Brainpy array does not support "where" argument in division

[CloudyDory]

• [x] Check for duplicate issues.
• [x] Provide a complete example of how to reproduce the bug, wrapped in triple backticks like this:
• [x] If applicable, include full error messages/tracebacks.

I am trying to implement the function in Brainpy: x / (1.0 - bm.exp(-x)). When x->0, the function has a theoretical limit of 1.0, but if we directly set x=0, the function will output NaN.

In Numpy, this can be handled by: np.divide(x, 1.0-np.exp(-x), out=np.ones_like(x), where=np.abs(x)>=1.0e-6).

However, in Brainpy, the similar expression does not work: bm.divide(x, 1.0-bm.exp(-x), out=bm.ones_like(x), where=bm.abs(x)>=1.0e-6). The error message is: TypeError: true_divide() got an unexpected keyword argument 'where'.

How can we deal with situations like this in Brainpy?

[chaoming0625]

Thank you for raising this missing functionality in brainpy.

Unfortunately, brainpy.math does not support where argument currently. We will add this in the near future.

However, this function can be rewritten to:

def f(x):
  return bm.where(bm.abs(x)<1.0e-6, 
                             bm.ones_like(x),
                             x / (1.0 - bm.exp(-x)))

Hoping this message will fix this issue.