Remove f values #3947

timothy-nunn · 2025-10-22T09:09:06Z

Removes F-values from PROCESS

Add a test for the constraints that runs and checks no attribute/type errors occur (fixed a few bugs that arose before the removal)
Removed f-values from constraints, iteration variables, inputs, scan variables, and their mention in the docs. The follow f-values have been kept but have been removed from their ability to iterate:
- fl_h_threshold
- fiooic
- fjohc
- fjohc0
- fdene
- fradpwr
Removed f-values from all regression tests
- st_regression.IN.DAT used some weird f-value bounds so I have updated the constraints (see comment below).
Added an additional constraint (22) to enforce L-mode (constraint 15 remains to enforce H-mode).

codecov-commenter · 2025-11-05T14:45:13Z

Codecov Report

❌ Patch coverage is 80.85106% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 46.22%. Comparing base (f59059d) to head (a400212).
⚠️ Report is 12 commits behind head on main.

Files with missing lines	Patch %	Lines
process/pfcoil.py	0.00%	4 Missing ⚠️
process/constraints.py	80.00%	3 Missing ⚠️
process/superconducting_tf_coil.py	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3947      +/-   ##
==========================================
+ Coverage   46.03%   46.22%   +0.19%     
==========================================
  Files         123      123              
  Lines       28970    28774     -196     
==========================================
- Hits        13335    13301      -34     
+ Misses      15635    15473     -162

☔ 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.

timothy-nunn · 2025-11-13T16:06:43Z

tests/regression/input_files/st_regression.IN.DAT

-ixc = 46
-fp_hcd_injected_max = 1.0
-boundl(46) = 0.6
-boundu(46) = 1.5
-* DESCRIPTION:   f-value for injected power variation range
-* JUSTIFICATION: Setup to allow the injected power to vary
-
 p_hcd_injected_max = 150.0 


When fp_hcd_injected_max = 1.5: $\frac{225.0}{150.0}-1.5 = 0$ therefore the new bound should be $225.0$

timothy-nunn · 2025-11-13T16:33:18Z

tests/regression/input_files/st_regression.IN.DAT

 *--------------------------Numerics Variables---------------------------*
 *‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾*

-epsvmc = 1.0E-11


This was much lower than other regression tests

timothy-nunn · 2025-11-13T17:02:47Z

How to update an input file to not use f-values

Ensure all equality constraint definitions (icc = ...) occur before any inequality constraints.
Set the variable neqns equal to the number of equality constraints.
Remove obsolete f-values from the input file.
Remove iteration variables that corresponded to f-values.

In 99% of cases, this will allow your file to run without f-values and without treating inequality constraints as equality. However, if the input file specified weird bounds on the f-value iteration variables, you will need to modify the constraint bound (max/min) to account for this change.

mkovari

See comments

examples/data/large_tokamak_IN.DAT

mkovari · 2025-11-14T13:54:13Z

documentation/io/input-guide.md

-

 !!! Info "Constraints"  
    A full list of constraints is given on the variable description page in the row labelled 


We need to update this in due course.

mkovari · 2025-11-14T13:54:50Z

process/io/obsolete_vars.py

+OBS_VARS_HELP.update(
+    dict.fromkeys(
+        fvalues_list,
+        "F-values are no longer supported, please use inequality constraints",


When the inequality constraints are documented we could add a link to the docs here.

mkovari · 2025-11-14T14:02:11Z

process/pfcoil.py

-                        numerics.boundu[37]
-                        * pfcoil_variables.j_cs_critical_flat_top_end
+                        # numerics.boundu[fjohc] * (which should have been 1)
+                        pfcoil_variables.j_cs_critical_flat_top_end


I don't think this can ever be false:
abs(pfcoil_variables.j_cs_flat_top_end) > 0.99e0 * abs(pfcoil_variables.j_cs_critical_flat_top_end)

Keep this warning (cslimit).
Keep fjohc and fjohc0 as variables (not iteration) and include them in the constraint.
Use them instead of the bounds in this line of code.

mkovari · 2025-11-14T14:03:35Z

process/pfcoil.py

                        "CS not using max current density: further optimisation may be possible"
                    )

-                # Check whether CS coil currents are feasible from engineering POV


Ideally these warnings should be recreated but it's no big deal.

mkovari · 2025-11-14T14:13:41Z

process/stellarator.py


        # The operation current density weighted with the global iop/icrit fraction
-        lhs[:] = constraint_variables.fiooic * jcrit_vector
+        lhs[:] = jcrit_vector


This isn't really correct anymore, I suspect, and the same may be true for the code below.

