Home/ Questions/Solving the error runtimewarning: invalid value encountered in double_scalars
Next
Answered
Alice Antoine
  • 23
Asked: 2022-05-11T18:35:28+00:00 In:

Solving the error runtimewarning: invalid value encountered in double_scalars

  • 23

. Advertisement .

..3..

. Advertisement .

..4..

I’m trying to run a new project. I do a couple of things like this:

import numpy
 
 d = numpy.array([[1089, 1093]])
 e = numpy.array([[1000, 4443]])
 answer = numpy.exp(-3 * d)
 answer1 = numpy.exp(-3 * e)
 res = answer.sum()/answer1.sum()
 print res

but in my program, I am getting the warning:

nan
 C:\Users\Desktop\test.py:16: RuntimeWarning: invalid value encountered in double_scalars
  res = answer.sum()/answer1.sum()

Can someone explain to me why the issue of the runtimewarning: invalid value encountered in double_scalars happened? Where have I gone wrong? Thank you!

runtimewarning
Share

2 Answers

  1. Best Answer
    Expert

    This is how I solve this problem using math manipulation. That is for the numerator.

    exp(-x)+exp(-y) = exp(log(exp(-x)+exp(-y)))
                = exp(log(exp(-x)*[1+exp(-y+x)]))
                = exp(log(exp(-x) + log(1+exp(-y+x)))
                = exp(-x + log(1+exp(-y+x)))

    Where x=3* 1089 and y=3* 1093 are respectively. The argument for this exponential is

    -x + log(1+exp(-y+x)) = -x + 6.1441934777474324e-06

    You could also use the same procedure for the denominator, but log(1+exp(-z+k)) has already been rounded up to 0. This means that the argument to the exponential function at denominator can be simply rounded up to -z=-3000. Now you have your result.

    exp(-x + log(1+exp(-y+x)))/exp(-z) = exp(-x+z+log(1+exp(-y+x)) 
                                   = exp(-266.99999385580668)

    This is very close to what you would get if only the two leading terms were kept (i.e. The numerator’s first number 1089 and the denominator’s first number 1000 are the 1089 codes.

    exp(3*(1089-1000))=exp(-267)

    Let’s take a look at how close we are to the solution to Wolfram alpha ( Link).

    Log[(exp[-3*1089]+exp[-3*1093])/([exp[-3*1000]+exp[-3*4443])] -> -266.999993855806522267194565420933791813296828742310997510523

    This difference is +1.7053025658242404e-13. The approximation that we made at denominator was therefore fine.

    The final result is

    'exp(-266.99999385580668) = 1.1050349147204485e-116

    From Wolfram Alpha is ( Link).

    1.105034914720621496.. × 10^-116 # Wolfram alpha.

    It is also safe to use numpy.

    • 17

  2. You can solve your problem now by:

    • You can choose a library that specializes in high-precision mathematics such as mpmath. This is less fun.
    • As an alternative to a larger weapon, you can do math manipulation as described below.
    • You can get a scipy/numpy function that is tailored to your needs
    • 0

Leave an answer

You must login to add an answer.

Need An Account,