kindred/create
1
1
Fork 0
Code Issues 37 Pull Requests Actions Packages Projects 9 Releases 2 Wiki Activity

Sketch: Solver: Extended Advanced Solver configuration

Browse Source 
========================================================

This is an advanced setting just for allowing increased choices to power users that have problems with a given sketch and want to
test different flavours of DogLeg algorithm.

This commit does not change the default behaviour of FreeCAD. It is only intended to give more options to power users.

The advanced solver configuration is extended to support three different Gauss-newton steps for DogLeg:

FullPivLU => h_gn = Jx.fullPivLu().solve(-fx);
LeastNormFullPivLU => h_gn = Jx.adjoint()*(Jx*Jx.adjoint()).fullPivLu().solve(-fx);
LeastNormLdlt => h_gn = Jx.adjoint()*(Jx*Jx.adjoint()).ldlt().solve(-fx);

This setting is applied only to DogLeg. It is applied to DogLeg as normal or redundant solver, if DogLeg is the selected solver.

Selecting a solver different from DogLeg for both normal and redundant disables the setting.

We have been told:
https://forum.kde.org/viewtopic.php?f=74&t=129439#p346104

that our default Gauss-Newton step in DogLeg may not be adequate in general (we generally deal with underconstraint systems
unless we have a fully constraint sketch, and even then it is many times overconstraint at least for redundant solving).

We have been told that maybe these LeastNorm options are more suitable for us (performance set aside). This enables you as power
user to test if it works fine with FreeCAD.
This commit is contained in:
Abdullah Tahiri
2015-11-26 15:19:03 +01:00
committed by wmayer
parent e0e8edf769
commit e28ca0847a
6 changed files with 100 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
src/Mod/Sketcher/App/planegcs/GCS.cpp
View File

@@ -198,6 +198,7 @@ System::System()
, convergence(1e-10)
, convergenceRedundant(1e-10)
, qrAlgorithm(EigenSparseQR)
, dogLegGaussStep(FullPivLU)
, qrpivotThreshold(1E-13)
, debugMode(Minimal)
, LM_eps(1E-10)
@@ -1453,6 +1454,7 @@ int System::solve_DL(SubSystem* subsys, bool isRedundantsolving)
<< ", tolx: " << tolx
<< ", tolf: " << tolf
<< ", convergence: " << (isRedundantsolving?convergenceRedundant:convergence)
<< ", dogLegGaussStep: " << (dogLegGaussStep==FullPivLU?"FullPivLU":(dogLegGaussStep==LeastNormFullPivLU?"LeastNormFullPivLU":"LeastNormLdlt"))
<< ", xsize: " << xsize
<< ", csize: " << csize
<< ", maxIter: " << maxIterNumber << "\n";
@@ -1505,7 +1507,20 @@ int System::solve_DL(SubSystem* subsys, bool isRedundantsolving)
h_sd = alpha*g;
// get the gauss-newton step
h_gn = Jx.fullPivLu().solve(-fx);
// http://forum.freecadweb.org/viewtopic.php?f=10&t=12769&start=50#p106220
// https://forum.kde.org/viewtopic.php?f=74&t=129439#p346104
switch (dogLegGaussStep){
case FullPivLU:
h_gn = Jx.fullPivLu().solve(-fx);
break;
case LeastNormFullPivLU:
h_gn = Jx.adjoint()*(Jx*Jx.adjoint()).fullPivLu().solve(-fx);
break;
case LeastNormLdlt:
h_gn = Jx.adjoint()*(Jx*Jx.adjoint()).ldlt().solve(-fx);
break;
}
double rel_error = (Jx*h_gn + fx).norm() / fx.norm();
if (rel_error > 1e15)
break;