HelperFunctions is added

HelperFunctions.ipynb

+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Here you can find some useful functions for working with the SpiNNaker ;) "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "~~~\n",
+    "    ExtractConnectionFile(Input_File, Connections_File,Non_Zero, Connections_shape, n_samples_to_plot)\n",
+    "~~~\n",
+    "\n",
+    " Using the CSNN simulator, by exporting the weights as binary files, using ***ExtractConnectionFile*** we can extract the weights as connection files to use as projections between the Populations on the SpiNNaker"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "import pathlib\n",
+    "import math\n",
+    "from mpl_toolkits.axes_grid1 import ImageGrid\n",
+    "import matplotlib.pyplot as plt\n",
+    "\n",
+    "def plot_and_save_grid(input_data,nrows_ncols,lable='Sample Plost',save=False,file_name=None):\n",
+    "    fig = plt.figure(figsize=(10, 10))\n",
+    "    grid = ImageGrid(fig, 111,  # similar to subplot(111)\n",
+    "                     nrows_ncols,  # creates nrow x ncol grid of axes\n",
+    "                     axes_pad=0.1,  # pad between axes in inch.\n",
+    "                     )\n",
+    "\n",
+    "    for ax, inx in zip(grid, list(range(nrows_ncols[0]*nrows_ncols[1]))):\n",
+    "        ax.imshow(input_data[inx],cmap=\"jet\")\n",
+    "        ax.set_axis_off()\n",
+    "    \n",
+    "\n",
+    "    if save:\n",
+    "        plt.savefig(file_name)\n",
+    "    return\n",
+    "\n",
+    "def Plot_extracted_weights(Input_File, Connections_File,Connections_shape={'x':28, 'y':28, 'z':400}, n_samples_to_plot=0):\n",
+    "    w = np.fromfile(Input_File, dtype=np.dtype(np.float32))#[28, 28, 1, 400]\n",
+    "    print(f'Input file shape = {w.shape}')\n",
+    "    w=w.reshape(Connections_shape['x'], Connections_shape['y'], Connections_shape['z'])#x, y, z\n",
+    "    w=np.array([w[:, :, i].transpose() for i in range(Connections_shape['z']) ])\n",
+    "    w=w.reshape(Connections_shape['z'], Connections_shape['x']*Connections_shape['y']).transpose()# 400rows of 28*28 to 400 rows of 784 and then transposed==> 784*400\n",
+    "    if n_samples_to_plot:\n",
+    "        w_to_plot=[w.reshape(Connections_shape['x'], Connections_shape['y'],-1)[:,:,i] for i in range(n_samples_to_plot)]\n",
+    "        plot_and_save_grid(w_to_plot,(int(math.sqrt(n_samples_to_plot)),int(math.sqrt(n_samples_to_plot))))\n",
+    "    return\n",
+    "\n",
+    "def ExtractConnectionFile(Input_File, Connections_File,Non_Zero=False, Connections_shape={'x':28, 'y':28, 'z':400}, n_samples_to_plot=0):\n",
+    "    extracted_weights=[]\n",
+    "    w = np.fromfile(Input_File, dtype=np.dtype(np.float32))#[28, 28, 1, 400]\n",
+    "    print(f'Input file shape = {w.shape}')\n",
+    "#    w=w.reshape(Connections_shape['x'], Connections_shape['y'], Connections_shape['z'])#x, y, z\n",
+    "#    w=np.array([w[:, :, i].transpose() for i in range(Connections_shape['z']) ])\n",
+    "    f=open(Connections_File, 'w')\n",
+    "    f.write('# columns = [\"i\", \"j\", \"weight\", \"delay\"]\\n')\n",
+    "    w=w.reshape(Connections_shape['x']*Connections_shape['y'], Connections_shape['z'])# 400rows of 28*28 to 400 rows of 784 and then transposed==> 784*400\n",
+    "    X,Y=w.shape\n",
+    "    for x in range(X):\n",
+    "        for y in range(Y):\n",
+    "            if Non_Zero:\n",
+    "                if w[x][y]:\n",
+    "                    line=f'{x} {y} {w[x][y]} 1\\n'\n",
+    "                    f.write(line)\n",
+    "                    extracted_weights.append(w[x][y])\n",
+    "            else:\n",
+    "                    line=f'{x} {y} {w[x][y]} 1\\n'    \n",
+    "                    f.write(line)\n",
+    "                    extracted_weights.append(w[x][y])\n",
+    "            \n",
+    "    f.close()\n",
+    "    if n_samples_to_plot:\n",
+    "        w_to_plot=[w.reshape(Connections_shape['x'], Connections_shape['y'],-1)[:,:,i] for i in range(n_samples_to_plot)]\n",
+    "        plot_and_save_grid(w_to_plot,(int(math.sqrt(n_samples_to_plot)),int(math.sqrt(n_samples_to_plot))))\n",
+    "\n",
+    "    print(f'Output file shape = {w.shape}')\n",
+    "    print(f'Connections File Path = {pathlib.Path(Connections_File).parent.resolve()}/{Connections_File}')\n",
+    "    return pathlib.Path(Connections_File).parent.resolve(), extracted_weights"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## HOw to use:\n",
+    "Just put the binary files in an array as below and call the function for all the files:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "files_name=['weights_fc1_0_0_25_0.7', 'weights_fc2_0_0_25_0.7', 'weights_fc1_1_1_25_0.7', 'weights_fc2_1_1_25_0.7']\n",
+    "weights=[[],[],[],[]]\n",
+    "shape=[{'x':28, 'y':28, 'z':400}, {'x':20, 'y':20, 'z':1600}]\n",
+    "Non_Zero=False\n",
+    "for i, file_name in enumerate(files_name):\n",
+    "    _, weights[i]=ExtractConnectionFile(Input_File=file_name, \n",
+    "                            Connections_File=f'{file_name}_Zero_Removed_{str(Non_Zero)}.txt',\n",
+    "                            Non_Zero=Non_Zero,\n",
+    "                            Connections_shape=shape[i%2],\n",
+    "                            n_samples_to_plot=0\n",
+    "                            )\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3.8.10 ('snnToolBox')",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "name": "python",
+   "version": "3.8.10"
+  },
+  "orig_nbformat": 4,
+  "vscode": {
+   "interpreter": {
+    "hash": "20a24a3a4007955ef5ee2dc9fb7716c1a03a2c0fc0e1cbee0d7177cafa6993b3"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}