Breaking news about Whisky

Last updated: 12th of January 2010

• The thorns EOS_Polytrope and EOS_Hybrid moved to CactusEOS.

Previous updates:

• A new public version saw the light of the day. This time, this includes some changes to the core part, which are likely to require changes in both, private thorns and parameter files. Here is a list of what you likely have to do after an update:
  • In general:
    • Whisky now uses TmunuBase and HydroBase (both from the CactusEinstein arrangement)
    • Whisky and HydroBase now use one variable group for each primitive and conserved variable, which means whisky_(prim|cons)_(scalar|vector)_var has to be replaced by the corresponding variable names
    • The following table summarizes changes in names and location of variables:

      old	new
      whisky::rho	hydrobase::rho
      whisky::eps	hydrobase::eps
      whisky::press	hydrobase::press
      whisky::velx	hydrobase::vel[0]
      whisky::vely	hydrobase::vel[1]
      whisky::velz	hydrobase::vel[2]
      whisky::sx	whisky::scon[0]
      whisky::sy	whisky::scon[1]
      whisky::sz	whisky::scon[2]

  • For parameter files:
    • Add the thorns TmunuBase and HydroBase to the list of active thorns (both from CactusEinstein).
    • Depending on your needs, add options for Tmunubase, e.g.:

      TmunuBase::stress_energy_storage = yes
      TmunuBase::stress_energy_at_RHS  = yes
      TmunuBase::timelevels            = 1
      TmunuBase::prolongation_type     = none

    • The same goes for HydroBase, e.g.:

      HydroBase::timelevels       = 3
      HydroBase::evolution_method = "whisky"

    • Look for use of old variable names and change them, e.g. in I/O parameters
  • For thorns:
    • Inherit from HydroBase or Whisky, being FRIEND to anything should not be necessary anymore
    • Look for use of old variable names and change them. Vector GFs are not yet nicely supported by Cactus, especially not for C/C++. You may want to have a look at Whisky_TOVSolverC/src/tov.c how the convertion can be done easily (but ugly for the moment)
• All the webpages have been updated and broken links have been fixed
• The webpages have gone through a bit of cleaning
• The Whisky Retreat IV was held
• We now have a mailinglist for questions/answers/announcements/discussions: users@whiskycode.org.
• We now have a mailinglist for questions/answers/announcements/discussions only among the developers: developers@whiskycode.org.
• The Whisky Retreat III is coming up, please register.
• Pictures from the conference call at the beginning of May are in can be found here.
• Pictures from the conference call at the end of February are in can be found here.
• Lots of Whisky talks at the Palma meeting.
• A new TOV solver thorn, Whisky_TOVSolverC, has been added to the repository. Written by Frank Löffler, it can do multiple boosted glued stars. Some binary NS simulations were performed that looked pretty good, but the system is totally unphysical.
• Using the new version of MoL plus modified versions of BSSN and Whisky, a stellar collapse run was performed using Carpet. Although spherically symmetric it was run without any symmetries and still managed to resolve the bounce and resulting NS correctly.
• Pictures from the conference call at the end of November can be found here.
• Pictures of the latest tests can be found here.
• A new thorn, Whisky_IVP, is added to the repository. This is a simple initial value problem solver using BAM written by Frank Löffler, Ian Hawke and Luca Baiotti.
• Joachim Frieben adds the analytic calculation of the left eigenvectors to the Roe and Marquina Riemann solvers. The compact flux formula for the Roe and Marquina solversare also added. This leads to considerable gains in efficiency.
• Whisky is altered to work with the new CactusEinstein arrangement (ADMBase and others). Auxilliary is replaced with ADMCoupling, and we no longer require a separate ADMConstraints thorn.
• Whisky passes TOV and migration tests, thanks to work by Luca Baiotti. See the pictures here.
• The Einstein thorn in the development Cactus distribution now includes timelevel support. This, combined with a few minor changes to Whisky and BSSN_MoL (no functionality changes) mean that we no longer need the branched version of Einstein.
• IMPORTANT: You will need to update a number of thorns for this to work. It would be safest to run gmake cvsupdate from your base Cactus directory to ensure that everything is updated. If this gives problems then the most important things to update are the flesh, PUGH, BSSN_MoL, MoL, Whisky and Whisky_Init_Data.
• The minor changes to thornlists and parameter files should be given by the examples on the downloads page.
• For the moment, the branched ADMConstraints thorn will remain, as will the (misspelt) Auxilliary thorn.
• ENO reconstruction added. This is arbitrary order, but the stencil width must be at least the order. Largely untested. Higher than second order gives poor results (slightly oscillatory) if reconstructing on the conserved variables.
• A few changes to MoL to remove warning messages on the Origin. No changes to functionality.
• Convergence tests of rotating stars seem to confirm second order convergence (TVD, Marquina, ICN). This is with the fully dynamical spacetime (static shift). This is slightly iffy due to the avoid origin problem, and these runs were performed without the constraints being output.
• Convergence tests of TOV stars confirm second order convergence using ICN or RK3 time evolution and either TVD or PPM reconstruction. RK2 give second order convergence in the Cowling approximation but not with a dynamical spacetime, reason unknown.
• Parameters added so that initial data can be specified just for the hydro. This allows Scott to do the shock tube on curved spacetime test, and Pedro the torus and Schwarzschild, and...
• Auxilliary thorn and branched ADMConstraints thorn added to EUHydro. Both will be merged straight back into CactusEinstein soon. Allows for standard constraint checking.
• EOS_Polytrope thorn added and additional EOS parameters and code to deal with polytropes. This probably isn't the most efficient way of implementing a polytropic EOS in Whisky, but it seems to work.
• BSSN_MoL will now do everything that ADM_BSSN would do, plus it couples to Whisky.

Impressum.