{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Benchmarks\n",
    "\n",
    "This notebook contains benchmarks against the [pynfft](https://pypi.python.org/pypi/pyNFFT) package, which is a Python wrapper of Fortran code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'0.1'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import from main directory\n",
    "import os, sys\n",
    "sys.path.append(os.path.abspath('..'))\n",
    "\n",
    "import nfft\n",
    "nfft.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.3.2'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pynfft\n",
    "pynfft.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Benchmarking the Forward Transform"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define some test data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def make_forward_data(M, N):\n",
    "    x = -0.5 + np.random.rand(M)\n",
    "    f_hat = np.random.randn(N) + 1j * np.random.randn(N)\n",
    "    return x, f_hat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a utility function around pynfft:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def pynfft_forward(x, f_hat):\n",
    "    M = len(x)\n",
    "    N = len(f_hat)\n",
    "    plan = pynfft.nfft.NFFT(N, M)\n",
    "    plan.x = x\n",
    "    plan.precompute()\n",
    "    plan.f_hat = f_hat\n",
    "    # Need copy because of bug in pynfft 1.x\n",
    "    # See https://github.com/ghisvail/pyNFFT/issues/57\n",
    "    return plan.trafo().copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "x, f_hat = make_forward_data(1000, 100000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out1 = nfft.nfft(x, f_hat)\n",
    "out2 = pynfft_forward(x, f_hat)\n",
    "np.allclose(out1, out2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10 loops, best of 3: 22.4 ms per loop\n",
      "10 loops, best of 3: 43.1 ms per loop\n"
     ]
    }
   ],
   "source": [
    "%timeit nfft.nfft(x, f_hat)\n",
    "%timeit pynfft_forward(x, f_hat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Benchmarking the Adjoint Transform"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define some test data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def make_adjoint_data(M):\n",
    "    x = -0.5 + np.random.rand(M)\n",
    "    f = np.random.randn(M) + 1j * np.random.randn(M)\n",
    "    return x, f"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a utility function around pynfft:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def pynfft_adjoint(x, f, N):\n",
    "    M = len(x)\n",
    "    plan = pynfft.nfft.NFFT(N, M)\n",
    "    plan.x = x\n",
    "    plan.precompute()\n",
    "    plan.f = f\n",
    "    # Need copy because of bug in pynfft 1.x\n",
    "    # See https://github.com/ghisvail/pyNFFT/issues/57\n",
    "    return plan.adjoint().copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "x, f = make_adjoint_data(1000)\n",
    "N = 100000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out1 = nfft.nfft_adjoint(x, f, N)\n",
    "out2 = pynfft_adjoint(x, f, N)\n",
    "\n",
    "np.allclose(out1, out2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10 loops, best of 3: 24.2 ms per loop\n",
      "10 loops, best of 3: 42.8 ms per loop\n"
     ]
    }
   ],
   "source": [
    "%timeit nfft.nfft_adjoint(x, f, N, sigma=3)\n",
    "%timeit pynfft_adjoint(x, f, N)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Fast Template Periodogram",
   "language": "python",
   "name": "ftp"
  },
  "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": 0
}