jax_compile.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import jax.numpy as np\n",
    "from jax import jit, lax, random\n",
    "from jax.util import partial"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/fehiepsi/jax/jax/lib/xla_bridge.py:122: UserWarning: No GPU found, falling back to CPU.\n",
      "  warnings.warn('No GPU found, falling back to CPU.')\n"
     ]
    }
   ],
   "source": [
    "x = random.normal(random.PRNGKey(0), ())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def f(x, a):\n",
    "    return x + a"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 12.6 ms, sys: 49 µs, total: 12.6 ms\n",
      "Wall time: 12 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(1.7941577, dtype=float32)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time f(x, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def g(fn, a):\n",
    "    return lax.fori_loop(0, 1000000, lambda i, v: fn(v), a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "@jit\n",
    "def h(x, a):\n",
    "    pf = lambda a: f(x, a)\n",
    "    return g(pf, a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 20.2 ms, sys: 13 µs, total: 20.2 ms\n",
      "Wall time: 19.3 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(-204124.73, dtype=float32)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time h(x, 3.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.05 ms, sys: 2 µs, total: 4.05 ms\n",
      "Wall time: 3.42 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(-204126.7, dtype=float32)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time h(x, 1.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = x + 0.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.43 ms, sys: 40 µs, total: 4.47 ms\n",
      "Wall time: 3.76 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(-106981.37, dtype=float32)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time h(y, 3.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.2 ms, sys: 15 µs, total: 4.21 ms\n",
      "Wall time: 3.74 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(-106980.33, dtype=float32)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time h(y, 4.)"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

I think this seems reasonable. Let me check a few other things I had in mind and get back to you on this.