Different results from python GetPFT and scuff-scatter

[jmllorens]

Hi Homer,

I found the new interface in libscuffSolver very useful and easy to handle. I am starting to migrate some of my models from command line to python, and I found a problem with the output of GetPFT which differs from that of scuff-scatter. The scattered power is the same, but the absorbed power and rest of magnitudes differ. I have been browsing the source code, but I haven't found any remarkable difference between the different calls.

In the attached zip file, there is a small example with a simple mesh file, the scuffgeo file, Args and a short python script.

Any help is welcome.

José.

test.zip

[jmllorens]

I think I have finally found the problems. The first one is that CachedIF in scuffSolver.cc always gets a copy of a PointSource. The second one is that the copy method of PlaneWave and PointSource does not copy the attributes of the base class IncField.

1. In scuffSolver.cc:684 the copy of the source to CachedIF takes place:

   else if (IF)
   { G->AssembleRHSVector(Omega, IF, KN);
    if (CachedIF){
      delete CachedIF;
    }
    PlaneWave *PW = (PlaneWave *) IF;
    if (PW) CachedIF=new PlaneWave(*PW);
    PointSource *PS = (PointSource *) IF;
    if (PS) CachedIF=new PointSource(*PS);
   
   }

   The two if clauses are always true and CachedIF ends up with a copy of PS.

2. The copy method of PlaneWave in PlaneWave.cc:50:

   PlaneWave::PlaneWave(const PlaneWave &PW)
   { 
   memcpy(E0, PW.E0, 3*sizeof(cdouble));
   memcpy(nHat, PW.nHat, 3*sizeof(double));
   SetRegionLabel(PW.RegionLabel);
   }

   does not include a copy of the IncField attributes. Therefore, Omega is equal to -1 +0j. The same problem happens for PointSource.

It would be nice to find a nice solution for both problems. I can only make a dirty patch at the moment.

José.

[jmllorens]

The pull-request #202 can fix this, although some refinements are welcome.