Matrix Operations with TensorFlow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Matrix Operations with TensorFlow\n",
    "#### http://vijayshinva.github.io"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matrix Addition\n",
    "\n",
    "![Matrix Addition](images/matrix_addition_one.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2], [3, 4]])\n",
    "matrix_two = tf.constant([[5, 6], [7, 8]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.add(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Broadcasting\n",
    "\n",
    "![Broadcasting](images/broadcasting_one.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n",
    "matrix_two = tf.constant([[10, 11, 12]]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.add(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Broadcasting](images/broadcasting_two.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n",
    "matrix_two = tf.constant([[10], [11], [12]]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.add(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Scalar Matrix Multiplication\n",
    "\n",
    "![Scalar Multiplication](images/scalar_multiplication.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "multiplier = tf.constant(5)\n",
    "matrix = tf.constant([[1, 2], [3, 4]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.multiply(multiplier, matrix)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(multiplier * matrix))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matrix Multiplication\n",
    "![Matrix Multiplication](images/matrix_multiply_one_correct.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2], [3, 4]])\n",
    "matrix_two = tf.constant([[5, 6], [7, 8]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matmul(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "## Donot use multiply or * with matrices\n",
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matmul(matrix_one, matrix_two)))\n",
    "    print('\\nThe following are wrong')\n",
    "    print(sess.run(matrix_one * matrix_two))\n",
    "    print(sess.run(tf.multiply(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Multiplication](images/matrix_multiply_four.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2], [3, 4]])\n",
    "matrix_two = tf.constant([[5, 6], [7, 8]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matmul(matrix_one, matrix_two)))\n",
    "    print(sess.run(tf.matmul(matrix_two, matrix_one)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Multiplication](images/matrix_multiply_two.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12], [13, 14, 15, 16]])\n",
    "matrix_two = tf.constant([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1]]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matmul(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Multiplication](images/matrix_multiply_three.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n",
    "matrix_two = tf.constant([[1], [2], [3]]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matmul(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Element wise Multiply\n",
    "![Broadcasting](images/matrix_multiply_five.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix_one = tf.constant([[1, 2, 3]])\n",
    "matrix_two = tf.constant([[2, 2, 2]]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.multiply(matrix_one, matrix_two)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matrix Transpose\n",
    "![Matrix Transpose](images/matrix_transpose_one.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1, 2], [3, 4]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_transpose(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Transpose](images/matrix_transpose_two.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1, 2], [3, 4], [5, 6]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_transpose(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matrix Determinant\n",
    "\n",
    "![Matix Determinant](images/matrix_det.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1.0, 2.0], [3.0, 4.0]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_determinant(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Determinant](images/matrix_det_two.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_determinant(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matrix Inverse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1.0, 2.0], [3.0, 4.0]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_inverse(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Matrix Inverse](images/matrix_inverse_two.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1.0, 2.0], [3.0, 4.0]])\n",
    "matrix_inverse = tf.matmul(matrix, tf.matrix_inverse(matrix))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(matrix_inverse))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Matrix Diagonal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "matrix = tf.constant([[1, 2, 3], [4, 5, 6], [7, 8, 9]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    print(sess.run(tf.matrix_diag(matrix)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### y = W * x + B"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "matrix_x = [[.9, .8, .7]]\n",
    "matrix_weights = [[1.0], [1.0], [1.0]]\n",
    "bias = 5\n",
    "with tf.Session() as sess:\n",
    "    matrix_y = tf.add(tf.matmul(matrix_weights, matrix_x), bias)\n",
    "    print(sess.run(matrix_y))"
   ]
  },
  {
   "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
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}