[Flexiblesusy-commits] [FlexibleSUSY/FlexibleSUSY] e7a9be: rename configure option for mass eigenvalue check

[GitHub]

  Branch: refs/heads/feature-diagonalization-error
  Home:   https://github.com/FlexibleSUSY/FlexibleSUSY
  Commit: e7a9be313b3f38fde6b803f1cb1d6b8022fec509
      https://github.com/FlexibleSUSY/FlexibleSUSY/commit/e7a9be313b3f38fde6b803f1cb1d6b8022fec509
  Author: Alexander Voigt <alexander.voigt at desy.de>
  Date:   2014-10-23 (Thu, 23 Oct 2014)

  Changed paths:
    M configure

  Log Message:
  -----------
  rename configure option for mass eigenvalue check


  Commit: 6e6577895dce6f91e80b322cce15921dd6e1a2d2
      https://github.com/FlexibleSUSY/FlexibleSUSY/commit/6e6577895dce6f91e80b322cce15921dd6e1a2d2
  Author: Alexander Voigt <alexander.voigt at desy.de>
  Date:   2014-10-23 (Thu, 23 Oct 2014)

  Changed paths:
    M meta/TreeMasses.m

  Log Message:
  -----------
  correcting mass eigenvalue error check

On 10/23/2014 08:05 PM, Jae-hyeon Park wrote:
> The error bound returned by fs_*() is an (approximate) upper limit on
> abs(MSd[i] - MSd_true[i]) common to all i.  Since this error is
> independent of i, a large hierarchy among the eigenvalues means a large
> relative error in the smallest eigenvalue.  I think this is what the
> referee said in his comment A-4-a.  Therefore, I would suggest checking
> the relative error of the eigenvalue with the smallest absolute value in
> the above code, e.g. like
>
>      eigenvalue_error > precision * std::abs(MSd[0]).
>
> This would also match the mass dimensions of both sides of the
> inequality, if precision is dimensionless.


Compare: https://github.com/FlexibleSUSY/FlexibleSUSY/compare/903db61c29ef...6e6577895dce