{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Layering Nested Operations\n",
    "\n",
    "We start by loading the necessary libraries and resetting the computational graph."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import os\n",
    "from tensorflow.python.framework import ops\n",
    "ops.reset_default_graph()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create a graph session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess = tf.Session()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create the Tensors, Constants, and Placeholders\n",
    "\n",
    "We start by creating an array to feed in to a placeholder (note the agreements on the dimensions).  We then declare some graph constants to use in the operations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create data to feed in\n",
    "my_array = np.array([[1., 3., 5., 7., 9.],\n",
    "                   [-2., 0., 2., 4., 6.],\n",
    "                   [-6., -3., 0., 3., 6.]])\n",
    "# Duplicate the array for having two inputs\n",
    "x_vals = np.array([my_array, my_array + 1])\n",
    "# Declare the placeholder\n",
    "x_data = tf.placeholder(tf.float32, shape=(3, 5))\n",
    "# Declare constants for operations\n",
    "m1 = tf.constant([[1.],[0.],[-1.],[2.],[4.]])\n",
    "m2 = tf.constant([[2.]])\n",
    "a1 = tf.constant([[10.]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Declare Operations\n",
    "\n",
    "We start with matrix multiplication (A[3x5] * m1[5x1]) = prod1[3x1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 1st Operation Layer = Multiplication\n",
    "prod1 = tf.matmul(x_data, m1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Second operation is multiplication of prod1[3x1] by m2[1x1], which results in prod2[3x1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 2nd Operation Layer = Multiplication\n",
    "prod2 = tf.matmul(prod1, m2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The third operation is matrix addition of prod2[3x1] to a1[1x1], This makes use of TensorFlow's broadcasting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 3rd Operation Layer = Addition\n",
    "add1 = tf.add(prod2, a1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate and Print Output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 102.]\n",
      " [  66.]\n",
      " [  58.]]\n",
      "[[ 114.]\n",
      " [  78.]\n",
      " [  70.]]\n"
     ]
    }
   ],
   "source": [
    "for x_val in x_vals:\n",
    "    print(sess.run(add1, feed_dict={x_data: x_val}))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create and Format Tensorboard outputs for viewing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "merged = tf.summary.merge_all(key='summaries')\n",
    "\n",
    "if not os.path.exists('tensorboard_logs/'):\n",
    "    os.makedirs('tensorboard_logs/')\n",
    "\n",
    "my_writer = tf.summary.FileWriter('tensorboard_logs/', sess.graph)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![layering_nested_operations](https://github.com/nfmcclure/tensorflow_cookbook/raw/master/02_TensorFlow_Way/images/02_Multiple_Operations.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    ""
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3.0
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}