What is the recourse here then? Should fiooic be added back? Or is there a way to calculate it from other variables in PROCESS (maybe the mentioned iop/icrit)?

Sorry, I really don't know anything about the stellarator code. I don't know what lhs and rhs are.

… and scan etc

ajpearcey · 2025-11-18T13:16:14Z

documentation/physics-models/plasma_h_mode.md


 This constraint can be activated by stating `icc = 15` in the input file.

-The scaling value `fl_h_threshold (ixc=103)` can be varied to set the required margin around the threshold.


fl_h_threshold should be kept and renamed. This new way of using it should be documented here.

I have updated the description and docs to document how I think fl_h_threshold is working now. We should probably go through this at some point (with some example input files) to ensure its working as before/expected.

In particular, we should carefully check how it was used to enforce L-mode/H-mode before.

tests/regression/input_files/large_tokamak_eval.IN.DAT

documentation/physics-models/plasma_density.md

examples/data/large_tokamak_IN.DAT

examples/data/large_tokamak_eval_IN.DAT

tests/integration/data/large_tokamak_eval.IN.DAT

tests/regression/input_files/large_tokamak_nof.IN.DAT

tests/unit/data/large_tokamak_IN.DAT

process/input.py

process/constraints.py

timothy-nunn · 2025-11-25T14:29:21Z

Hi @ajpearcey @mkovari

Thank you both for the reviews. I have resolved everything that I have actioned. The following comments are still outstanding:

fiooic and another f-values use in warnings/outputs
A condition that can (now?) never be false: Remove f values #3947 (comment)
fl_h_threshold

timothy-nunn · 2025-11-25T16:01:01Z

@chris-ashe has pointed me to f_c_tf_turn_operating_critical which is possibly the same as fiooic.

Looking at an MFile, they have very similar values:

fiooic___________________________________________________________________ (itvar024)_____________________ 7.28529047768881166e-01

Actual_current_/_critical_current________________________________________ (f_c_tf_turn_operating_critical)_ 7.28529093232043401e-01 OP

So I can replace the warning and its use in stellarator with f_c_tf_turn_operating_critical.

This will not work for the Stellarator (#3947 (comment)) because f_c_tf_turn_operating_critical is not set, nor are a lot of variables related to the TF critical current (@ym1906 any ideas since you're probably most familiar with the Stellarator models? I could add an input in that is specifically a margin for the Stellarator model that would not iterate and would default to say 0.7).

mkovari · 2025-11-26T10:18:43Z

There are two separate constraint equations for enforcing the L-H threshold: icc = 15 and 73.
In both cases it would be tidier to have separate constraints for L and H mode.
In any case there needs to be a margin defined by a parameter (not an iteration variable) to replace fl_h_threshold.

timothy-nunn · 2025-11-26T10:28:25Z

There are two separate constraint equations for enforcing the L-H threshold: icc = 15 and 73. In both cases it would be tidier to have separate constraints for L and H mode. In any case there needs to be a margin defined by a parameter (not an iteration variable) to replace fl_h_threshold.

@ajpearcey @chris-ashe @jonmaddock do we have any thoughts on this? I'm not sure we can use fl_h_threshold as set out here.

I think this requires the margin (f-value) to be iterating.

mkovari · 2025-11-25T14:03:54Z

process/stellarator.py


        # The operation current density weighted with the global iop/icrit fraction
-        lhs[:] = constraint_variables.fiooic * jcrit_vector
+        lhs[:] = jcrit_vector


Sorry, I really don't know anything about the stellarator code. I don't know what lhs and rhs are.

mkovari · 2025-11-26T10:02:55Z

process/pfcoil.py

-                        numerics.boundu[37]
-                        * pfcoil_variables.j_cs_critical_flat_top_end
+                        # numerics.boundu[fjohc] * (which should have been 1)
+                        pfcoil_variables.j_cs_critical_flat_top_end


Keep this warning (cslimit).
Keep fjohc and fjohc0 as variables (not iteration) and include them in the constraint.
Use them instead of the bounds in this line of code.

