Added more agressive caching, regained about another 50% performance gain. Solar model over 10Gyr now takes roughly 400ms (as opposed to 1 second before)
GridFire failed to compile on gcc and gnu stdlibc++ this has been
resolved. Further, the boost dependency has been removed since we no
longer use boost at all. This should dramatically simplify installation.
Finally we have added some build system checks to ensure that the
correct version of a C++ and fortran compiler are present on the system
Neutrino loss is essential for neutrino cooling. Started adding
framework to track this. Reaclib reactions use a simple heuristic where
electron capture reactions loss 100% of their energy to neutrinos
whereas beta decay reactions loose 50% of their energy to neutrinos
Previously Jacobians were stored by engines and accessed through engine
accessors (e.g getJacobianMatrixEntry); however, this resulted in
desynced jacobian states. We have changed to a pattern of Engine creates
a jacobian and returns it to the caller. The caller can then do what
they will with it. Because of this the getJacobianMatrixEntry method has
been removed.
BREAKING CHANGE:
- There is no longer any getJacobianMatrixEntry method on
DynamicEngine classes
- the generateJacobian method signature has changed to return a
NetworkJacobian object. Internally this uses an Eigen Sparse Matrix to
store its data.
one major issue was that QSE solving was only running at each partition. This was creating effectivley infinite sources of partitioned species. Now we partition when engine updates are triggered, however, solveQSEAbundance is called every timestep. This has major performance implications and so has required a lot of optimization to make it even somewhat viable. For now construction is much slower. Time per iteration is still slower than it was before; however, it is tractable. There is also currently too much stiffness in the network. This is likeley a bug that was introduced in this refactoring which will be addressed soon.
This entailed a major rewrite of the composition handling from each engine and engine view along with the solver and primer. The intent here is to let Compositions be constructed from the same extensive property which the solver tracks internally. This addressed C0 discontinuity issues in the tracked molar abundances of species which were introduced by repeadidly swaping from molar abundance space to mass fraction space and back. This also allowed for a simplification of the primeNetwork method. Specifically the mass borrowing system was dramatically simplified as molar abundances are extensive.
Added new ways to select exactly what types of reactions (strong, beta+, beta-, electron capture, positron capture, or any combination thereof) can be turned on at network construction time. There are a few quality of life masks added as well such as weak which addes all weak type reactions, and all which adds weak + strong reactions. The default is to just add strong reactions for now.
previousl the sparsity calculations for the jacobian matrix were completly broken. The method subgraph_sparsity was returning that all derivities were only depenednt on temperature and density. It should have been reporting that they also depended on some of the abundances. This was resolved by switching to a different structural sparsity engine (for_jac_sparsity). This bug had turned the solver into a fixed point iteration solver which failed for the stiff system we have. Now that it is resolved the solver can once again evolved over Gyr timescales.
previously recordADTape was duplicating a lot of work which recordEpsTape also needed to do. Now all derivs are being recorded into m_rhsADFun so that only one tape recording phase is needed per network build stage.
essentially all callers can now inform the graph engine about which species they hold active and graph engine then uses those to define a sparsity pattern and only calculate the jacobian along that sparsity pattern
All jacobian calculations were broken because the indexing used to record the AD tape was broken (see not parallel to) the indexing used by the composition object. A fix for this was to sort the network species by mass. However, more generally we should introduce a mechanism to ensure these two indexed sets always remain parallel
bigs were introduced by the interface change from accepting raw molar abundance vectors to using the composition vector. This commit resolves many of these, including preformant ways to report that a species is not present in the composition and unified index lookups using composition object tooling.
BREAKING CHANGE:
Major weak rate progress which includes: A refactor of many of the public interfaces for GridFire Engines to use composition objects as opposed to raw abundance vectors. This helps prevent index mismatch errors. Further, the weak reaction class has been expanded with the majority of an implimentation, including an atomic_base derived class to allow for proper auto diff tracking of the interpolated table results. Some additional changes are that the version of fourdst and libcomposition have been bumped to versions with smarter caching of intermediate vectors and a few bug fixes.
