spm_Gpdf function attempts to calculate log of zero

[TYH0517]

When running neurodesign, the process fails with the following error:

C:\Users\User\anaconda3\lib\site-packages\neurodesign\neurodesign.py:561: RuntimeWarning: divide by zero encountered in log
  s = np.array(s)

This corresponds to the spm_Gpdf function. In this, the line

res = (h - 1) * np.log(s) + h * np.log(l) - l * s - np.log(gamma(h))

appears to be trying to calculate the log of zero as the first value of the variable s is 0. s is generated in the canonical function from the line

s = np.array(range(int(np.ceil(p[6] / dt)))).

It is not clear why this behaviour occurs since it doesn't appear to be a problem for anyone else. I've altered the code slightly to make it compatible with python3, but there isn't an obvious link between those changes and this behaviour.

The input used was:

    TR = 2.0, 
     n_trials = 200, 
     P = [0.5, 0.5], 
     C = [[1.0, 0.0], [0.0, 1.0], [-0.5, 0.5], [0.5, -0.5], [1.0, 0.0], [0.0, 1.0], [0.5, -0.5]], 
     duration = None, 
     n_stimuli = 2, 
     rho = 0.3,  
     stim_duration = 2.0, 
     restnum = 40, 
     restdur = 20.0, 
     ITImodel = 'uniform', 
     ITImin = 2.0, 
     ITImean = 3.0, 
     ITImax = 4.0, 
     confoundorder = 10, 
     maxrep = 10, 
     hardprob = False, 
     t_pre = 1.0, 
     t_post = 1.0, 
)```

These values were obtained through the neurodesign web tool.

Any clues to what might be the problem would be appreciated!