Convert iteration_variables.f90 to Python

[timothy-nunn]

Convert iteration_variables.f90 to Python.

Similar to #3552, I have created a config class and created an entry (in a dict this time) for each existing iteration variable.

Notes

• I did not convert this q95/qbar multiple iteration variable name feature

  PROCESS/source/fortran/iteration_variables.f90

  Lines 4176 to 4177 in 1069123

  	if ((ixc(i) == 18).and.(i_plasma_current /= 2)) name_xc(i) = 'q95'
  	if ((ixc(i) == 18).and.(i_plasma_current == 2)) name_xc(i) = 'qbar'

• The following if statement always resolved true

  PROCESS/source/fortran/iteration_variables.f90

  Lines 4203 to 4207 in 1069123

  	if (abs(xcm(i)) > 1.0D-99) then
  	scale(i) = 1.0D0/xcm(i)
  	else
  	scale(i) = 1.0D0
  	end if

  because of this previous check

  PROCESS/source/fortran/iteration_variables.f90

  Lines 4189 to 4193 in 1069123

  	if (abs(xcm(i)) <= 1.0D-12) then
  	idiags(1) = i ; idiags(2) = ixc(i)
  	write(*,*) 'Iteration variable ',ixc(i),'(',trim(lablxc(ixc(i))),') is zero.'
  	call report_error(55)
  	end if

• Validation in convxc now uses the actual iteration value not its scaled value (where xc(i) below is a scaled iteration variable)

  PROCESS/source/fortran/iteration_variables.f90

  Lines 4438 to 4450 in 1069123

  	if (abs(xc(i)) <= 1.0D-12) then
  	idiags(1) = i ; idiags(2) = ixc(i)
  	write(*,*) 'Iteration variable ',ixc(i),'(',trim(lablxc(ixc(i))),') is zero.'
  	call report_error(58)
  	end if
  	! Crude method of catching NaN errors
  	!if ((abs(xc(i)) > 9.99D99).or.(xc(i) /= xc(i))) then
  	if (isnan(xc(i)).or.(abs(xc(i))>9.99D99)) then
  	idiags(1) = i ; idiags(2) = ixc(i) ; fdiags(1) = xc(i)
  	call report_error(59)
  	end if

• Removed validation of scale in convxc because validation in the loading should protect against very small scales

  PROCESS/source/fortran/iteration_variables.f90

  Lines 4452 to 4455 in 1069123

  	if (abs(scale(i)) < 1.0D-99) then
  	idiags(1) = i ; idiags(2) = ixc(i)
  	call report_error(60)
  	end if

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 47.50000% with 42 lines in your changes missing coverage. Please review.

Project coverage is 36.13%. Comparing base (05d827e) to head (ac5eca1).
Report is 116 commits behind head on main.

Files with missing lines	Patch %	Lines
process/iteration_variables.py	45.71%	38 Missing ⚠️
process/optimiser.py	50.00%	2 Missing ⚠️
process/caller.py	50.00%	1 Missing ⚠️
process/main.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3570      +/-   ##
==========================================
- Coverage   36.54%   36.13%   -0.42%     
==========================================
  Files          87       88       +1     
  Lines       22190    22106      -84     
==========================================
- Hits         8110     7988     -122     
- Misses      14080    14118      +38     

☔ 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]

This section of code could have been simplified

PROCESS/source/fortran/iteration_variables.f90

Lines 4202 to 4210 in 159ac62

	do i = 1,nvar
	if (abs(xcm(i)) > 1.0D-99) then
	scale(i) = 1.0D0/xcm(i)
	else
	scale(i) = 1.0D0
	end if
	scafc(i) = 1.0D0/scale(i)
	xcm(i) = xcm(i)*scale(i)
	end do

As you will see, we are essentially performing the following calculations
$s=1/x$
and then later on
$x:=x\times s$

so it follows that when $x$ is set at the end of the snippet, it is to $1$. While this is correct, simplifying the code as such causes a difference in the variable value that causes differences in once-throughs. My guess is that this happens because of some floating point error that accumulates. As a result, some weird expressions exist within the code that look like they should have been simplified.

[jonmaddock]

Due to my inaction on this, there are now some conflicts. Could you resolve these before I review?

[jonmaddock]

Great, nice to get the dicts working nicely too.