Closed jonmaddock closed 5 months ago
As the optimisation parameters are now sorted immediately after reading in from the input file, this causes the solution vector in the mfile to be re-ordered. The (re-ordered) solution vector is very similar, but not the same; this is likely to be to do with the non-idempotence effect previously observed when running the models multiple times or re-ordering the optimisation parameters. This can be reduced or eliminated by re-running the models until the outputs converge (https://github.com/ukaea/PROCESS-usage/issues/12). I think this effect is separate to this issue, and therefore although the resulting MFILE is (slightly) different, it is no less valid than the current regression reference.
Note that comparing itvar*
and xcm*
values in the test do not make sense due to the re-ordering of the inputs.
@jonmaddock should we try and do for the inputs what we did for the constraints (#3038)
Ensures consistent order and makes output comparison easier.