itsme - should i change STEPSIZEREDUCTIONFACTOR...
3 Replies
huh what solver is that in?
oh is that the path simplification thing?
i mean you can adjust parameters all day but the bug will just show up somewhere else, parameters just hide the problem by changing the process
i would look deeper to figure out why the bug is happening, something might be "going negative" 🤷
i think the bug is because the solver algorithm assumes that the feasibility of routing in one go is monotonic. so in this case, it just skips the region where it can route it in one go. should i make a non-multiplier cap for how much it can reduce step size so that it doesn't skip things basically adding a new parameter?
or should i make it determine the reduction factor based on the situation?
sorry which solver is this? Is it path simplification?
yea that makes sense to me RE: step size having an absolute minimum i think
i think i understand what you're saying, basically the algorithm itself may not be sound