At the moment, the loadSolution method takes a solution from a simulation with modelconceptA to read in as initial conditions for another simulation with modelConceptA.
The method isn't necessarily limited to this. If a modelConceptAB existed, where an additional, or multiple additional primary variables (B) were added on top of the primary variables included in modelconceptA , the loadSolution method could load the solution covering the primary variables in modelConceptA, and the usual initial conditions function could be used for the remaining primary variables from the B scope of the modelConceptAB.
I've added a test that performs an example of this. modelConceptA is a ransmodel with 2 components (rans2C), and the target modelConceptAB is a nonisothermal ransmodel with 2 components (rans2cni). The rans2c solution is read in, and the initial conditions function in problem is used for the nonisothermal model.
This would allow to set up more complicated initial conditions for complicated models with solutions from base model runs.
It could also allow us to reduce testing time, in that initial conditions from a base solution can be used for each test in a specific suite. This would reduce the time required to create physical output.
todos:
-
Make this backwards compatible -
Check for mismatched primary variables. Only load mathing PV names. (This should already exist/work, need to test)