regarding low order elements (order<2)

[sungho91]

Hi all,

I have a quick question
Can we make Remhos work for low-order elements like (order<2)?

Best,

Sungho

[vladotomov]

@sungho91 it's supposed to work with -o 1, and even probably with -o 0.

[sungho91]

@vladotomov

Really? If I command line like this mpirun -np 8 remhos -m ./data/cube01_hex.mesh -p 10 -rs 1 -o 1, it does not work.
Also, I turned off the high-order solver by setting zero for the option and tried every option for the low-order solver, but it eventually was terminated.

[vladotomov]

@sungho91 you are correct, there are some inaccuracies in the setup.

Please try #45 (branch fix-order-less-2) and let me knows if it works.

[sungho91]

@vladotomov

I used mpirun -np 8 remhos -m ./data/cube01_hex.mesh -p 10 -rs 2 -o 1 -dt 0.02 -tf 0.8 -ho 1 -lo 1 -fct 2 -visit or mpirun -np 8 remhos -m ./data/cube01_hex.mesh -p 10 -rs 1 -o 2 -dt 0.02 -tf 0.8 -ho 1 -lo 1 -fct 2 -visit and got those result. They look similar so, I think it wroks correctly.

By the way, since Remhos has some options for choosing solvers, do we have the best options for running Remhos as a rule of thumb?

[vladotomov]

By the way, since Remhos has some options for choosing solvers, do we have the best options for running Remhos as a rule of thumb?

For orders up to three, with standard matrix-based assembly, I'd say the best (sharpest / most robust / most tested) option is
-ho 3 -lo 2 -fct 1

[sungho91]

@vladotomov thank you very much