PyREC scaffolding (from YREC development, 7be4133)/ alfilein.fGlossary

alfilein.f

412 lines · 2 definitions · on GitHub (7be4133)

Python version: alfilein.py, 353 lines, on the right (switch to Python above the code).

Variables on this page (19)

alatm_alpha, alatm_feh, alphas, fehs, gls, ioatma, ishort, laltptau100, loldnextgen, nalpha, nfeh, nga, ngl, nta, nteff, p100ls, pls, t100ls, teffls

Purpose

This file reads a table of model atmospheres by France Allard and collaborators (its comments name the 1999 NextGen format and a newer "Allard" format of 2007) and puts it into the arrays that the atmosphere lookup ALSURFP uses. For each $\log T_\mathrm{eff}$ and $\log g$ the table gives $\log P$ at $T = T_\mathrm{eff}$, and, in the new format, $\log P$ and $\log T$ at $\tau = 100$.

How it fits

  • Input: the file named FALLARD, opened on unit IOATMA. Both come from the run's settings: FALLARD is an argument of PARMIN and SETUPS, IOATMA is in IOUNITS_MOD, and the requested metallicity and $\alpha$-enhancement (ALATM_FeH, ALATM_Alpha) are in NMLPAR_MOD, set by PARMIN.
  • Who calls it: SETUPS, once at start-up, and only when KTTAU = 4 (Allard atmospheres, which the comment there says are used instead of Kurucz below 10,000 K).
  • What it calls: SORT_SHELL (in this file) and ALTABINIT, which checks the table and derives its limits.
  • Output: ALATM01_MOD (the grid TEFFLs, GLs, FeHs, ALPHAs and the tables PLs, P100Ls, T100Ls, with their sizes), read by ALTABINIT, ALPRINT and ALSURFP.
  • Modules: ALATM01_MOD (ALTABINIT, ALPRINT, ALSURFP), NMLPAR_MOD for the requested [Fe/H] and $\alpha$ (formerly /ALATM03/) and the flag LALTPTau100, IOUNITS_MOD for the units (formerly /LUOUT/ and /ALATM03/).

The method

  1. Fill every table entry with $-999$, the mark of an invalid entry.
  2. Read the first record to decide the format: 3.50 in columns 4–7 means the old NextGen format, ALLARD in columns 4–9 the new one. Anything else stops the run.
  3. Old NextGen: fixed sizes, 54 temperatures and 5 gravities; reads $\log g$, the temperatures and $\log P$ only. This format has no $\tau = 100$ values. The temperatures are stored as read, with no conversion: The old file in the repository, input/atmos/allard/Nextgen.p00, already gives $\log T_\mathrm{eff}$ (3.23044892 to 4.00000000, that is 1,700 to 10,000 K, the range its @@README names), so TEFFLs holds $\log T_\mathrm{eff}$ in both formats (lines 173–175↗; settled from the code and the input file, Nov 9, 2026).
  4. New format: one record per ($T_\mathrm{eff}$, $\log g$, [Fe/H], $\alpha$). The routine reads the file twice. The first pass keeps only the records with the requested [Fe/H] and $\alpha$ and collects the distinct $T_\mathrm{eff}$ and $\log g$ values, which it then sorts. The second pass places each record's three values at its indices, and stores $\log_{10} T_\mathrm{eff}$ in TEFFLs.
  5. Call ALTABINIT.

Gaps in the grid (combinations of $T_\mathrm{eff}$ and $\log g$ absent from the file) keep the value $-999$; ALTABINIT then finds each row's valid range.

Design

Every error ends the run with STOP 3, the exit status for input errors, after a message to the screen and to ISHORT. Some messages name other routines ("AtTabInit Pass 1 File Error"), and the closing comment says "END OF ALINITTAB," left from earlier names.

Suspected bugs, checked by running the code

Each bug is marked by what running the code showed: (changes results, large) or (small), the fix moving a run clearly beyond a null change; (stops the run); (wrong output), the fix changing only what is written; (harmless in the runs), within the noise; (never occurred), the code ran but the fault did not; (not reached); (cannot be reached), a guard prevents it; (withdrawn), the suspicion was wrong.

1. (never occurred) Two tolerances for [Fe/H] and $\alpha$.

(lines 209–210↗; lines 296–297↗; lines 326–327↗) Both passes skip a record when $|\Delta[\mathrm{Fe/H}]| > 10^{-5}$ or $|\Delta\alpha| > 0$ (an exact match), measured from the requested ALATM_FeH, ALATM_Alpha; the second pass then drops the record again when $|\Delta[\mathrm{Fe/H}]| \ge 10^{-6}$ or $|\Delta\alpha| \ge 10^{-6}$. For $\alpha$ the second test changes nothing; for [Fe/H], a request between $10^{-6}$ and $10^{-5}$ from the file's value adds the records' $T_\mathrm{eff}$ and $\log g$ to the grid in the first pass, but stores none of them, so every entry stays at $-999$ and ALTABINIT stops the run with "Bad input Allard Table," not with a message that names the [Fe/H]. Only with KTTAU = 4 and a requested [Fe/H] that is not exactly a value of the file (whose values have two decimals, format 915).

