MATLAB engine API for Python¶
As SystemLab|Design is based on the Python programming language, it is possible to call MathWorks MATLAB ® as a computational engine from a functional block script. After creating an instance of the MatlabEngine class you can call MATLAB functions and algorithms and interchange data between the Python script and MATLAB workspace.
The following procedure (for Windows 10) describes how to setup and run the MATLAB ® Engine API for Python ® .
To run the MATLAB Engine API for Python you must have a supported version of MATLAB R2014A or later running on the same operating system as where SystemLab|Design has been installed.
Additional information on the use of this feature and its installation can be found on the MathWorks ® web site: Calling MATLAB from Python.
How to install the MATLAB Engine API for Python for SystemLab|Design
Launch the Windows command prompt window by entering in the search field from the taskbar command or cmd
Below the Command Prompt select Run as administrator.
From the command prompt type:
Note: “path_to_matlab_root_dir” is for example “C:\Program Files\MATLAB\R2019a”.
From the command prompt type:
python setup.py install --prefix="C:\path_to_systemlab_application\systemlab_design"
Note: “path_to_systemlab_application” is the location where the “systemlab_design” folder is installed on your computer.
To verify that the MATLAB API was correctly installed check that there is a new folder matlab under the …\systemlab_design\Lib\site-packages folder.
Select the folder matlab and drag and drop it to the folder systemlab_design. [This step is important as it will ensure that the matlab.engine can be located and successfully imported from any functional block script.]
Test that the MATLAB engine can be called from a functional block script
Launch a new application of SystemLab|Design by double left-clicking on the SystemLab-Design.exe executable file.
Add a new functional block to the default project scene with the following settings:
Add one port to the functional block with the following settings: Port name: ‘Out’, Cardinal point: East, Direction: Out, Signal type: Electrical (remember to click Apply before saving and closing the Add ports dialog).
Functional block name/ID: MATLAB Test
Script module name: Matlab_Run
Open a session of SciTE (or equivalent) by selecting Edit/Open code/script editor.
Within the editing panel (under “1 Untitled”) copy and paste the code as shown in the box below.
Note: In the code below we initiate a new MATLAB process by calling eng = matlab.engine.start_matlab(). We then build an array of ones within the MATLAB workspace and access this array using the command sig_array = matlab.double(eng.ones(1,n)).
To ensure that the array matches the Numpy array format we use the command sig_array = np.asarray(sig_array) to convert the Python list to a Numpy array. The data from this array is then returned to the simulation algorithm and allocated to the output port of the functional block.
import numpy as np import config import matlab.engine def run(input_signal_data, parameters_input, settings): '''==PROJECT SETTINGS===================================================''' module_name = 'MATLAB Test' n = settings['num_samples'] n = int(round(n)) time = settings['time_window'] fs = settings['sampling_rate'] carrier = 0 sig_type = 'Electrical' parameters_input =  '''==CALCULATIONS=======================================================''' time_array = np.linspace(0, time, n) noise_array = np.zeros(n) if config.sim_status_win_enabled == True: config.sim_status_win.textEdit.append('Starting MATLAB Engine... ') config.app.processEvents() # Initiate a new MATLAB process eng = matlab.engine.start_matlab() # Build array (within MATLAB workspace) and access locally sig_array = matlab.double(eng.ones(1,n)) # Change array from python list format to numpy array format sig_array = np.asarray(sig_array) # eng.desktop(nargout=0) if config.sim_status_win_enabled == True: config.sim_status_win.textEdit.append('Quiting MATLAB Engine... ') config.app.processEvents() eng.quit() matlab_results =  '''==RETURN (Output Signals, Parameters, Results)==================================''' electrical_signal = [1, sig_type, carrier, fs, time_array, sig_array, noise_array] return ([electrical_signal], parameters_input, matlab_results)
Save the file as Matlab_Run.py within the folder systemlab_design (make sure to include the suffix .py in the File Name field when saving) and close the session of SciTE.
Start running a simulation [The simulation should complete as shown below]
Double left click on the output port of MATLAB Test to verify that an output signal is created as shown below:
About the MATLAB engine API for Python…
As shown partially in this example, you can call any MATLAB function or a custom script (m-file) directly from the functional block Python script. Data arguments can be passed from Python to the MATLAB engine (in this example we have used the number of samples ‘n’) and returned by declaring local variables in Python (in this example we used ‘sig_array’). Also, any data types within the MATLAB workspace can be accessed through the command a = eng.workspace(‘data’) where ‘a’ is the local variable in Python and ‘data’ is a data variable located within the MATLAB workspace.
Further information on how to exhange data between Python and MATLAB can be found on the MathWorks ® web site: Calling MATLAB from Python.
NOTE: To initiate a MATLAB process from within a functional block script the import matlab.engine statement MUST be included at the beginning of the script. Also, once calculations linked to the MATLAB process have been completed, the active session of the MATLAB engine MUST be closed by using the eng.quit() command. If the process is not terminated, the functional block script will remain in a suspended state!