Closed johnoel closed 5 years ago
In linad99/fvar_a50.cpp line 34, the computed mfexp values for x > 60 is
vtmp.elem_value(i)=exp(60.)*(1.+2.*x)/(1.+x);
and the incorrect computed derivative value is
dfv1(i)=dfvtmp(i)*exp(60.0)*(1./(1.+b*b));
from line 74 in adjoint function DF_dvmfexp.
The computed derivative should be
dfv1(i)=dfvtmp(i)*exp(60.0)*(1./((1.+b)*(1.+b)));
Dave confirmed it is okay. Merged ac4072bc4671e421130c75a5e193318006b76477
In linad99/fvar_a50.cpp line 34, the computed mfexp values for x > 60 is
and the incorrect computed derivative value is
from line 74 in adjoint function DF_dvmfexp.
The computed derivative should be