provide a means for fixing the seed of random number generator in various functions

tulip-control / polytope

Geometric operations on polytopes of any dimension

https://pypi.org/project/polytope

Other

74 stars 19 forks source link

provide a means for fixing the seed of random number generator in various functions #13

Open necozay opened 10 years ago

necozay commented 10 years ago

for reproducibility... In particular, for the following functions:

esp.py
quickhull.py
glpk (not sure about the source of randomness but we have observed inconsistencies in the output for multiple repeated runs - glpsol seems to accept a seed value)

necozay commented 8 years ago

projection_iterhull

(see also related tulip issue)

slivingston commented 8 years ago

Is this a "bug"?

Fixing the seed is crucial for many tests, so providing the feature requested in this issue is well motivated. However, is randomization being used wrongly somewhere?

johnyf commented 8 years ago

The "inconsistencies" mentioned in the OP may have led to this characterization.