🎨 Constraint function plots in `plot_proc` #4049

chris-ashe · 2026-01-19T10:35:49Z

This pull request enhances the post-optimization reporting for constraint handling in process/scan.py. The main improvements involve adding more detailed output for both equality and inequality constraints, including physical values, units, symbols, and bounds, which should make debugging and analysis easier.

Enhanced constraint reporting:

Added output of residuals, constraint values, and units for equality constraints using process_output.ovarre, improving the detail of post-optimization logs.
For inequality constraints, now calculate and output the physical value, symbol, units, and physical bound for each constraint, and include these in the output file for better traceability.

Output formatting improvements:

Updated the table headers for inequality constraint reporting to include "Physical constraint bound" for clearer presentation of results.

Checklist

I confirm that I have completed the following checks:

My changes follow the PROCESS style guide
I have justified any large differences in the regression tests caused by this pull request in the comments.
I have added new tests where appropriate for the changes I have made.
If I have had to change any existing unit or integration tests, I have justified this change in the pull request comments.
If I have made documentation changes, I have checked they render correctly.
I have added documentation for my change, if appropriate.

…djust font sizes and add new plots in main function

…d normalization and improve axis annotations

codecov-commenter · 2026-01-19T10:45:14Z

Codecov Report

❌ Patch coverage is 1.82927% with 161 lines in your changes missing coverage. Please review.
✅ Project coverage is 46.62%. Comparing base (5ea7543) to head (41f0c58).
⚠️ Report is 7 commits behind head on main.

Files with missing lines	Patch %	Lines
process/io/plot_proc.py	1.97%	149 Missing ⚠️
process/scan.py	0.00%	12 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4049      +/-   ##
==========================================
+ Coverage   46.49%   46.62%   +0.12%     
==========================================
  Files         123      123              
  Lines       28777    29530     +753     
==========================================
+ Hits        13381    13769     +388     
- Misses      15396    15761     +365

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…lace dashed line with arrow annotation, and remove relative change text display

…on to ensure proper string formatting

…function for improved clarity

…ght, add z-order for layering, and improve text annotation with dynamic color and background

jonmaddock · 2026-01-21T14:07:21Z

@chris-ashe could you add an image of the new output here?

chris-ashe · 2026-01-21T14:41:03Z

@chris-ashe could you add an image of the new output here?

Have added to main PR text now

jonmaddock

Great to have this working, such an improvement to our visualisation! Main points are about calculating stuff that has either been calculated already (I think) or should have been, e.g. normalisation of constraint/bound values.

process/scan.py

jonmaddock · 2026-01-21T14:40:24Z

process/scan.py

+                constraint_bound = con2[i]
+
+                # assumes f-value is 1
+                # -cc because sign is reversed in constraint_eqns


Can you be totally clear here: after the sign change here, a constraint > 0 is violated, right? i.e. the convention.

From my understanding and by the data the plots make it seems to be the correct way. @timothy-nunn what do you think?

In PROCESS, a negative cc indicates that a constraint is violated

jonmaddock · 2026-01-21T14:41:58Z

process/scan.py

+                    constraint_value = constraint_bound * (-numerics.rcm[i] + 1)
+                else:
+                    raise ProcessValueError(
+                        f"Unknown constraint direction '{sym[i]}' for inequality constraint"


Not sure this could ever happen: this is already tested in constraints, and doesn't need to be re-done here I believe.

jonmaddock · 2026-01-21T14:45:13Z

process/scan.py


            for i in range(numerics.neqns, numerics.neqns + numerics.nineqns):
                name = numerics.lablcc[numerics.icc[i] - 1]
+                constraint_bound = con2[i]


What's con2?

Its a horrible name but it is the value of the equality constraint. So for a machine of major radius 8m, the radial build consistency constraint value is 8m. Have added a comment

jonmaddock · 2026-01-21T14:49:25Z

process/scan.py

+                if sym[i] == ">=":
+                    constraint_value = constraint_bound * (1 - -numerics.rcm[i])
+                elif sym[i] == "<=":
+                    constraint_value = constraint_bound * (-numerics.rcm[i] + 1)


rcm is the normalised constraint residual, right? This seems rather confusing to verify without a comment. Aren't the absolute constraint values already stored on the constraint object? If not, they should be, rather than being re-calculated here.

