Output arrays in format that can be compared with Python numpy

Question

Is there a way to convert an array in Mathematica such that its output format resembles that of Python's numpy arrays? I want to go from:

{{0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 0.0625, -0.0625, 0.0625, 
  0, 0, -0.125, 0.0625, 0, 0}, 
 {-0.0625, 0.1875, 0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 
  0, 0, 0, -0.0625, 0, 0}}

to something that looks like this:

np.array([
  [0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 0.0625, -0.0625, 0.0625, 
   0, 0, -0.125, 0.0625, 0, 0], 
  [-0.0625, 0.1875, 0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 
   0, 0, 0, -0.0625, 0, 0]])

Related: How to convert a Mathematica list in a Python list?. — ybeltukov, Dec 03 '15 at 14:46

Jason B. · Accepted Answer · 2015-12-03T14:48:04.897

Here is a quick and dirty method to do it,

input = {{0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 
   0.0625, -0.0625, 0.0625, 0, 0, -0.125, 0.0625, 0, 0}, {-0.0625, 
   0.1875, 0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 0, 0, 
   0, -0.0625, 0, 0}}
(* {{0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 0.0625, -0.0625,
   0.0625, 0, 0, -0.125, 0.0625, 0, 0}, {-0.0625, 0.1875, 
  0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 0, 0, 0, -0.0625,
   0, 0}} *)

First I convert the list to a string, then I use StringReplace and copy the result to the clipboard to paste wherever I need it,

CopyToClipboard@
 StringReplace[
  ToString@input, {"{{" -> "np.array([[", "}, {" -> "], [", 
   "}}" -> "]])"}]

The output has no line breaks,

np.array([[0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 0.0625, -0.0625, 0.0625, 0, 0, -0.125, 0.0625, 0, 0], [-0.0625, 0.1875, 0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 0, 0, 0, -0.0625, 0, 0]])

A more polished, functional form, which can deal with differently shaped arrays, and with large and small numbers, is

toNumpy = CopyToClipboard[
    "np.array(" <>
     StringReplace[
      ToString[
       Map[FortranForm, #, {-1}]]
      , {"{" -> "[", "}" -> "]"}] <> ")"
    ] &;

Some test cases,

toNumpy@RandomReal[10^13, {2, 2, 2}]

np.array([[[6.515662614124779e12, 5.006958562518258e12], [9.038456654284738e12, 6.155528971184453e12]], [[8.352734900856066e12, 3.183407977590039e11], [9.3300341220833e12, 4.83956443633259e12]]])

and

toNumpy@RandomReal[10^-13, {5}]

np.array([2.5981516628601564e-14, 8.152269536832112e-14, 5.284716124977113e-16, 5.3549719092414376e-14, 2.877505407608681e-14])

I try to run the toNumpy versions and they don't give any output on my computer. — Alejandro Marcos Aragon, Dec 03 '15 at 14:49
Sorry, I assumed you wanted to paste it somewhere to compare with python, or paste into your python code directly, so I put CopyToClipboard in the function. Just hit paste and it will come out. Just erase that part if you'd rather have it in the notebook — Jason B., Dec 03 '15 at 14:59
The result is in the clipboard, you have to paste it somewhere... — Alejandro Marcos Aragon, Mar 14 '16 at 14:29

score 2 · Answer 2 · answered Apr 22 '17 at 02:26

I've prepared a Module to handle such conversions (original answer https://mathematica.stackexchange.com/a/144200/48312). It would work like this for Alejandro's question:

ToPython[{{0.1875, -0.0625, 0.0625, -0.0625, -0.1875, 0.0625, -0.0625,
0.0625, 0, 0, -0.125, 0.0625, 0, 0}, {-0.0625, 0.1875, 
0.0625, -0.0625, 0.0625, -0.1875, -0.0625, 0.0625, 0, 0, 
0, -0.0625, 0, 0}}]

The output in Mathematica would be a string (that is directly copied to clipboard)

numpy.array( [numpy.array( [0.1875e0,(-0.625e-1),0.625e-1,(-0.625e-1),         (-0.1875e0),0.625e-1,(-0.625e-1),0.625e-1,0,0,(-0.125e0),0.625e-1,0,0,]     ),numpy.array( [(-0.625e-1),0.1875e0,0.625e-1,(-0.625e-1),0.625e-1,(-0.1875e0),    (-0.625e-1),0.625e-1,0,0,0,(-0.625e-1),0,0,] ),] )

Once copied and interpret by Numpy...

array([[ 0.1875, -0.0625,  0.0625, -0.0625, -0.1875,  0.0625, -0.0625,
     0.0625,  0.    ,  0.    , -0.125 ,  0.0625,  0.    ,  0.    ],
   [-0.0625,  0.1875,  0.0625, -0.0625,  0.0625, -0.1875, -0.0625,
     0.0625,  0.    ,  0.    ,  0.    , -0.0625,  0.    ,  0.    ]])

Output arrays in format that can be compared with Python numpy

2 Answers2