New WWG System

[SAnsell]

Introduction

The WWG system previously worked [sort of] with a single wwg mesh system. MCNP (with adjustments) suported multiple WWG but annoyingly this was run by running CombLayer multiple times to make multiple WWINP files and then using them in MCNP.
Fluka and PHITS did not have support for WWG.

The latest commit to master has changed this to allow multiple wwg windows for all three codes. This has resulted in a different method of setting up your WWG.

Currently the code for MCNP (default) only supports a SINGLE wwg. window, PHITS only support 6 wwg windows and FLUKA supports none. I have patches for FLUKA/PHITS to solve this problem. I have an old patch for MCNP (6.3 but not later).
Just ask. I am investigating with FLUKA/PHITS if I can publish them fully.

Quick Start Guide

Assuming you just want it to work, these are the main steps done with an example.

~/CombLayerGit/Master/maxiv \
    --FLUKA   \
  --noLengthCheck                                       \
  --defMagnet TDCline                                   \
  --defaultConfig Linac Segment29  \
   --energyCut electron 1.0                  \
   --energyCut photon 1.0                  \
   --energyCut neutron 3000.0                           \
   --sdefType Beam                                      \
   --sdefObj TDC29BellowAA -1             \
   --sdefMod particle electron                          \
   --sdefMod energy 3000.0                              \
   --sdefMod yStep -1.0                                 \
   --sdefMod radius 0.001                               \
   -T mesh electron free ${fullXY}   \
   -T mesh photon free ${fullXY}   \
   -TMod binary mesh*   \
   --wBIAS user all all 3 2.0 \
    -w    \ 
    --wWWG \
   --wwgVTK SS0 testA.vtk  \
   --wwgRPtMesh  TDC29BellowAA 0rigin  \
   --wwgSource S0 TDC29BeamStopA:front  'Vec3D(0,0,-10)' \
   --wwgSource S1 TDC29BeamStopA:back   'Vec3D(0,0,10)' \
   --wwgMesh M1 'Vec3D(-600,9700,-8)' 'Vec3D(-100,10500,8)' 80 50 1
   --comment "mesh particle grid energyCut density r2length r2power"  \
   --wwgCalc SS0 all M1 S0 0.0 0.009 1.0 1.5 \
   --wwgCalc TS1 all M1 S1 0.0 0.009 1.0 1.5 \
   --wwgCADIS SS0ut SS0 TS1 S0 S1 0.009 1.0 1.5 \
   --wwgNorm SS0ut 20 -15 -1 \
   --wwgActive SS0ut Log \
  -n 1e1 \
  -m 1  \
  -s 1839831 \
  AA 
}

The basic steps required are (they can be ordered in anyway in the input):

1. -w :: Start a weight calculation [either cell based or wwg or both]
2. --wWWG :: Initialise a mesh based weight window system calculation
3. --wwgSource :: define a "source point". The parameters are NAME [this can be anything] and then any valid point [in this case it is a named FixedPoint with a vector displacement.
4. --wwgMesh :: Define a 3D grid. This is lower corner and upper corner and the XYZ sizes.
5. --wwgCalc :: Create a weighted mesh bias - SS0 is the name [anything] , the next key is the particles required, M1 is the grid name S0 is the source point, then there are four values: minEnergy to use, density (scalar), r2Length scalar and r2 power scalar.
6. --wwgCADIS [optional] :: Merge two mesh biases into a new group. SS0ut is name of result then the source / adjoint mesh, and then the source / adjoint sources followed by the densityScale, r2scale, r2Power.
7. --wwgNorm [optional] :: The scale is spead/compessed over a range of exp(20) using the result range exp(-15) to exp(-1).
8. --wwgActive :: Write out SSOut for FLUKA [only those selected are written to FLUKA] The option Log implies that results are in log space.

Detailed Description of System

wwgCalc

The principle of this is to calculate the direct beam that can come from a source point into the cells of the grid. This basically does this using an approximate total scattering cross section. Given the approximation [awful] a density scaling and a r2 power scaling are provided.

Obviously that is not likely to be enough for the wwg unless the material is very uniform and light. If a much more detailed build up is needed, then use wwgMarkov to build the value. But often using the CADIS formalism the values obtained from CADIS is very close to that generated by a full calculation, in particular if the region is mono dimensional.

wwgCADIS

This follows the Adjoint method of Wagner. wwgCADIS works by assuming one unit is the source and one is the adjoint (tally) point or plane. The calculation make the result within a single unit. This can be further processed with wwgMarkov and then further CADIS if wanted to accentual the buldup/multscatter.