Before when we had f values we couldn't have the absolute value of the constraint variable as the f values were not tied to it. @timothy-nunn has a better explanation for it as we tried to make this plot with f values and found we couldnt

This always confuses me as when you look into the code people often misuse this parameter and put in whatever they feel like. This is why I am proposing to standardise and abstract our constraints as much as possible (#3937).

jonmaddock · 2026-01-21T14:52:46Z

process/io/plot_proc.py

+            / (itvar_upper - itvar_lower)
+            if itvar_final != itvar_lower
+            else 0
+        )


I don't think we should be calculating this again here: I believe this is already calculated, and we should use that value.

I don't think this is in the output. We have the final / initial value given by xcm. This is then used to find the initial value and where that sits within the normalised bounds

process/io/plot_proc.py

jonmaddock · 2026-01-21T14:58:33Z

process/io/plot_proc.py

    # Plot main plasma information
    plot_main_plasma_information(
-        figs[3].add_subplot(111, aspect="equal"),
+        figs[4].add_subplot(111, aspect="equal"),


This churn is a bit of a pain. Could figs just be appended to?

jonmaddock · 2026-01-21T15:00:55Z

This looks fantastic now!

timothy-nunn

I'm not sure the equality constraint plot is doing anything for me. Its quite hard to meaningfully plot the residuals of these constraints especially given the different orders of magnitude. Would this be better as just a table?

The inequality bar chart should show margin so the bars should come from the right for $\leq$ constraints.

…pper bound

chris-ashe added 3 commits January 18, 2026 12:37

Enhance plotting functions for equality and inequality constraints; a…

7d35f9d

…djust font sizes and add new plots in main function

Add normalized relative change plotting and enhance axis formatting

e5216cd

Enhance plotting functions for iteration and constraint variables; ad…

ee77603

…d normalization and improve axis annotations

chris-ashe requested review from jonmaddock and timothy-nunn January 19, 2026 10:35

chris-ashe self-assigned this Jan 19, 2026

chris-ashe added the Input/Output Files Issues related to the input and output data files label Jan 19, 2026

chris-ashe added 4 commits January 21, 2026 10:25

Refactor plot_iteration_variables function: change color to blue, rep…

11a207e

…lace dashed line with arrow annotation, and remove relative change text display

Fix unit type extraction in plot_equality_constraint_equations functi…

eea102e

…on to ensure proper string formatting

Adjust tolerance levels and color coding in plot_iteration_variables …

b3850b0

…function for improved clarity

Enhance plot_inequality_constraint_equations function: adjust bar hei…

235a486

…ght, add z-order for layering, and improve text annotation with dynamic color and background

jonmaddock requested changes Jan 21, 2026

View reviewed changes

timothy-nunn assigned timothy-nunn and unassigned chris-ashe Jan 21, 2026

Requested changes

3f49ae3

chris-ashe requested a review from a team as a code owner January 22, 2026 09:54

timothy-nunn reviewed Jan 22, 2026

View reviewed changes

Make the bar plots originate from the right if the constraint is an u…

41f0c58

…pper bound

jonmaddock mentioned this pull request Jan 22, 2026

WIP: Resolve "Add solution comparison tool" - [opened] #2770

Closed

🎨 Constraint function plots in plot_proc #4049

Are you sure you want to change the base?

🎨 Constraint function plots in plot_proc #4049

Conversation

chris-ashe commented Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Checklist

Uh oh!

codecov-commenter commented Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

jonmaddock commented Jan 21, 2026

Uh oh!

chris-ashe commented Jan 21, 2026

Uh oh!

jonmaddock left a comment

Choose a reason for hiding this comment

Uh oh!

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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jonmaddock commented Jan 21, 2026

Uh oh!

timothy-nunn left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

🎨 Constraint function plots in `plot_proc` #4049

🎨 Constraint function plots in `plot_proc` #4049

chris-ashe commented Jan 19, 2026 •

edited

Loading

codecov-commenter commented Jan 19, 2026 •

edited

Loading

timothy-nunn left a comment •

edited

Loading