TransWikia.com

Problem passing a list from Mathematica to external python function

Mathematica Asked by tom houston on March 6, 2021

comp1a = Flatten[Join[{{0 + 0 i}}, {comp1}]];
Flatten[comp1a[[1 ;; 6]]]

This creates a list of complex values -> comp1a

The first 6 terms in the list are:

{0, -2.04375 + 1.33694 I, -1.0492 + 3.3731 I, -4.95694 +
1.01694 I, 4.61625 – 4.52896 I, 6.66754 – 4.1488 I}

Then I try to pass the list to;

ExternalEvaluate["Python", "import numpy as np; timhist = np.fft.ifft(comp1a); timehist.tolist()"]

and the result is;

Failure
Message:"name ‘comp1a’ is not defined
Tag:PythonError

My question is how do I feed a list developed in Mathematica to a python function called externally

One Answer

This will not work. Python does not know what mylist is:

mylist = {1, 2, 3, 4};
ExternalEvaluate["Python", "sum(mylist)"]

You must get the value of mylist into Python somehow. This could be done with string replacement:

ExternalEvaluate["Python", "sum(" <>
  StringReplace[ToString[mylist], {"{" -> "[", "}" -> "]"}]
  <> ")"]

... though as Jason .B noted in the comments, it's much simpler to write this as a TemplateExpression:

ExternalEvaluate["Python", "sum(<* mylist *>)"]

For complex numbers, you'll need to rewrite them as a complex type first before you can feed them to Python:

createcomplex[z_?NumericQ] := 
 "complex(" <> ToString[Re[z]] <> "," <> ToString[Im[z]] <> ")"

createcomplex[-2.7 + 3.5 I]
(* result: "complex(-2.7,3.5)" *)

Or more simply, using templates:

toPyComplexStr[z_?NumericQ] := 
 TemplateApply["complex(<* Re[z] *>, <* Im[z] *>)", z]

There is however, a better way in the documentation which doesn't depend on text substitution and it uses an external session:

session = StartExternalSession["Python"];
ExternalEvaluate[session, <|
  "Command" -> "sum",
  "Arguments" -> {NumericArray[{1 + I, 2.3 - I, 5.7 - 2.3 I, 7}, 
     "ComplexReal32"]}
  |>]
(* result: 16. - 2.3 I *)

Let's use it for your problem:

session = StartExternalSession["Python"];
result = Last@
  ExternalEvaluate[
   session, {"import numpy as np", <|"Command" -> "np.fft.ifft", 
     "Arguments" -> {NumericArray[{0, -2.04375 + 1.33694 I, -1.0492 + 
          3.3731 I, -4.95694 + 1.01694 I, 4.61625 - 4.52896 I, 
         6.66754 - 4.1488 I}, "ComplexReal32"]}|>}]
Normal[result]
(* result: {0.538983 - 0.491797 I, -1.01814 - 2.38259 I, -1.15996 + 0.0605042 I, 
 0.650033 + 0.10651 I, -1.85749 + 0.939762 I, 2.84658 + 1.76761 I} *)

... and notice that this result matches up with InverseFourier if you use FourierParameters -> {1, -1}:

InverseFourier[{0, -2.04375 + 1.33694 I, -1.0492 + 3.3731 I, -4.95694 + 1.01694 I, 4.61625 - 4.52896 I, 6.66754 - 4.1488 I}, 
  FourierParameters -> {1, -1}]

(* result: {0.538983 - 0.491797 I, -1.01814 - 2.38259 I, -1.15996 + 0.0605042 I, 
 0.650033 + 0.10651 I, -1.85749 + 0.939762 I, 2.84658 + 1.76761 I} *)

Answered by flinty on March 6, 2021

Add your own answers!

Ask a Question

Get help from others!

© 2024 TransWikia.com. All rights reserved. Sites we Love: PCI Database, UKBizDB, Menu Kuliner, Sharing RPP