Francesco Casella
Francesco Casella
@FranceHub we also recently implemented a diagnostic method to identify bad start values. The method is described in [this paper](https://arxiv.org/abs/1911.12433), also published on [Applied Mathematics and Computation](https://doi.org/10.1016/j.amc.2021.125991). I presented a...
> Even if it breaks the “problem-solver separation principle” I don't think it actually does it. min-max limits are declarative. We'll definitely look into this. Unfortunately at the moment we...
> Related to #12193 Definitely, but there's more to it. Not only we should select good initial points, but also ensure that each iteration does not bring the solution estimate...
Discussion continues on #14104.
@arun3688 please have a look ASAP.
> Does this wrong RHS calculation also affect ME FMUs? It shouldn't, the ME FMUs do not call any ODE solver internally, integration is performed by the caller.
Also see OpenModelica/OpenModelicaLibraryTesting#193, we should definitely test this all the time.
OK, we just need to wait for @arun3688 to get back to work
Adding @matteodepascali to the loop
@arun3688 this would be a very important fix. If @bricolo100 is happy with it, I would recommend to port f77c4b0 to maintenance/v1.25 so we get this in 1.25.0 or in...