Hi everyone, I have a stochastic test model (2 stages, 4 scenarios) with some stochastic values on S_COM_PROJ et S_FLO_FUNC. Stochastic, sensivity analysis, SPINES and SPINES(EV) work with only the S_COM_PROJ activated, but neither of the SPINES variant work when I activate the S_FLO_FUNC. I understand from the TIMES documentation that those variants are experimental : is this problem in the "experimental area" or should it work ? Thanks,
To be more accurate : - "do work" = optimal solution found, and results ok. - "do not work" = run aborted due to "***Errors in gams compile***" with error messages in .lst (see attached file)
I just tested it myself, and I can see it working both under the normal stochastic mode and under normal SPINES. But, looking at your listing file, it seems that you must have used the $SET SOLVEDA 1 setting (aka SPINES(EV)?), and by testing that combination, I can see that for now, S_FLO_FUNC does not work with it. This will be fixed in the next version.
But I think you mentioned that you had a problem with it also when running SPINES without $SET SOLVEDA 1? Are you sure? I am not seeing the problem there.
And, most importantly, S_FLO_FUNC should definitely work under the regular stochastic mode, and I am not seeing any problem with it there. Can you confirm that?
Yes the regular stochastic mode work fine with S_FLO_FUNC. Indeed I meant $SET SOLVEDA 1 by SPINES(EV), is this correct ? I I am, I just realized that the case manager always add $SET SOLVEDA 1 in the RUN file even when the simple SPINES variant is selected. Is this the way it should work ? When I make sure that $SET SOLVEDA 1 is not in the RUN file, the simple SPINES variant work fine with S_FLO_FUNC. Using the simple SPINES mode is enough at the moment for me, but I would be very interested in the next version when S_FLO_FUNC will work with SOLVEDA 1.
I'm retaking this topic since VEDA also seems to set automatically the SOLVEDA option for me. In my case, the SOLVEDA option is set to 'YES' in the run file if I set VEDA to deterministic and SPINES-non-EV mode through the control panel. It is set to 1 in SPINES-EV and simple stochastic mode. Thus I have two basic questions: 1°) Is 'YES' the same as 1 (is SOLVEDA only a boolean)? It seemed so to me since the resolution times are very similar for me in the SPINES and SPINES-EV runs. What are the other uses/values of SOLVEDA? It is mentioned in TIMES Doc Part III that it triggers the solution dump for VEDA-FE, what does that mean? 2°) How to override this switch? I tried to write $ SET SOLVEDA 'NO' in the template, but VEDA is writing $ SET SOLVEDA 'YES' some lines after in the Run file after when I hit 'solve' in the case manager. I guess I could try to write it at the bottom of the file, but maybe there is a better way to do it?
1°) No, 'YES' is not the same as 1. For SPINES, $SET SOLVEDA 1 activates the EV mode.
2°) If VEDA is putting
$SET
SOLVEDA 1 into the RUN file, and you don't want that, remove the switch in VEDA which is causing the setting. I guess it is in Tools
→
Options.
Your listing file indicated to me that you were, indeed, using the $SET SOLVEDA 1 setting, when you encountered the error.
Thank you Antti for your answer (although the part about the listing file is more aimed at François I think :-) )
In a general manner and out of curiosity, is there a way to modify VEDA-written switches of the RUN file (like this SOLVEDA) other than through the VEDA control panel?
Ahh... you were overtaking the thread, not simply retaking it!
In general, you can modify the switches manually in the RUN file, or in the RUN file template. However, doing so requires some carefulness with respect to the position of the switch. Some switches need to be positioned already before the call for initsys.mod, while most switches can be placed anywhere before the call for maindrv.mod. Therefore, it is safest to let VEDA change the settings.
And thank you for the answer!