How to use python code with numpy in Matlab?

Hello, I'm trying to use python code in matlab.

As How can I use Python pandas in MATLAB? - (mathworks.com) shows, I installed anaconda and try to run my python code with pyrunfile.

clear all;
pyrunfile('test.py');

However, numpy seems to occur some errors below.

Error using __init__><module>
Python Error: ImportError: Unable to import required dependencies:
numpy:

IMPORTANT: PLEASE READ THIS FOR ADVICE ON HOW TO SOLVE THIS ISSUE!

Importing the numpy C-extensions failed. This error can happen for
many reasons, often due to issues with your setup or how NumPy was
installed.

We have compiled some common reasons and troubleshooting tips at:

    https://numpy.org/devdocs/user/troubleshooting-importerror.html

Please note and check the following:

  * The Python version is: Python3.9 from "C:\Program Files\MATLAB\R2022a\bin\win64\MATLAB.exe"
  * The NumPy version is: "1.20.3"

and make sure that they are the versions you expect.
Please carefully study the documentation linked above for further help.

Original error was: DLL load failed while importing _multiarray_umath: The specified module could not be found.


Error in <string>><module> (line 25)

From anaconda prompt, python is version 3.9 and numpy is version 1.20.3.

And, Matlab is connected to the anaconda

>> pyenv
ans = 

  PythonEnvironment with properties:

          Version: "3.9"
       Executable: "C:\Users\middl\anaconda3\python.EXE"
          Library: "C:\Users\middl\anaconda3\python39.dll"
             Home: "C:\Users\middl\anaconda3"
           Status: Loaded
    ExecutionMode: InProcess
        ProcessID: "17884"
      ProcessName: "MATLAB"

It looks like pandas is working.　Would you tell me why I cannot use numpy?

Answer 1

[Peter O]

To my knowledge, the Anaconda distribution remains an unsupported version of Python for MATLAB. Numpy fails with Anaconda because Anaconda maps the DLL's in a non-standard spot and adds them to the path when you start Anaconda's Prompts or Navigators. Pandas is straight Python. To be fast, numpy uses compiled C-code, hence those extra DLL's. The workaround I've used successfully in the past is to launch Matlab from an Anaconda prompt.

Again, this is an unsupported way to do things. I've noticed that matplotlib has a tendency to crash this way. Of course, MATLAB's plotting engine and tools are better :)

Steps:

1. Launch Anaconda Prompt or Anaconda Powershell Prompt
2. Type matlab at the prompt. Matlab will launch with extra environment variables configured by Anaconda's startup.
3. Test in MATLAB:

py.math.sqrt(2)
py.numpy.sqrt(2)

Hope this helps!

[Yasuyuki Hamanaka]

Thanks to you, the program worked well. I appreciate your help!!

[Luc VIGNAUD]

Worked also for me

but " unsupported way to do things" is somehow unsupportable from Mathworks !

[Peter O]

Yeah, my hunch would be that because the Anaconda distribution is bundled by a third party, which is neither Python Foundation nor MathWorks, it would add a lot of overhead to support and potentially interfacing with a third organization. For instance locating the source of a configuration or versioning bug just got twice as long. To be clear, you can achieve everything offered from the Anaconda distribution with stock python and pip. The Anaconda version adds ease of use for newcomers, data science specific package configuration, and built-in version/environment containerization.