Python-EXP中遇到的溢出

Question

I am using a formula in a model I'm developing, which, when it runs produces RuntTimeWarning: overflow encountered in exp. My understanding from what I've read is that this due to excessively large numbers being used in the calculation for the data type.

My formula stores to a numpy array, so is there a method for changing the datatype so that these values can be computed?

Thanks

Answer 1

This sounds like an XY problem .

While higher precision datatypes could possibly solve this issue, this often comes at a performance and usability cost. This is not possible with numpy directly, but can be done with mpmath. More details can be found in this question .

Before resorting to such measures:

1. Verify that these large numbers are actually correct behavior and do not result from a bug further up.

2. See if it is possible to log-transform your formulas so that large numbers are avoided.