[Flexiblesusy-commits] [FlexibleSUSY/FlexibleSUSY] fdeb81: adding setter for scale_getter function

[GitHub]

  Branch: refs/heads/feature-2.0
  Home:   https://github.com/FlexibleSUSY/FlexibleSUSY
  Commit: fdeb81d7276aefd6245d7e0b25453a4ee05c0a4d
      https://github.com/FlexibleSUSY/FlexibleSUSY/commit/fdeb81d7276aefd6245d7e0b25453a4ee05c0a4d
  Author: Alexander Voigt <Alexander.Voigt at desy.de>
  Date:   2016-12-02 (Fri, 02 Dec 2016)

  Changed paths:
    M src/two_scale_convergence_tester_drbar.hpp

  Log Message:
  -----------
  adding setter for scale_getter function


  Commit: 676a662e786159e3ab1b9f2a06f2314de3daa556
      https://github.com/FlexibleSUSY/FlexibleSUSY/commit/676a662e786159e3ab1b9f2a06f2314de3daa556
  Author: Alexander Voigt <Alexander.Voigt at desy.de>
  Date:   2016-12-02 (Fri, 02 Dec 2016)

  Changed paths:
    M templates/two_scale_high_scale_spectrum_generator.cpp.in
    M templates/two_scale_low_scale_spectrum_generator.cpp.in

  Log Message:
  -----------
  optimize convergence tester for FlexibleSUSY

The two-scale solver checks for convergence after all sliders have
been run.  In FlexibleSUSY the last slider stops at the SUSY scale.
This is the default scale where we want to calculate the spectrum and
thus the scale where we want to check for convergence.  Thus, by
default we don't need to specify a specific scale where to check for
convergence.

If we still do it, we decrease performance, because the SUSY scale
changes from iteration to iteration (if it is a dynmamic scale).
Thus, in every iteration we need to run all parameters to the new SUSY
scale inside the convergence tester (which has changed a bit).  This
leads to a performance drop.


Compare: https://github.com/FlexibleSUSY/FlexibleSUSY/compare/b44c4ca909fe...676a662e7861