use characteristic speed +/-c for Bx in HLLEM

Issue #898 closed
Roland Haas created an issue

the attached patch (against trunk) adds a new parameter decouple_normal_Bfield which defaults to yes (and changes the default behaviour) to GRHydro's MHD code when divergence cleaning is used. Reasons for this are given eg in Mignonoe et al. arXiv:1001.2832v2 section 3.1. This patch allows us to run the rotor test without having to resort to the LLF flux formula (which also uses +/- characteristic speed but for all variables) and does not seem to cause any problems in the TOV tests that Philipp ran. Bruno says it helps with the cylindrical blast as well.

Ok to include in the release (given that it is tunable at runtime) even though we are in the "no major change" phase?

Keyword: GRHydro

Comments (5)

  1. Bruno Mundim
    • removed comment

    This is indeed a bug fix for the magnetized HLLE solver with divergence cleaning. Not only the rotor test improves but also preliminary tests of the cylindrical explosion shows now that it is possible to use HLLE solver for this stringent test.

    Please apply this patch.

  2. Log in to comment