Python numpy Runtime warning using np.where np.errstate and warnings 'error'

Question

I keep getting the runtime warning about division. In the code below, I used the answer of a question from this forum and even imported warnings error from this.

def alpha_n(V):
    with np.errstate(divide='ignore'):
        alph = np.where(V!= -55, 0.01*(V+55)/(1-np.exp(-0.1*(V+55))), 0.1)
    return alph

RuntimeWarning: invalid value encountered in true_divide

How could I define the function properly to avoid warnings?

What version of python do you use? Also, could you provide an example of data you send through your function alpha_n — Christian
– Christian, Commented Jan 19, 2021 at 17:11
Using np.where to avoid bad values is not a good idea. They are still evaluated. An alternative is use np.divide ufunc with its own where (and out) parameters. — hpaulj
– hpaulj, Commented Jan 19, 2021 at 17:39

hpaulj · Accepted Answer · 2021-01-19 19:53:41Z

1

In [26]: def alpha_n(V):
    ...:     with np.errstate(invalid='ignore'):
    ...:         alph = np.where(V!= -55, 0.01*(V+55)/(1-np.exp(-0.1*(V+55))), 0
    ...: .1)
    ...:     return alph
    ...: 
In [27]: alpha_n(np.array([1,2,3,-55]))
Out[27]: array([0.56207849, 0.5719136 , 0.58176131, 0.1       ])

Divide by 0 is different from invalid value:

In [28]: 1/(1-np.exp(-0.1*(np.array([-55])+55)))
<ipython-input-28-ed83c58d75bb>:1: RuntimeWarning: divide by zero encountered in true_divide
  1/(1-np.exp(-0.1*(np.array([-55])+55)))
Out[28]: array([inf])
In [29]: 0/(1-np.exp(-0.1*(np.array([-55])+55)))
<ipython-input-29-0d642b423038>:1: RuntimeWarning: invalid value encountered in true_divide
  0/(1-np.exp(-0.1*(np.array([-55])+55)))
Out[29]: array([nan])

answered Jan 19, 2021 at 19:53

hpaulj

233k14 gold badges260 silver badges392 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Python numpy Runtime warning using np.where np.errstate and warnings 'error'

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related