A possible fix:

alfilein.f

@@ lines 323–330 @@

 C we now verify that we have the correct FeH and Alpha

-      if (DABS(FeH -ALATM_FeH) .ge. 1D-6)  goto 440     ! Wrong FeH, bypass item
-      if (DABS(Alpha -ALATM_Alpha) .ge. 1D-6)  goto 440 ! Wrong Alpha, bypass item
+C 11/09/26 KC via Claude
+C code changed to drop the second test of FeH and Alpha, whose 1D-6
+C dropped records that the test after label 410 (1D-5) had kept.
+C     if (DABS(FeH -ALATM_FeH) .ge. 1D-6)  goto 440
+C     if (DABS(Alpha -ALATM_Alpha) .ge. 1D-6)  goto 440


 C We now have the correct indices for our tables, i1 for the Teff-direction

Checked by running YREC (Nov 9, 2026; a line added at each suspected path counted how often it was taken, the outputs otherwise unchanged): Of 34 runs (26 of YREC's cases, from a 0.04 $M_\odot$ brown dwarf and the solar models to 9 $M_\odot$, rotation, and helium burning, and 8 made-up ones; one of these, with FCORE = 1.0, stopped in its first second), only testsuite/Test_solarGS98_allard uses the Allard table (Nextgen2.all, KTTAU = 4). There the second test ran 316 times, each with [Fe/H] = 0 requested and found, and it never dropped a record. No case sets ALATM_FeH (the default is 0), and a value written as in the file reads to the same number as the file's two-decimal field, so the condition needs a request between $10^{-6}$ and $10^{-5}$ off a value of the file, which no case or document makes; it was not run. No fix was run. Intent: none found.

2. (never occurred) Loop bounds for the new [Fe/H] and $\alpha$ values.

(lines 248–262↗) The checks for new [Fe/H] and $\alpha$ values loop do i = 1,nGL instead of 1,nFeH and 1,nALPHA, and so compare with entries of FeHs and ALPHAs that were never set. Harmless today: Only records within $10^{-5}$ of the requested [Fe/H] and with exactly the requested $\alpha$ get this far, and since the file's values have two decimals they are all equal to the first, found at i = 1; nFeH and nALPHA stay 1, and only ALPRINT prints them. It would count wrong values once the filter let several [Fe/H] or $\alpha$ through.

A possible fix:

alfilein.f

@@ lines 246–252 @@


 c Now check FeHs, increase counter if needed
-      do i = 1,nGL  ! Skip out if any old FeH is a match
+C 11/09/26 KC via Claude
+C code changed to loop over the FeH and ALPHA values found so far,
+C not over the GL values.
+      do i = 1,nFeH  ! Skip out if any old FeH is a match
         if (DABS(FeH-FeHs(i)) .lt.1D-6) goto 314
       enddo
         nFeH = nFeH+1      ! count the next FeH
@@ lines 254–260 @@
  314  CONTINUE

 c Now check ALPHAs, increase counter if needed
-      do i = 1,nGL  ! Skip out if any old ALPHA is a match
+      do i = 1,nALPHA  ! Skip out if any old ALPHA is a match
         if (DABS(ALPHA-ALPHAs(i)) .lt.1D-6) goto 316
       enddo
         nALPHA = nALPHA+1      ! count the next ALPHA

Checked by running YREC (Nov 9, 2026; a line added at each suspected path counted how often it was taken, the outputs otherwise unchanged): In testsuite/Test_solarGS98_allard, the only Allard run, the two loops ran for 315 records, all with [Fe/H] = 0 and $\alpha$ = 0; the first entry matched every time, so no unset entry was compared. It would need records of several [Fe/H] or $\alpha$ to pass the filter, which no case or table does. No fix was run. Intent: none found.

Definitions

subroutineALFILEIN

lines 32–370 · GitHub

Reads the Allard atmosphere file, old or new format, keeps the records with the requested [Fe/H] and $\alpha$, fills ALATM01_MOD and calls ALTABINIT. Stops the run on any error.

Uses modules ALATM01_MOD, IOUNITS_MOD, NMLPAR_MOD

Called by SETUPS

Flowchart of its calls, in order (5)
SORT_SHELL278→SORT_SHELL279→SORT_SHELL280→SORT_SHELL281→ALTABINIT350

subroutineSORT_SHELL

lines 377–408 · GitHub

Shell's sort from Numerical Recipes, ascending, in place, with the increments $1, 4, 13, 40, \dots$ ($h \to 3h + 1$).

Called by ALFILEIN ×4

Change log

The changes since rp_1

Changed since rp_1. Declarations only: COMMON blocks to modules (/ALATM01/; /ALATM03/ split between NMLPAR_MOD and IOUNITS_MOD; /LUOUT/ to IOUNITS_MOD; /ALATM04/, which held only dummies, removed), IMPLICIT NONE; the EXTERNAL sort_shell line, which a misplaced column had made a continuation of the /ALATM04/ declaration, is now a statement of its own. The misspelled flag of former question 1 was then declared as a local LOGICAL, so the question stands.

Changed since a8a5950. The old-format branch now tests the namelist flag LALTPTau100 (from NMLPAR_MOD) instead of the misspelled, never-set local LATMTPTau100, so an old NextGen 1 file is refused when $\tau = 100$ values are requested (this settles former question 1). The error exit stops with STOP 3, the code for input errors, instead of STOP 9999 (KNOWN_ISSUES.md, issue 38).