def _decide(self, metrics):
cv = to_numpy([e["cv"] for e in metrics])
delta_cv = to_numpy([e["delta_cv"] for e in metrics])
n_feasible = (cv <= 0).sum()
# if the whole window had only feasible solutions
if n_feasible == len(metrics):
return False
# transition period - some were feasible some were not
elif 0 < n_feasible < len(metrics):
return True
# all solutions are infeasible
else:
return delta_cv.max() > self.tol
def _decide(self, metrics):
delta_f = to_numpy(metrics)
return delta_f.max() > self.tol
def _decide(self, metrics):
return to_numpy(metrics).mean() > self.tol