niklasschmitz

#!/bin/bash
# 1. Install CUDA
echo "Installing CUDA..."
# Only install if CUDA is not already installed.
if ! sudo dpkg-query -W cuda; then
  # The 18.04 installer
  sudo curl -O http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-repo-ubuntu1804_10.0.130-1_amd64.deb
  sudo dpkg -i ./cuda-repo-ubuntu1804_10.0.130-1_amd64.deb
  sudo apt-get update
  sudo apt-get install cuda -y

niklasschmitz

import jax
from jax import core
from graphviz import Digraph
import itertools


styles = {
  'const': dict(style='filled', color='goldenrod1'),
  'invar': dict(color='mediumspringgreen', style='filled'),
  'outvar': dict(style='filled,dashed', fillcolor='indianred1', color='black'),

niklasschmitz

import plot

-- Conjugate gradients solver
def solve (mat:m=>m=>Float) (b:m=>Float) : m=>Float =
  x0 = for i:m. 0.0
  ax = mat **. x0
  r0 = b - ax
  (xOut, _, _) = fold (x0, r0, r0) $
     \s:m (x, r, p).
       ap = mat **. p

niklasschmitz

using NLsolve
using Zygote
using ChainRulesCore
using IterativeSolvers
using LinearMaps
using SparseArrays
using LinearAlgebra
using BenchmarkTools
using Random
Random.seed!(1234)

niklasschmitz

import torch
import jax
import jax.dlpack
import torch.utils.dlpack
from jax.config import config
config.update("jax_enable_x64", True)

def jax2torch(x):
    return torch.utils.dlpack.from_dlpack(jax.dlpack.to_dlpack(x))

niklasschmitz

# rrule from frule (transposition)
using Zygote
using ChainRulesCore
using LinearAlgebra

function f(x)
    a = sin.(x)
    b = sum(a)
    c = b * a
    return c

niklasschmitz

import plot
import png

key = new_key 1234
Height = Fin 100
Width = Fin 300
img_ = for i:Height. for j:Width. n_to_f ((ordinal i) + (ordinal j))
img = img_ / (n_to_f (size Height + size Width))
:html matshow img
:t img

niklasschmitz

using Images
# CC0 license: https://commons.wikimedia.org/wiki/File:Wikicommons_kamera_rollei35_rrt877.jpg
file = download("https://upload.wikimedia.org/wikipedia/commons/thumb/3/38/Wikicommons_kamera_rollei35_rrt877.jpg/640px-Wikicommons_kamera_rollei35_rrt877.jpg")
img = load(file)
open("camera.ppm", "w+") do io
     println(io, "P6")
     println(io, size(img, 2), " ",  size(img, 1))
     println(io, 255)

     for row in eachrow(img)

niklasschmitz

import png

' ## Colormaps
  Here we model colormaps as sequences of RGB colors which are then
  piecewise linearly interpolated in the interval [0,1] upon evaluation.

RGB = (Fin 3)=>Float

cmap_viridis = [[253.0, 231.0,  37.0]
               ,[ 94.0, 201.0,  98.0]