Add random vector sampling method to MGA by bsergi · Pull Request #162 · ReEDS-Model/ReEDS

bsergi · 2026-07-27T20:36:27Z

Summary

This PR adds a new capability to the existing MGA method to support random vector sampling across subcategories of the variables being explored.

Technical details

The current MGA method applies the minimization (or maximization) to the sum of a variable across all modeled regions (e.g., total U.S. solar capacity or fossil generation). The random vector method assigns randomly-sampled weights to the regional quantities applied before the summation. This technique allows for exploring the regional distribution of capacity and generation results in the near-optimal space.

As an illustration, the plot below shows the range of natural gas capacity by region (using the 54-zones) from 100 runs with the MGA random vector method with a 1% cost threshold. The MGA max/min squares indicate the results from the previous, nationwide-method; the original min/max approach is better suited for getting a wider range of total capacity, whereas the new method is better geared toward understanding uncertainty at a regional level.

Implementation notes

The weights are generated by themain_mga_rv function in mcs_sampler.py, with the number of samples (runs) controlled by GSw_MGA_RV_runs.
Regional weights are drawn from a support of {-1,1} with equal probability; including both negative and positive values allows for exploring minimization in one region and maximization in another.
- The method in mcs_sampler.py can support linear sampling from [-1,1] using a uniform distribution. Initial tests suggested this didn't yield as much of a signal so we haven't implemented a switch to control this.
- The regionality at which unique weights are sampled is controlled by GSw_MGA_RV_region and can be any level in the hierarchy file.
Currently this method is only supported for GSw_MGA_Objective = (capacity or generation), with weights generated by region and technology group (i_subtech).
- If GSw_MGA_SubObjective is an aggregated tech category (e.g., gentech), unique weights will be sampled for the different i_subtech groups within the larger category.
Weights are read in by b_inputs.gms and applied to the MGA objective function in d_mga.gms.

Additional changes

Added a run using the random vector method to cases_test.csv.

Switches added/removed/changed

GSw_MGA_RV_runs: number of random weight vectors to draw; corresponds to the number of ReEDS runs.
GSw_MGA_RV_region`: level of regionality at which to sample the random weights.

Relevant sources or documentation

The random vector method has been used in previous MGA studies (examples here and here

Validation, testing, and comparison report(s)

Will run reference case tests once the HPC is back in service. Will also link a slide deck with some results using this method.

Checklist for author

Details to double-check

General information to guide review

Zero impact on results of default case
No large data file(s) added/modified
No substantive impact on runtime for full-US reference case
No substantive impact on folder size for full-US reference case
No change to process flow (runreeds.py, reeds/core/solve/solve.py)
No change to code organization
No change to package requirements (environment.yml or Project.toml)

Did you use LLM tools (chatbot or copilot) in the preparation of this PR? If so, describe how

Tag points of contact here if you would like additional review of the relevant parts of the model

bsergi · 2026-07-27T20:41:19Z

                    print(f"\n[Warning] Case Column: '{c}'")
                    print(
-                        "You are attempting to run a Monte Carlo simulation with "
+                        "You are attempting to run a Monte Carlo or MGA Random Vector simulation with "


This probably doesn't matter since the MGA runs should all have the same inputs, but I left it in for now in case there's a non-deterministic component to the representative day sampling.

If the inputs are unchanged then the rep days will be unchanged, so I'm pretty sure we can remove the check for RV and only keep it for MCS, which makes it a little easier to use.

patrickbrown4

Looks great! The only critical comment is in d_mga.gms; all the rest are minor cleanups. Approving now but should still confirm zero change before merging.

patrickbrown4 · 2026-07-29T16:57:54Z

 inflation.csv,inputs/financials/inflation_{inflation_suffix}.csv,1,ignore,ignore,,,,0,0,,,,,,
 interconnection_queues.csv,inputs/capacity_exogenous/interconnection_queues.csv,1,ignore,ignore,r,"tg,r",,1,0,,,,,,
-jtype.csv,inputs/sets/jtype.csv,1,ignore,ignore,,,,,,,,set,jtype,job types used in model (construction and om),
+jtype.csv,inputs/sets/jtype.csv,1,ignore,ignore,,,,,0,,,set,jtype,job types used in model (construction and om),


jtype.csv doesn't have a header so I think the changes to this file should be reverted

patrickbrown4 · 2026-07-29T17:03:28Z

                    print(f"\n[Warning] Case Column: '{c}'")
                    print(
-                        "You are attempting to run a Monte Carlo simulation with "
+                        "You are attempting to run a Monte Carlo or MGA Random Vector simulation with "


