search

tapios / risk-networks

Code for risk networks: a blend of compartmental models, graphs, data assimilation and semi-supervised learning
Other
2 stars 2 forks source link

Numba seed/reproducibility issue #70

Closed lubo93 closed 4 years ago

lubo93 commented 4 years ago

Running the same epidemic simulation twice doesn't produce the same output. As an example, I used "super_simple_epidemic.py" and obtained

super_simple_epidemic

and (for the same parameters):

super_simple_epidemic1

The origin of this issue may be the numba parallelization routines in "ContactSimulator", as pointed out by @glwagner:

https://github.com/numba/numba/issues/3249