For Nash: after infeasibilities continue in debug mode before aborting

[dklein-pik]

Purpose of this PR

For Nash only:

So far, REMIND aborted after the maximum number of consecutive infeasibilities/failures (cm_abortOnConsecFail) was reached in at least one region. Now, it switches from parallel to serial (debug) mode after the maximum number was reached and tries to solve (only the infeasible regions, see #1522). If all regions become magically feasible in the debug mode, it continues in parallel mode. If not, it finally aborts. All this is done within GAMS, not in R, to keep the full information from the previous nash and solver iteration. However, this also means that for the duration of the debug mode, which only runs on one core, the other cores are idle.

Type of change

• New feature

Checklist:

• My code follows the coding etiquette
• I performed a self-review of my own code
• I explained my changes within the PR, particularly in hard-to-understand areas
• I checked that the in-code documentation is up-to-date
• I adjusted the reporting in remind2 where it was needed
• I adjusted forbiddenColumnNames in readCheckScenarioConfig.R in case the PR leads to deprecated switches
• All automated model tests pass (FAIL 0 in the output of make test)
• The changelog CHANGELOG.md has been updated correctly

Further information (optional):

• Test runs are here: /p/tmp/dklein/remind-dklein/output/SSP1-PkBudg650_2024-03-27_12.24.58/

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

Nice.

[orichters]

One comment, see below.

main.gms, start.R and start_bundle_coupled.R look good.

By the way: Wasn't there the idea to move the switch definitions to a different file anyway at some point to avoid all this redundant work in the standalone templates that are unusable anyway?

[orichters]

I actually don't get the comment. Setting solprint to on disables it by default?

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

Lazy commenting.

diff --git a/core/loop.gms b/core/loop.gms
index a7758ac..d57b3f7 100644
--- a/core/loop.gms
+++ b/core/loop.gms
@@ -63,7 +63,8 @@ if( (cm_startyear gt 2005),
 ***         SOLVE 
 ***--------------------------------------------------------------------------
 ***this disables solprint in cm_nash_mode=debug case by default. It is switched on in case of infes in nash/solve.gms
-$IF %cm_nash_mode% == "debug" option solprint = off ;
+*RP* for faster debugging, turn solprint immediately on
+$IF %cm_nash_mode% == "debug" option solprint = on ;
 
 o_modelstat = 100;
 loop(sol_itr$(sol_itr.val <= c_solver_try_max),

(You will have to look on the Gitlab repo.)