If the inputs are unchanged then the rep days will be unchanged, so I'm pretty sure we can remove the check for RV and only keep it for MCS, which makes it a little easier to use.

patrickbrown4 · 2026-07-29T17:06:01Z

+                if int(dfcases[c].MCS_runs) > 0:
+                    mcs_sampler.general_mcs_dist_validation(reeds.io.reeds_path, mcs_dist_path, sw)
+                    numruns = int(dfcases[c].MCS_runs)
+                    run_type = 'MC'
+                else:                    
+                    numruns = int(dfcases[c].GSw_MGA_RV_runs)
+                    run_type = 'R'


MCS and RV are mutually incompatible, right (at least in the current implementation, which is totally ok)? If so, should probably add a check to runreeds.check_compatibility() that raises an exception if MCS_runs and GSw_MGA_RV_runs are both nonzero.

patrickbrown4 · 2026-07-29T17:08:04Z

 import pandas as pd
 import scipy.stats
 import sys
+import re


(minor cleanup) can remove since unused

Suggested change

import re

patrickbrown4 · 2026-07-29T17:08:54Z

+
+# %%


Suggested change

# %%

patrickbrown4 · 2026-07-29T17:46:07Z


+By default the MGA min/max is applied to the sum of the variable across all regions being modeled.
+The MGA approach also supports an option to randomly sample of a vector of weights to apply to the regional values of the variable being optimized.
+This method can be useful to characterizing the uncertainty in the regional distribution of the results.


Suggested change

This method can be useful to characterizing the uncertainty in the regional distribution of the results.

This method can be used to characterize the uncertainty in the regional distribution of the results.

patrickbrown4 · 2026-07-29T17:48:15Z

+By default the MGA min/max is applied to the sum of the variable across all regions being modeled.
+The MGA approach also supports an option to randomly sample of a vector of weights to apply to the regional values of the variable being optimized.
+This method can be useful to characterizing the uncertainty in the regional distribution of the results.
+Weights are sampled as discrete values from a support of {-1,1} to allow for simultaneous minimization and maximization.


Just to check, is "support" the right word here? (I'm unfamiliar with this usage but also don't know much stats)

patrickbrown4 · 2026-07-29T17:48:47Z


+The MGA random vector capability is controlled by the following switches:
+- `GSw_MGA_RV_runs` (default `0`): Number of random samples of weight vectors to draw; corresponds to the number of runs.
+- `GSw_MGA_RV_region` (default `r`): Regionality level (specified by hierarhcy file) over which to sample the random weights. 


Suggested change

- `GSw_MGA_RV_region` (default `r`): Regionality level (specified by hierarhcy file) over which to sample the random weights.

- `GSw_MGA_RV_region` (default `r`): Regionality level (specified by hierarchy file) over which to sample the random weights.

patrickbrown4 · 2026-07-29T17:51:16Z

+*TODO: add different handling for other subobjectives with different dimensions
+* should we also include error checking?
+
+


Best to move to issue if you still want to address it (or just remove if it's old)

Suggested change

*TODO: add different handling for other subobjectives with different dimensions

* should we also include error checking?

patrickbrown4 · 2026-07-29T17:53:47Z

+* ---------------------------------------------------------------------------
+
+$elseif.mgaobj %GSw_MGA_Objective% == 'generation'
+Equation eq_MGA_Objective "--MW-- Defines generation for MGA" ;
+Variable MGA_OBJ "--MWh-- Generation of technology to be minimized/maximied" ;
+eq_MGA_Objective$Sw_MGA..
+    MGA_OBJ
+    =e=
+    sum{(i,v,r,h,t)
+        $[tmodel(t)
+        $valgen(i,v,r,t)
+        $%GSw_MGA_SubObjective%(i)],
+        GEN(i,v,r,h,t) * hours(h) 
+        * sum{i_subtech$i_subsets(i,i_subtech), mga_weights(r,i_subtech)}


