#3264 closed defect (fixed)
Solver for algebraic loops
| Reported by: | Rüdiger Franke | Owned by: | Marcus Walther |
|---|---|---|---|
| Priority: | high | Milestone: | 1.9.4 |
| Component: | Cpp Run-time | Version: | trunk |
| Keywords: | Cc: |
Description (last modified by )
Currently an FMU with algebraic loop fails because the solver "Newton" is not known. The first hurdle can be taken by fixing the capatalization in FMU(2)GlobalSettings.h from "Newton" to "newton" -- or better configure "kinsol" as in OMCFactory::ReadSimulationParameter (r25379, r25381)?
Btw. LinSolverOMCFactory only knows "umfpack" -- is the default "kinsol" in OMCFactory::ReadSimulationParameter correct?
Anyway, the FMU now wants to load libOMCppKinsol.so dynamically at runtime. It appears that the setting in SimulationRuntime/Core/System/CMakeLists.txt, Line 16 is wrong:
set_target_properties(${SystemName}_FMU PROPERTIES COMPILE_DEFINITIONS "DRUNTIME_STATIC_LINKING")
Changing to COMPILE_DEFINITIONS "RUNTIME_STATIC_LINKING", without D, the resulting libOMCppSystem_FMU contains an AlgLoopSolverFactory with the intended constructer. But that one crashes.
r25380 adds the rpath for now, so that FMUs with algebraic loops work. This is considered a temporary fix though.
How can one link a reasonable solver for algebraic loops into the FMU?
Change History (6)
comment:1 by , 10 years ago
| Description: | modified (diff) |
|---|
comment:2 by , 9 years ago
| Milestone: | 1.9.3 → 1.9.4 |
|---|
comment:3 by , 9 years ago
@rfranke: I think this ticket is already fixed, what is your opinion? The default solver for FMU(2) is Newton. If Kinsol should be used as solver, we can pass "-DFMU_SUNDIALS=ON" to the cmake-script of the c++ runtime. This will put the kinsol library and a licence text to the generated FMU(2).
comment:4 by , 9 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Moved to new milestone 1.9.4