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

[necozay]

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]

• projection_iterhull

(see also related tulip issue)

[slivingston]

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]

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