mkovari · 2025-11-26T10:05:08Z

process/pfcoil.py

+                        "j_cs_pulse_start / j_cs_critical_pulse_start shouldn't be above 0.7 "
+                        "for engineering reliability"
+                    )
+


These warnings are fine. Remember the constraints on current density may not be applied (if T margin is used instead).

mkovari · 2025-11-26T10:07:36Z

process/superconducting_tf_coil.py

+                "f_c_tf_turn_operating_critical shouldn't be above 0.7 for engineering reliability"
+            )
+
        po.ovarre(


This warning is fine.

mkovari · 2025-11-26T10:10:23Z

process/constraints.py

    args : output structure : residual error; constraint value;

-    fiooic: f-value for TF coil operating current / critical
    j_tf_wp_critical: critical current density for winding pack (A/m2)


We still need fiooic:
The current density margin can be set using fiooic.

chris-ashe · 2025-12-03T08:38:30Z

documentation/physics-models/plasma_density.md

 This constraint can be activated by stating `icc = 5` in the input file.

-The value of `i_density_limit` can be set to apply the relevant limit . The scaling value `fdene` can be varied also. 
+The value of `i_density_limit` can be set to apply the relevant limit. The variable `fdene` can be set to scale the constraint bound: `nd_plasma_electrons_vol_avg` / `nd_plasma_electrons_max` <= `fdene`.


i_density_limit isnt always limits that apply to the volume averaged density.

documentation/physics-models/plasma_h_mode.md

chris-ashe · 2025-12-03T08:47:14Z

documentation/physics-models/plasma_h_mode.md

+$$

-**Therefore it is recommended to always use `icc = 15` if trying to simulate a plasma scenario specifically in L or H-mode**
+Again, `fl_h_threshold` can be used to add a margin to the constraint. For example, `fl_h_threshold = 1.25` ensures that `p_plasma_separatrix_mw` is at least 25% less than the threshold power.


Should it not be 25% more?

chris-ashe · 2025-12-03T08:53:08Z

examples/data/scan_example_file_IN.DAT

+******************
+nsweep = 17 * b_tf_inboard_max, maximum peak toroidal field 
+isweep = 6
+sweep = 10.5, 10.4, 10.3, 10.2, 10.1, 10.0


Do we know how close the solution is initially to this inboard leg limit? Would it be better to have a scan example that directly changes a variable instead of a limit?

I think for this PR I will leave it but we should probably create an issue to make a more meaningful example

examples/scan.ipynb

chris-ashe · 2025-12-03T08:58:35Z

process/constraints.py

    )


+@ConstraintManager.register_constraint(22, "MW", ">=")


For the sake of numerical sanity should this not be further down?

timothy-nunn · 2025-12-05T11:16:25Z

Chris also recommends having two separate margins for constraint 15 and 22

timothy-nunn · 2025-12-15T13:53:08Z

Hi All, I have added in the two separate variables and updated the docstrings/docs accordingly. Please check my work carefully to ensure this new way of constraining on the LH threshold makes sense.

I have had to make some changes in physics.py because fl_h_threshold was used there, I am not sure if what I have done is correct?

documentation/physics-models/plasma_h_mode.md

mkovari · 2025-12-15T15:26:15Z

documentation/physics-models/plasma_h_mode.md

+\mathtt{fl\_h\_threshold} \times \underbrace{\mathtt{p\_l\_h\_threshold\_mw}}_{\text{Power from scaling}}  >= \mathtt{p\_plasma\_separatrix\_mw} 
+$$

-**Therefore it is recommended to always use `icc = 15` if trying to simulate a plasma scenario specifically in L or H-mode**


Really icc = 15 is mandatory (unless a different mode is selected), since the plasma must be in some mode.

process/constraints.py

mkovari · 2025-12-15T15:36:51Z

process/constraints.py

+        data_structure.current_drive_variables.n_beam_decay_lengths_core_required * cc,
+    )
+
+


Why is this here all of a sudden?
(Equation to fix number of NBI decay lengths to plasma centre)

I moved the equations around so that they were in order

mkovari · 2025-12-15T15:43:06Z

process/constraints.py

+        - data_structure.physics_variables.p_l_h_threshold_mw
+        / data_structure.constraint_variables.l_mode_threshold_margin,
+    )
+


This code really is a cumbersome way to write a simple equation.
It would be easier to read if each of the return quantities had a name.
The code is fine but I think you may need to rename the margin. l_mode_threshold_margin is a fraction, not a margin. If l_mode_threshold_margin = 0.8 then the margin is 20%.

