maximum_matrix_size_estimate

sub_row_memory_estimation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import math\n",
    "import numpy as np\n",
    "import scipy.stats as stats\n",
    "\n",
    "mean = 1.6\n",
    "sd = 0.75\n",
    "\n",
    "max_row_length = 1024\n",
    "num_rows = 1000\n",
    "max_delay = 0.7\n",
    "\n",
    "dist = stats.truncnorm\n",
    "params = {\"loc\": mean, \"scale\": sd, \"a\": (0.1 - mean) / sd, \"b\": (np.inf - mean) / sd}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Probabity in first sub-row:0.094469, maximum synapses:149\n"
     ]
    }
   ],
   "source": [
    "prob_first_sub_row = dist.cdf(max_delay, **params)\n",
    "row_probability = 0.9999 ** (1.0 / float(num_rows))\n",
    "max_first_sub_row_synapses = stats.binom.ppf(row_probability, max_row_length, prob_first_sub_row)\n",
    "print \"Probabity in first sub-row:%f, maximum synapses:%u\" % (prob_first_sub_row, max_first_sub_row_synapses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Subsequent sub-row maximum synapses:972\n"
     ]
    }
   ],
   "source": [
    "max_subsequent_sub_row_synapses = stats.binom.ppf(row_probability, max_row_length, 1.0 - prob_first_sub_row)\n",
    "print \"Subsequent sub-row maximum synapses:%u\" % (max_subsequent_sub_row_synapses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extension delay range:5.398744, max sub rows:8, max sub row length:121\n"
     ]
    }
   ],
   "source": [
    "# If any sub-rows are required\n",
    "if max_subsequent_sub_row_synapses > 0:\n",
    "    # Calculate the maximum range of delays\n",
    "    # this many synapses is likely to have\n",
    "    max_probability = 0.9999 ** (1.0 / float(max_subsequent_sub_row_synapses))\n",
    "    extension_delay_range = dist.ppf(max_probability, **params) -\\\n",
    "        dist.ppf(1.0 - max_probability, **params)\n",
    "\n",
    "    # Convert this to a maximum number of sub-rows\n",
    "    max_sub_rows = max(1, int(math.ceil(extension_delay_range /\n",
    "                                        0.7)))\n",
    "\n",
    "    # Divide mean number of synapses in row evenly between sub-rows\n",
    "    max_sub_row_length = max_subsequent_sub_row_synapses // max_sub_rows\n",
    "\n",
    "    print(\"Extension delay range:%f, max sub rows:%u, max sub row length:%u\"\n",
    "                          % (extension_delay_range, max_sub_rows, max_sub_row_length))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}