Closed jondea closed 6 years ago
I think choosing a seed makes more sense. Then we can still be rather stringent on the boundary test.
I agree we should choose a seed eventually, but it worries me slightly that it fails so often. Do you know where it is failing?
Displacement jumps are suppose to be strictly decreasing in k, but here's an example where it ever so slightly isn't: 0.00013567, 0.000133472
Pretty minimal tbh.
I've solved this by using one specific seed to generate the random particles. I forgot to tag this issue in my commit...
The boundary conditions test uses randomly generated particles, and this sometimes cause the test to fail (example). We need to make the test more permissive (but still strong), or alternatively use a seed which is known to pass.