqsdh1lAwpSeoio0aN4aunHLzR29yQXXcdz5Ili2loaGCXXd7BwoWP0a9fv1eG8XbYYRxf//pl\nbL31CNYXQuvOXd/DEzcW1+6TJHW5Tl8FPSIagcuBnYDVwMzMXFBzfDfgIirRuwg4OjNfLqoeSVL3\nU+TEielA38ycAJxOJZAAiIgG4Argw5k5Cfg5MLrAWiRJ3VCRITURmAuQmfOA8TXH3g48D5wYEb8A\nNs/MLLAWSVI3VGRIbQYsr9luqQ4BAgwFJgCXAvsAUyJi7wJrkSR1Q0XO7lsODKzZbszMtdX3zwN/\nWdd7ioi5VHpadxZYjySViqugb1iRIXUvcBBwQ0TsDjxQc+xRYEBEjK1OppgEfKPAWiSpdBYufJRT\nbz2D/s0DN3xyHVY+u4ILDj7bVdDr9ANgakTcW90+JiKOBAZk5pUR8RHguuokinsz87YCa5GkUurf\nPJCBW2/e1WWUVmEhlZmtwCfa7H645vidwLuKur4kqftzxQlJ6kFWrvwb55//Zf72txU899yzHHro\n4Wy77XZceunFrF27lubmZs444xxOOukEhgzZghUrlnPBBZdw3nln8/TTi2hpWcsRRxzFlClTufnm\nG5g7dw6NjY1st90OfPrTJ3PXXXfwne/Mpnfv3gwd2sxZZ53bZlWKN8aQkqQeZNGip5gyZV/23HNv\nnnvuOY4//qP069ePs846l5EjRzFnzq08/vhjNDQ0MHXqfkyatBc33fQ9Bg8ewhlnzGLVqlXMmPEB\nxo/fjdtu+xEnnfRZtttue2655UZaWlr42c9+ylFHHc2ee76buXPnsHLlSgYMGPCm6zWkJKkHGTx4\nCN///ne5++47aGoaQEtLC0uXLmXkyFEAHHDAq2sJrtv3+OMLGT++cnemqamJ0aNHs2jRU3z2s1/i\n+uuvZfHiRYwbtxOtra2ccMJnuOaab3PDDdczatRoJk/e6x+q15CSpC608tkVndrW9dd/h3HjdmT6\n9MP47W/nc999v2To0GaeeupJ3vKWt3LddbN5y1tGAq8uHrvNNqP5/e9/x+TJe7Fq1UoWLPgLw4eP\nYPbsqzj55M/St29fTjzxBB566AHmz/8fZsw4lsGDB3Phhedy1113Mm3agW/6OxlSktRFRo0awwUH\nn73R2+zIxImTuOSSC7n77l8wevQY+vfvz8knn855551NQ0MDQ4c2c/jhR3LDDd995TOHHHIo559/\nDscdN5PVq1e/EkJjx47lk5+cSVNTf5qbt2SHHcaxcuVKTj310zQ19aepqYmJE/+xx8y7Crokqcut\nbxV0n8wrSSotQ0qSVFqGlCSptLrNxIkFCx7p6hIkqcfqqoVru83EicmfO7B1Yy3CKEmqX2csXNvp\nj4/f2FyEUZJ6Hu9JSZJKy5CSJJWWISVJKi1DSpJUWoaUJKm0DClJUmkZUpKk0jKkJEmlZUhJkkrL\nkJIklZYhJUkqLUNKklRahpQkqbQMKUlSaRlSkqTSMqQkSaVlSEmSSsuQkiSVliElSSqthtbW1q6u\noS7jx49vt9Axp+zW7vmPXvjrdvd7vud7vud7/ps7/7bb7mx3/7Rpe//D58+fP7+hvXPtSUmSSqvb\n9KSmfeXI1oFbb97VZUhSj7Ni8V/50r+cwtix2xZ2jebmgfakJEndiyElSSotQ0qSVFqGlCSptAwp\nSVJpGVKSpNIypCRJpWVISZJKy5CSJJWWISVJKi1DSpJUWoaUJKm0DClJUmkZUpKk0jKkJEmlZUhJ\nkkrLkJIklZYhJUkqLUNKklRahpQkqbQMKUlSaRlSkqTSMqQkSaVlSEmSSsuQkiSVVu8NnRARDwFX\nA9dk5pLiS5IkqaKentSBQD/gzoj4SUQcHhF9Cq5LkqQNh1RmLszMszNze+BK4GJgSURcEhFbFF6h\nJKnHqme4byBwGPBBYATwX8D3gP2A24HxRRYoSeq5NhhSwKPAHOBM4JeZ2QoQEV8D9l3fhyKiEbgc\n2AlYDczMzAXtnHcF8HxmfvYNVy9J+qdWT0h9JDNvrd0REYdm5s3A9A4+Nx3om5kTIuJdwEVtz4+I\njwHjgF+8oaolST3CekMqIt4HbAKcFRGbAw1AK9AH+Bxw8wbangjMBcjMeRHxmmHBiJgAvBP4OrDd\nm/0CkqR/Xh31pDYDJlRf967Zv4ZKSG3IZsDymu2WiGjMzLURMRw4A/hX4Ig3VrIkqadYb0hl5hXA\nFRExJTN//ibaXg4MrNluzMy11feHAUOBnwDDgKaI+FNmzn4T15Ek/ZPqaLjvysz8KPCFiPhCm8Ot\nmfnuDbR9L3AQcENE7A48sO5AZl4KXFq9zoeA7QwoSVJbHQ33fa36+mUqQ3y1Wuto+wfA1Ii4t7p9\nTEQcCQzIzCvfRHuSpB6mo+G+31TfXpiZu7zRhqtT1T/RZvfD7Zx39RttW5LUM9SzLNKSiJgcEZsU\nXo0kSTXq+Z3UeKq/Y4qIdftaM7NXQTVJkgTUEVKZ2dwZhUiS1FY9a/dtBRwF9Kfyg95ewOjMPLrg\n2iRJPVw996RuBnamssBsf+Bg4Kkii5IkCeoLqaGZ+SHgR1Smle8F7FZkUZIkQX0htbT6msBOmbmM\nymoRkiQVqp7ZfXdExA3AycBPI+IdVB69IUlSoep5Mu/ngdMz83Hg/cCfgUOLLkySpI7W7vsQry5X\n1BARe1TfLwX2AVxrT5JUqI6G+/am4zX1DClJUqE6Wrvvw51YhyRJr9PRcN+czDwgIh5r53BrZo4p\nsC5Jkjoc7vto9XXvdo75aA1JUuE6Gu5bXH37NLAfsHl1u4FKSHlPSpJUqHp+J3Vb9fXxNvsNKUlS\noeoJqS0yc+fCK5EkqY16lkW6IyKmRkQ950qStNHU05N6ArgdfOihJKlz1RNSnwZGZeYTRRcjSVKt\neobwnuLVldAlSeo09fSkFgMPRcS9wMvVfa2ZOaO4siRJqi+k5lT/1PLHvJKkwnW0LNKwzFwC3Ekl\nlBpqDhtSkqTCddST+iZwAHAXrw+lVsC1+yRJhepoWaQDqq+jOq0aSZJqbPCeVES8C5gIXAb8CNgF\n+ERm3lhwbZKkHq6eKej/CfwGeC/wIvAO4PQii5IkCeoLqcbMvIvK/ambqj/qdbUJSVLh6gmpVRFx\nMjAF+HFEfApYUWxZkiTVF1JHAU3AoZm5FBgGvL/QqiRJor4f8z4L/DAzfx8RRwF9gJZiy5Ikqb6e\n1LXAYdVZfmcCy4CriyxKkiSoL6RGZ+YXqczu+2ZmzgIGF1uWJEn1hVSviBgKTAfmRMRwKveoJEkq\nVD0hdSEwD/hJZj4I/AKYVWRRkiRBHRMnMvM64LqaXdtn5triSpIkqaKjVdDnZOYBEfFYO8daM9MF\nZiVJheqoJ3VPRBxNZUafJEmdrqOQGgFsDWwPvA24BVgDHAj8GaehS5IK1tGjOo4HiIi7gV0y84Xq\n9tnAbZ1TniSpJ6tndt9WwPKa7ZeA5mLKkSTpVfUsi/Qj4OcRcSOVUHsf8N1Cq5IkiTp6Upl5MnAp\nEFTuTZ2XmV8qujBJkurpSZGZNwE3FVyLJEmvUc89KUmSuoQhJUkqLUNKklRahpQkqbQMKUlSaRlS\nkqTSMqQkSaVlSEmSSsuQkiSVliElSSotQ0qSVFqGlCSptAwpSVJpGVKSpNIypCRJpWVISZJKy5CS\nJJWWISVJKq26Hh//ZkREI3A5sBOwGpiZmQtqjh8JfApYAzwIHJeZrUXVI0nqforsSU0H+mbmBOB0\n4KJ1ByKiHzAL2Csz9wAGAQcWWIskqRsqMqQmAnMBMnMeML7m2EvAv2TmS9Xt3sCLBdYiSeqGigyp\nzYDlNdst1SFAMrM1M58FiIgTgP6Z+bMCa5EkdUOF3ZOiElADa7YbM3Ptuo1qYF0AvA14b4F1SJK6\nqSJ7UvcC+wNExO7AA22Ofx3YBPjXmmE/SZJeUWRP6gfA1Ii4t7p9THVG3wBgPjADuBu4IyIAvpqZ\ntxRYjySpmykspKrTyT/RZvfDNe97FXVtSdI/B3/MK0kqLUNKklRahpQkqbQMKUlSaRlSkqTSMqQk\nSaVlSEmSSsuQkiSVliElSSotQ0qSVFqGlCSptAwpSVJpGVKSpNIypCRJpWVISZJKy5CSJJWWISVJ\nKi1DSpJUWoaUJKm0DClJUmkZUpKk0jKkJEmlZUhJkkrLkJIklZYhJUkqLUNKklRahpQkqbQMKUlS\naRlSkqTSMqQkSaVlSEmSSsuQkiSVliElSSotQ0qSVFqGlCSptAwpSVJpGVKSpNIypCRJpWVISZJK\ny5CSJJWWISVJKi1DSpJUWoaUJKm0DClJUmkZUpKk0jKkJEmlZUhJkkrLkJIklZYhJUkqLUNKklRa\nhpQkqbQMKUlSaRlSkqTSMqQkSaVlSEmSSsuQkiSVliElSSotQ0qSVFqGlCSptAwpSVJpGVKSpNIy\npCRJpWVISZJKy5CSJJVW76IajohG4HJgJ2A1MDMzF9QcPwj4IrAGuCozv1FULZKk7qnIntR0oG9m\nTgBOBy5adyAi+gAXA1OBPYFjI2LLAmuRJHVDhfWkgInAXIDMnBcR42uObQ/8JTOXAUTEPcBk4Mb1\nNXbVhy+hsdHRSUnqbGvXrmVI0+b07t2n069dZEhtBiyv2W6JiMbMXFs9tqzm2ApgUEeNDR+8VcPG\nL1GSVGZFdk2WAwNrr1UNKKgEVO2xgcALBdYiSeqGigype4H9ASJid+CBmmN/BraNiMER0ZfKUN+v\nCqxFktQNNbS2thbScEQ08OrsPoBjgHcAAzLzyog4EDiDSlB+MzP/q5BCJEndVmEhJUnSP8rpcpKk\n0jKkJEmlZUhJkkrLkJIklZYhJUkqLUNKklRahpQkqbT+D0pSgCvPAJoMAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15afaaf0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAUsAAAEYCAYAAADVrdTHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x163d9330>"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Training and testing sets"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Notice -- and this is very important -- that we **always** split data into two sets, and one of them is used for training *only*, while the other is used for testing *only*. This way, we train SVM on one data set and then, using another set, ask how well it learned the separation between conditions. What do you think its performance would be if we trained and tested on the same dataset?\n",
      "\n",
      "Answer: usually you would get a very good performance, not 100%, but good nonetheless because for most datasets SVM will be able to find some sort of a separating hyperplane that does indeed separate the two conditions to some extent. However, the whole point is to see if this hyperplane is actually useful at all in general, not just for this training set. That's why we use a separate testing set to check whether there is a general separation rule."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Cross-validation"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "So what is this cross-validation? In `psychopy_ext`, we split data randomly into training and test samples. If we did it in a more systematic manner this would yield the cross-validation. For example, a popular cross-validation technique is know as *leave-one-out* where all but one runs are used in the training set and the remaining one is the testing set. In our case, we would take the first run as the testing set and the remaining seven runs as the training set. In the next iteration, we would take the second run as the testing run and the other seven as the training set. And so forth. In the end, this would be an *eight-fold cross-validation* as if we folded the data in eight different ways.\n",
