Finished the partial implementation of implicit equations from a few years ago.
Right now, we store the old solution value for all implicit equation type fields with a boost::unordered_map. This should allow for straightforward extension to higher order time schemes (by indexing previous time states).
The performance is about the same, if not slightly worse, since we have to check whether we have a map entry that needs to be copied for each field index in solveIncrement.cc. This can likely be circumvented by another bool entry to the variable_info struct.
Finished the partial implementation of implicit equations from a few years ago.
Right now, we store the old solution value for all implicit equation type fields with a
boost::unordered_map
. This should allow for straightforward extension to higher order time schemes (by indexing previous time states).The performance is about the same, if not slightly worse, since we have to check whether we have a map entry that needs to be copied for each field index in
solveIncrement.cc
. This can likely be circumvented by another bool entry to thevariable_info
struct.Closes #228