Ensuring shares sum to 1

This update will make the following changes to how capacity or share additions are made in FTT stand alone, and how regulations are corrected for the stretching effect:

Endogenous shares are calculated for every model following the substitutions made in the shares equation.
From here, endogenous capacity, generation, or useful energy demand is calculated. This difference depends on what the shares are shares of, e.g. in FTT: H market shares are calculated as shares of generation, so corrections should be made in generation space.
Share or capacity additions are corrected to the appropriate units for the given model, and scaled to the number of time steps. For this method, in the first year absolute differences from baseline will very closely match capacity additions (tested for TWSA). After the first year, the additions have stimulated demand and absolute differences will outstrip capacity additions.
Note that the use of MWKA in FTT: P means the treatment is different. Setting MWKA to zero is now immediately effective, however all other values of MWKA may be adjusted due to the stretching effect. Nevertheless, even after setting >80% of capacity using MWKA, the model adjusts very well, and most targets are met after two years. This does not make a large impact of total capacity or total generation, which suggests the implementation works well.
As all corrections are done in capacity (or generation or UED) space, new shares are calculated using the new capacities. This ensures that if zero capacity has been mandated, there will be zero market shares. It ensures shares sum to 1 and prevents negative market shares. Any non-zero capacities will be subjected to the stretching effect, which will happen regardless of methodology.

All files are diffable so please critique the code and perform your own tests to confirm policy decisions are coming through as intended. If share/capacity additions every iteration becomes an issue, Cormac's method of only adding in the last iteration is also successful. Perhaps a switch for this should be a further development?

BONUS COMMIT

I have committed a launch.json file which, if debugging in vscode, will disable numba for you, so you do not need to comment any appearances of njit out if you want to debug the code. When debugging, select the Python: Debug File configuration.

I believe you can continue to commit in your branch, which automatically updates the pull request. May not have time till Friday.

From: Rosie Hayward @.> Sent: 16 August 2023 15:31 To: cpmodel/FTT_StandAlone @.> Cc: Nijsse, Femke @.>; Review requested @.> Subject: Re: [cpmodel/FTT_StandAlone] Ensuring shares sum to 1 (PR #51)

CAUTION: This email originated from outside of the organisation. Do not click links or open attachments unless you recognise the sender and know the content is safe.

@rh-camecon commented on this pull request.

In SourceCode/Industrial_Heat/ftt_mtm_main.pyhttps://github.com/cpmodel/FTT_StandAlone/pull/51#discussion_r1296009058:

     isReg = np.zeros([len(titles['RTI']), len(titles['ITTI'])])
division = np.zeros([len(titles['RTI']), len(titles['ITTI'])])

division = divide((data_dt['IWK3'][:, :, 0] - data['IRG3'][:, :, 0]),

data_dt['IRG3'][:, :, 0])

isReg = 0.5 + 0.5np.tanh(21.25*division)

isReg = np.where(data['IRG3'][:, :, 0] > 0.0,

(np.tanh(1 +

(data_dt['IUD3'][:, :, 0] - data['IRG3'][:, :, 0])

/ data['IRG3'][:, :, 0])),

0.0) isReg[data['IRG3'][:, :, 0] == 0.0] = 1.0

I'm happy to go with that equation if you have tested it, there is nothing wrong with ugly maths if it is ensuring the model works as intended.

I can't change my commit, but it looks like this will work for all the models (just checked it made sense for heat, where HREG is in shares for some reason) so if you want to put it in as a suggested change in all of them, I'm happy to add it to this pull!

— Reply to this email directly, view it on GitHubhttps://github.com/cpmodel/FTT_StandAlone/pull/51#discussion_r1296009058, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AGHDII4WKDY5UTD32PQBOHLXVTKUZANCNFSM6AAAAAA3SNK6X4. You are receiving this because your review was requested.Message ID: @.***>

cpmodel / FTT_StandAlone

Ensuring shares sum to 1 #51