      "\n",
      "Importantly, data always remains independent."
     ]
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Nonlinear classifiers"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "So far we only used linear classifiers in our examples, i.e., those that are only capable of drawing a hyperplane between two classes. But what if the distributions of the signal look like this:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ds = mvpa2.suite.pure_multivariate_signal(npoints, 2)\n",
      "ds.samples += 6\n",
      "plot_data(ds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAREAAAERCAYAAABclJOOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncuGFiMfjycX/B0fgSj896vOnAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15b86910>"
       ]
      }
     ],
     "prompt_number": 13
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "There is no way you can separate these two populations linearly. But you can clearly see a difference between two populiations so surely there must be a way to do that. And surely there is because this is a very common situation that machine learning has to deal with. Instead of using linear classifiers we can simply use their nonlinear variants. In the example below, we use an SVM with the radial basis function kernel."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "res, supvecs = train_svm(ds, cl=mvpa2.suite.RbfNuSVMC)\n",
      "plot_data(ds, supvecs=supvecs, contour=res)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPcAAAEMCAYAAAAYriA9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nt45OKeX/ATPweCYdpqDGAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x158585d0>"
       ]
      }
     ],
     "prompt_number": 14
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Awesome! Nonlinear classifiers works amazingly in machine learning. But should you use this in neuroimaging? Probably not. To illustrate that, imagine taking images with and without faces. You can go ahead and train your nonlinear classifier on these images or, if you want to relate it to brain responses, first filter them with Gabor-line filters just like in V1. The classifier will likely be able to differentiate between the two categories -- but will you want to make a claim that V1 does face processing? From what we know, V1 does not detect face information. Yet we can decode faces from it. Why?\n",
      "\n",
      "Well, the information about faces present or absent is in the signal (the image) itself by definition, so given a powerful enough model I will find this difference. But it does not mean that V1 makes use of this information. For its purposes, this information is unusable. The whole point of visual processing is to deconvolve signals into something usable, something *easily* discriminable. In neuroscience, we tend to assume that a linear separation is easy enough for the next layer in the visual hierarchy to perform.\n",
      "\n",
      "Not everybody necessarily agrees with this resctriction though."
     ]
    }
   ],
   "metadata": {}
  }
 ]
}