Using 'parsim' with a group of parameters

Good afternon, i have 3 tables with a set of parameters (each table has a different number of parameters) that I want to simulate in parallel with the parsim command. I have managed to extract the data and create the groups to simulate but at the moment of using parsim, either all the commands do not appear or they appear mixed.

I still don't seem to fully understand how to use the commands to simulate in parallel.

The parameters are fixed, i.e. they are not a range.

Thanks in advance

Answer 1

[Vijeta]

Hi,

Here's an example of how you can use `parsim` to simulate different sets of parameters in parallel using MATLAB.

Assuming you have three tables of parameters, `table1`, `table2`, and `table3`, each with a different number of parameters and the function you want to evaluate for each set of parameters is `myFunction`:

% Initialize tables of parameters 

table1 = readtable('table1.csv');
table2 = readtable('table2.csv'); 

table3 = readtable('table3.csv'); 

% Combine tables into a cell array of tables 

paramTables = {table1, table2, table3}; 

 % Define the number of simulations to run 

numSims = length(paramTables); 

 % Initialize results variable 

results = cell(numSims, 1); 

% Create parallel pool 

parpool(); 
% Evaluate function for each set of parameters in parallel 

parfor i = 1:numSims 

    results{i} = myFunction(paramTables{i}, otherInputs); 

end 
% Close parallel pool 

delete(gcp); 

In this example, the `readtable` functions load data from CSV files into MATLAB tables. These tables are then combined into a cell array called `paramTables` which serves as input to the `parsim` function.

The `parfor` loop evaluates the `myFunction` function for each set of parameters in parallel, where `otherInputs` is a variable containing any additional inputs required by the function.

The results are stored in a cell array called `results`. Note that the results are stored as a cell array because each set of parameters may have a different number of output variables or different sizes of output variables.

Finally, the parallel pool is closed with the `delete(gcp)` command.