timothy-nunn linked an issue Oct 22, 2025 that may be closed by this pull request

Remove f-values #3936

Open

timothy-nunn force-pushed the 3936-remove-f-values branch from c6ca56b to 484d46e Compare October 29, 2025 17:31

timothy-nunn self-assigned this Oct 30, 2025

timothy-nunn force-pushed the 3936-remove-f-values branch from 484d46e to 56efc15 Compare November 5, 2025 14:01

timothy-nunn force-pushed the 3936-remove-f-values branch 3 times, most recently from 1bd708f to 6302f99 Compare November 6, 2025 13:45

timothy-nunn force-pushed the 3936-remove-f-values branch from 665329e to cfacb82 Compare November 13, 2025 16:32

timothy-nunn commented Nov 13, 2025

View reviewed changes

timothy-nunn marked this pull request as ready for review November 13, 2025 16:54

timothy-nunn requested review from ajpearcey, chris-ashe, clmould, jonmaddock, mkovari and stuartmuldrew November 13, 2025 16:55

mkovari reviewed Nov 14, 2025

View reviewed changes

timothy-nunn added 7 commits November 17, 2025 16:11

Add simple unit test for constraint functions

f12bf70

Remove F-values from constraint equations

f0cccbb

Remove f-values from input files, input handler, iteration variables,…

759f7b2

… and scan etc

Update reference/test input files to use inequality constraints

3d8fd9b

Update scan example to a more meaningful scan

6b60f61

Add fl_h_threshold back to constraints

f583d7c

Update st_regression.IN.DAT to allow it to converge

ebfbd2a

ajpearcey requested changes Nov 18, 2025

View reviewed changes

timothy-nunn added 3 commits November 25, 2025 11:55

Correct/enhance fl_h_threshold use docs

d200413

Add fdene back

b34dd21

Add fradpwr back

c9b3d71

timothy-nunn force-pushed the 3936-remove-f-values branch from cfacb82 to c9b3d71 Compare November 25, 2025 14:21

Correct dr_shld_inboard in large tokamak regression test

91c4eaf

Add back warnings for fiooic, fjohc, and fjohc0

0a4a3c2

mkovari reviewed Nov 26, 2025

View reviewed changes

Add back fiooic, fjohc, and fjohc0

f4d507b

timothy-nunn force-pushed the 3936-remove-f-values branch from d7f5b38 to f4d507b Compare November 26, 2025 13:47

Add additional constraint to enforce L-mode

a9350cf

chris-ashe requested changes Dec 3, 2025

View reviewed changes

Correct mistakes in HMode docs and scan example

a7fd265

Split fl_h_threshold into two separate margins

54c6aaa

timothy-nunn requested review from ajpearcey, chris-ashe and mkovari December 15, 2025 13:54

mkovari reviewed Dec 15, 2025

View reviewed changes

Correct constraint 22

a400212



		!!! Info "Constraints"
		A full list of constraints is given on the variable description page in the row labelled


		This constraint can be activated by stating `icc = 15` in the input file.

		The scaling value `fl_h_threshold (ixc=103)` can be varied to set the required margin around the threshold.

		data_structure.current_drive_variables.n_beam_decay_lengths_core_required * cc,
		)

Remove f values #3947

Are you sure you want to change the base?

Remove f values #3947

Uh oh!

Conversation

timothy-nunn commented Oct 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-commenter commented Nov 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

timothy-nunn commented Nov 13, 2025

How to update an input file to not use f-values

Uh oh!

mkovari left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

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!

timothy-nunn Nov 25, 2025 • 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

timothy-nunn Nov 25, 2025 • 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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

timothy-nunn commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

timothy-nunn commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mkovari commented Nov 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

timothy-nunn commented Nov 26, 2025

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

timothy-nunn commented Oct 22, 2025 •

edited

Loading

codecov-commenter commented Nov 5, 2025 •

edited

Loading

timothy-nunn Nov 25, 2025 •

edited

Loading

timothy-nunn Nov 25, 2025 •

edited

Loading

timothy-nunn commented Nov 25, 2025 •

edited

Loading

timothy-nunn commented Nov 25, 2025 •

edited

Loading

mkovari commented Nov 26, 2025 •

edited

Loading

timothy-nunn commented Dec 15, 2025 •

edited

Loading