adam-sim-dev

I have not found a "perfect" solution yet. Well, I do find that for some models (I set beta to a small value, not zero), AMS appears to be indefinite...

> > AMS appears to be indefinite or singular. Is that expected? > > Not if the original problem is symmetric and $\beta > 0$. The reason could be non-convergent...

You can use CUDA_FLAGS = -x=cu --expt-extended-lambda -arch=$(CUDA_ARCH) -allow-unsupported-compiler in user.mk if using make. I have not tested CMake.

> @adam-sim-dev , I am assuming there is need for this? If so I can revive the branch and work on in a bit to get it ready I work...

> Another option is `mem.pages`, which records the virtual memory size and RSS as reported by Linux. It's not perfectly accurate but typically lower-overhead, and should at least give you...

Hi @daboehme, what I want to do is like: ``` char* caliper_config = "mem.pages"; CALI_MARK_BEGIN("My code block"); # My code block CALI_MARK_END("My code block"); ``` to record the memory usage...

Thanks @mlstowell for the great addition! I use complex operations frequently. It would be great if add and subtract functions could be implemented for ComplexVector too . https://github.com/mfem/mfem/blob/7aa327174abea231cf3136a58d68594e0349df4f/linalg/vector.hpp#L373-L390