Duplicates the next block below; can instead just add * sum{i_subtech$i_subsets(i,i_subtech), mga_weights(r,i_subtech)} to the existing $elseif.mgaobj %GSw_MGA_Objective% == 'generation' block

atpham88 and others added 30 commits April 16, 2026 12:10

move files to github.com

b91d8e0

use gentech as MGA subobjective instead of all techs

261d0f4

resolve conflicts

0149846

Update user_guide.md

c980617

resolve conflicts

54fac41

Merge branch 'main' into ap_MGA

864ec0e

clean up

4edc116

remove all cap MGA max/min option

145d3c2

add gentech to i_subtech set

6f8216b

resolve to pass github tests

ffd3c6f

Update futurefiles.csv

255c39f

Merge branch 'main' into ap_MGA

76a989d

reduce number of employment output files

ab732e8

consolidate employment output files

ea35ef8

clean up

a010fcc

clean up

de30597

fix construction job-years calculation

98f2e43

no longer need encoding when writing log file

13224b8

Merge remote-tracking branch 'origin/ap_MGA' into bs/mga_rv

2c88390

Merge branch 'main' into ap_MGA

111be85

revert changes

a827879

revert changes

61e2c5d

Update log.py

3e52966

Merge remote-tracking branch 'origin/ap_MGA' into bs/mga_rv

bb14377

Merge remote-tracking branch 'origin/main' into bs/mga_rv

45be271

fix handling of jtype set for employment factors

11ff01c

setup of multiple runs for MGA RV and creation of weights

111fabf

handling of MGA weights in GAMS

303544f

add MGA random vector test case, group switches

7aeda82

remove electrolyzer/dac/smr from gentech

eab1b63

bsergi added 13 commits June 8, 2026 10:20

update switch name for MGA RV approach

81a033a

Merge remote-tracking branch 'origin/main' into bs/mga_rv

3a71b8b

fix bug with monte carlo run naming

e5c225f

remove unneeded files from runfiles.csv

ebc1bb2

reformat and add descriptive names to LHS MGA RV file

5bda41d

draw weights from uniform(-1,1)

eeeaad6

Merge remote-tracking branch 'origin/main' into bs/mga_rv

e05039d

add option for discrete sampling with MGA random vector

1305692

Merge branch 'main' into bs/mga_rv

69c647b

Add option to sample weights at any level in the hierarchy file

af1b962

Move MGA random vector compatibility check up

7085ff9

Add details on MGA random vector to user guide

611d2bf

remove "download everything" requirement for MGA random vector

71e5ab1

bsergi commented Jul 27, 2026

View reviewed changes

bsergi added 3 commits July 27, 2026 16:47

revert unrelated change to transmission employment

ead44e0

remove upper case files

64986af

revert line ending changes

b6aae52

bsergi marked this pull request as ready for review July 27, 2026 20:54

bsergi requested a review from patrickbrown4 July 27, 2026 20:54

github-actions Bot added data_changes model_changes docs switches labels Jul 27, 2026

bsergi requested a review from atpham88 July 27, 2026 20:54

patrickbrown4 approved these changes Jul 29, 2026

View reviewed changes

	This method can be useful to characterizing the uncertainty in the regional distribution of the results.
	This method can be used to characterize the uncertainty in the regional distribution of the results.

	- `GSw_MGA_RV_region` (default `r`): Regionality level (specified by hierarhcy file) over which to sample the random weights.
	- `GSw_MGA_RV_region` (default `r`): Regionality level (specified by hierarchy file) over which to sample the random weights.

		*TODO: add different handling for other subobjectives with different dimensions
		* should we also include error checking?

Uh oh!

Conversation

bsergi commented Jul 27, 2026

Summary

Technical details

Implementation notes

Additional changes

Switches added/removed/changed

Relevant sources or documentation

Validation, testing, and comparison report(s)

Checklist for author

Details to double-check

General information to guide review

Did you use LLM tools (chatbot or copilot) in the preparation of this PR? If so, describe how

Tag points of contact here if you would like additional review of the relevant parts of the model

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

patrickbrown4 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants