Sameer Agarwal

@jiapei-nexera we are working on fixing our build issues. cc: @sergiud

Yes i am aware of that. On Fri, Jun 17, 2022, 2:24 PM jiapei-nexera ***@***.***> wrote: > @sandwichmaker > > By the way, did you notice that > https://github.com/DrTimothyAldenDavis/SuiteSparse also...

is this issue still open?

@tipf you are right that the documentation for this is not great. In part because I have been the only user of this API for internal debugging. Would you care...

Sorry about the delayed reply. The reason we do the dumping inside the trust region strategies rather than inside the trust region minimizer is because to be able to debug...

Thank you @EXing I will take a look.

I see what is happening, yes currently ceres solver removes the constant parameter blocks from the ordering. This happens at https://github.com/ceres-solver/ceres-solver/blob/master/internal/ceres/trust_region_preprocessor.cc#L182.

cc: @DmitriyKorchemkin

Thats a fair observation, there is no rush to remove anything I was just hoping for simplification. I think we can still get rid of the no_threads backend. Lets come...

The discussion and benchmarks on https://ceres-solver-review.googlesource.com/c/ceres-solver/+/23140 indicates that at this point the difference between c++ threads and openmp is not particularly large and we will reduce complexity by making it...