Matlab ODE function solving using Python Scipy library.

Question

Simon Mwakitabu on 13 Feb 2024

1
Link

Direct link to this question

https://www.mathworks.com/matlabcentral/answers/2081553-matlab-ode-function-solving-using-python-scipy-library

Commented: Simon Mwakitabu on 22 Apr 2024 at 8:39

I would like to solve a system of stiff ODE equations from Matlab with other methods from the scipy library such as LSODA or Sundials rather than Rosenbrock (ode23s).

A matlab ODE complex function defined as;

function [dpdt]=damper_rebo(t,P,amp,freq,pars)
.....

with amp, frequency constants and pars a cell of parameters with constants parameters.

It works fine with;

tic
[tsol,ysol]=ode23s(@(t,P) damper_rebo(t,P,amp,freq,pars),[0 0.5/freq],P_0,options);
toc

I am trying to compare with other methods to solve ODE with stiff equations like LSODA and BDF.

I loaded the python environment and installed all the path requirements, then I tried this:

py.importlib.import_module("scipy")
%--Call out the constant parameters from the function--%%
pars=params7();
pyrun("from scipy.integrate import LSODA")
%-----------------------------------------%
%---Excitation Signal Properties of the Damper---%
amp=25.25e-3;                            % Modal excitational amplitudes (m)
v_n=0.151;
freq=v_n/2/pi/amp;                               % Modal frequency (Hz)
%-------------------------------------%
%--Initial Pressure conditions--%
Pg_0=pars{68};
P_0=[0.01,0.001,1e5,2e5,Pg_0];           % At equilibrium three chambers at the same pressure
%-------------------------------%
% %-------------------------------%
%--Atmospheric Surrounding Temp and Pressure--%
T_atm=15;                                        % Fluid Initial Properties by the supplier
p_atm=101325;
%---------------------------------------------%
Tspan = py.list([0.0, 0.5/freq]);
pyfun = py.str('lambda t,P: damper_rebo(t,P,amp,freq,pars)');
or
pyfun = py.function_handle(@(t,P) damper_rebo(t,P,amp,freq,pars));
sol=py.scipy.integrate.solve_ivp(pyfun,Tspan, py.numpy.array(P_0),pyargs(method="BDF", rtol=0.00001, atol=1e-06));

The first `pyfun`option brings an error that `

Python Error: TypeError: 'str' object is not callable`

The second option brings an error too with

`Unable to resolve the name 'py.function_handle'.`

1. How can I pass the Matlab ODE function to a callable python function?

2. How can I pass the extra arguments in a python manner?

0 Comments
Show -2 older commentsHide -2 older comments

Sign in to comment.

Sign in to answer this question.

Answer 1

Tejas on 5 Apr 2024 at 5:39

1
Link

Direct link to this answer

https://www.mathworks.com/matlabcentral/answers/2081553-matlab-ode-function-solving-using-python-scipy-library#answer_1436881

Edited: Tejas on 5 Apr 2024 at 5:42

Hello Simon,

It appears from the code shared above, that you are attempting to pass a MATLAB function handle as an input argument to a Python function, which has led to the two issues mentioned.

In the first issue, ‘py.str’ converts the function call into a string, which subsequently renders it uncallable from any function. The second issue arises due to an attempt to pass a MATLAB function handle as an input argument to a Python function, a process for which there currently isn't a direct method. For further understanding of what is supported and what isn't when interacting with Python from MATLAB, please refer to the following documentation: https://www.mathworks.com/help/releases/R2022b/matlab/matlab_external/passing-data-to-python.html .

This documentation also includes details on how MATLAB variables are interpreted as Python variables, offering guidance on how to pass input arguments when calling a Python function from MATLAB.

To address the above two issues, I suggest the following workaround:

Create a Python function that calls the ‘damper_rebo’ function using the MATLAB Engine API.
Pass this Python function as an input argument to the ‘solve_ivp’ function.