Change variables in the base workspace through a Matlab Function block

Hi,

I'm trying to change a matrix (of type double) in the base workspace through a Matlab Function block in Simulink. To access the matrix in the base workspace within the Matlab function block, I have set the variable type in the Symbols pane of the Matlab function block to parameter data. Then I change the parameter and assign it to the base workspace with the assignin function, which I found in past solutions for the same problem.

Here is an example of my problem:

function updateBaseWorkspaceVariable(Workplan, TasksToDo)
    % Modify the base workspace variable using the parameters
    assignin('base', 'myVariable', Workplan + TasksToDo);
end

However when running my code and this example code I get the following error message:

Error:Function 'assignin' not supported for code generation.

Function 'MATLAB Function' (#343.124.176), line 3, column 5:

"assignin('base', 'myVariable', Workplan + TasksToDo)"

Launch diagnostic report.

Since this is a solution already used in past Matlab versions I'm wondering, if it's a problem within the Matlab version that I'm using (23a) or whether there is something I'm missing.

Thanks in advance for your help!

Kind regards

Lisa

Answer 1

[Jon]

I would suggest using the "Interpreted MATLAB Function" block for this purpose. It doesn't need to get compiled. It is under

Simulink>User-Defined Functions in the Library Browser.

In the Parameters field MATLAB function, enter something like myfun(u).

In a directory that is on the MATLAB path, make and save an .m file called myfun(u), that has your code it in it, e.g.

function y = myfun(u)
y = u^2;
assignin('base', 'y',y);

[Jon]

I have attached an example .slx model and associated .m file with the function to assign in the base workspace

[Jon]

You can also use the MATLAB function block using the coder.extrinsic function like this

function y = fcn(u)
coder.extrinsic("assignin")
y = u^2;
assignin("base","y",y)

[Jon]

Ohh, I guess our messages crossed, but I see now that @Harald has also suggested this, hopefully my example above makes the details clear

[Lisa P]

Hi Jon,

Thank you for the quick reply!

Both solutions work perfectly.

[Jon]

Hi Lisa, if you are all set, please accept an answer so others will know that an answer is available. Thanks

Answer 2

[Harald]

Hi Lisa,

you can declare a function as extrinsic to address the error message, using coder.extrinsic.

I would question the workflow of writing to the base workspace every time the block is executed. If you only need to have the variable in the workspace after the simulation, you can use model callbacks.

Best wishes,

Harald

[Lisa P]

Hi Harald,

Thank you for your quick reply as well!

Also thank you for the tipp on model callbacks. I might use those in other places. For my question the writing to the base workspace at every time of the blocks execution was intended. I'm using the table in the workspace as an input for a state chart. I want the table to be updated at certain discrete points in time to symbolise tasks being added to a To-Do-List. Then there is a state in the state chart that checks whether there are tasks in the list and when there are tasks the state 'working' is reached. Otherwise the state 'waiting' remains active. Therefore I wanted the list to be updated at every timestep.

My solution might not be the best, but I was just trying to make an idea work (:

Kind regards,

Lisa