Return two variables with different data types from MATLAB engine in Python

I am analyzing oscilloscope .csv files with Python. I have a signal that has one or two negative peaks (baseline is around 0 value), depending on the waveform analyzed. I use inside Python a MATLAB engine to call a MATLAB file that contains a function to find those peaks and return their position on the x (time) and y (amplitude) axis.

Inside MATLAB, the two cases (one / two peaks) give (for one variable, be it x or y; i will exemplify with x in this post) in the workspace a 1x1 double (eg: 0.020000000000000) or a 2x1 double ([0.016000000000000;4.628000000000000]) for one/two peaks (values and representation are copied from the workspace). So, for one waveform, MATLAB identified a peak at 0.02 seconds ; and for another waveform two peaks at 0.016 seconds (one of them) and at 4.628 seconds (the other one). All my .csv files contain one OR two peaks, and I need to study them further, so I have no way of knowing how many peaks a waveform has.

The MATLAB peak finder function has the following structure:

    function [times, amplitudes] = peak_find_matlab(file, x, y)

where x and y come from the file and denote the oscilloscope times and amplitudes for each individual point (~10000 points per array). In Python I call the MATLAB engine and call the function as such:

    eng = matlab.engine.start_matlab()
    [times, amplitudes] = eng.peak_find_matlab(file, x_list, y_list, nargout=2)

For the cases where two peaks are found, everything is fine, the engine returns the following:

    type times=  <class 'matlab.mlarray.double'> with value =  [[0.016],[4.628]]
    type amplitudes =  <class 'matlab.mlarray.double'> with value =  [[-0.16],[-0.062]]

I convert the 'matlab.mlarray.double' into a list to further use it in the Python script and other scripts, using:

    np.array(times._data).tolist()
    np.array(amplitudes._data).tolist()

In the case of one peak, MATLAB engine returns a 'float' type:

    type times=  <type 'float'> with value =  0.02
    type amps =  <type 'float'> with value =  -0.0792016896

Trying to convert this to list, yields different errors ( AttributeError: 'float' object has no attribute '_data', or 'float' object has no attribute '_getitem__'_ and IndexError: too many indices for array) if I try to use the float data type or try to convert it to simple numpy array or accessing any element of this array...

I am trying to understand why MATLAB outputs different data types for this operation and how to tell him to keep consistent with, for example 'matlab.mlarray.double' class which yields the necessary results. I would assume I will have, in the case one waveform has one peak, to tell MATLAB to output one value, and an empty value for the other one, something like: [[0.016],[]]. If so, how can this be done?

For consistency, the rest of my python code, where I plot a vertical line at each time value for the peaks and fill a histogram, depending on whether the script found a file with one peak (which I call primary event), or a file with two peaks (for which the first peak is coined a "primary" and the second one is a ... "secondary" :D).

secondaryTimes = []
primaryAmplitudes = []
secondaryAmplitudes = []

eng = matlab.engine.start_matlab()
[times, amplitudes] = eng.peak_find_matlab(file, x_list, y_list, nargout=2)           
peakTimes = np.array(peaks._data).tolist() 
peakAmps = np.array(amplitudes._data).tolist()
for peak in peaks._data: plt.axvline(x=peak, color='r', linestyle='-.')    # plot a vertical line where the peak was found
primaryAmplitudes.append(peakAmps[0])  
### need more than two values in peakTimes, the second value (index 1) is the time at which the secondary occurred
if len(peakTimes) > 1:
     secondaryTimes.append(peakTimes[1])
     secondaryAmplitudes.append(peakAmps[1])
if len(primaryAmplitudes) > 0:
     making_some_nice_histogram   # "highest amplitude" for primaries
if len(secondaryTimes) > 0:
     making_another_nice_histogram   # "timing histogram", in case there were secondary discharges

Thank you very much for your time!

Answer 1

[Vlad Cristescu]

You could change peak_find_matlab or do a wrapper over it in order to check the type of the returned values.

myVar1 = []
myVar2 = []
[times, amplitudes] = eng.peak_find_matlab(file, x_list, y_list, nargout=2)
if isinstance(times, float):
    myVar1.append(times)
else:
    myVar1=times

if isinstance(amplitudes, float):
    myVar2.append(amplitudes)
else:
    myVar2 = amplitudes

And then you continue your program using myVar1 and myVar2.

Hope this solves the problem