neutrino.f

1. (changes results, small) Checked by KC (Nov 10, 2026) Metals other than C and O are left out.

(lines 52–58↗; nulosses.f, lines 87–95↗) Nitrogen, neon, iron, and the rest of $Z$ are not passed (ZC and ZO are ¹²C + ¹³C and ¹⁶O + ¹⁸O, engeb.f, lines 1231–1232↗), so the four mass fractions sum to less than one, and AZBAR does not renormalize them: $\bar A = 1/\sum_i X_i/A_i$ is too large by the factor $1/\sum_i X_i$ against the four fractions renormalized, while $\bar Z = \sum_i Z_i X_i/A_i \big/ \sum_i X_i/A_i$ does not depend on the sum, so $Y_e = \bar Z/\bar A$ is too small by the factor $\sum_i X_i$ against them; against the real mixture, $Y_e$ misses the omitted metals' electrons, about $(1 - \sum_i X_i)/2$. What follows: The neutrino losses, which SNEUT computes from $\bar A$, $\bar Z$, and $Y_e$, are those of a mixture with too few electrons. Only with LNULOS1 (off by default; set in many of YREC's cases) and above $\log T$ = TCUT(5) (7.5 in YREC's cases).

A possible fix, two ways. (a), the one the run below used, puts the metals other than carbon and oxygen in the oxygen slot, so that the mass fractions sum to one, as AZBAR assumes. It treats every nucleus other than hydrogen as having $Z/A = 1/2$, as YREC's older loss routines do (EMUE = 0.5*(1.0+X), engeb.f, line 1273↗), and gives the missing metals $A = 16$, which the code does not specify anywhere:

neutrino.f

@@ lines 52–58 @@
          xmass(1) = x
          xmass(2) = y
          xmass(3) = zc
-         xmass(4) = zo
+C 11/10/26 KC via Claude
+C code changed to put the metals other than C and O in the oxygen slot,
+C so that the mass fractions sum to one, as AZBAR assumes.
+         xmass(4) = 1.0d0 - x - y - zc


        call nulosses(temp,den,snu,xmass,ymass,aion,zion,dsnudt,dsnudd)

Or (b), not run: AZBAR divides by the sum of the mass fractions it is given, so that abar and zbar are the means over the four species even when they do not sum to one. This is the same as renormalizing the four fractions: It needs no assumption about the missing metals, but gives them the four species' mean $Z/A$, hydrogen's included:

nulosses.f (another file)

@@ lines 82–97 @@
       integer          i,ionmax
       double precision xmass(ionmax),aion(ionmax),zion(ionmax),
-     1                 ymass(ionmax),abar,zbar,zbarxx,ytot1
+     1                 ymass(ionmax),abar,zbar,zbarxx,ytot1,xtot


       zbarxx  = 0.0d0
       ytot1   = 0.0d0
+      xtot    = 0.0d0
       do i=1,ionmax
        ymass(i) = xmass(i)/aion(i)
        ytot1    = ytot1 + ymass(i)
        zbarxx   = zbarxx + zion(i) * ymass(i)
+       xtot     = xtot + xmass(i)
       enddo
-      abar   = 1.0d0/ytot1
-      zbar   = zbarxx * abar
+C 11/10/26 KC via Claude
+C code changed to divide by the sum of the mass fractions given, so that
+C abar and zbar are the means over those species even if it is not one.
+      abar   = xtot/ytot1
+      zbar   = zbarxx/ytot1
       return
       end

How the code itself shows it. AZBAR forms ytot1 as the sum of xmass(i)/aion(i) and sets abar = 1.0d0/ytot1 and zbar = zbarxx * abar (nulosses.f, lines 87–95↗), so abar is the mean nucleon number only when the xmass sum to one (the mean charge zbarxx/ytot1 is a ratio and does not depend on it), and SNEUT takes ye = zbar*abari (sneut.f, line 192↗), which therefore does. NEUTRINO passes x, y, zc, and zo alone (lines 52–55↗), and ENGEB sets ZC = XC12+XC13 and ZO = XO16+XO18 (engeb.f, lines 1231–1232↗), so ¹⁴N and the metals heavier than oxygen are missing.

Checked by running YREC (Nov 10, 2026, the runs repeated the same day with a private start model; a line added at each suspected path counted how often it was taken, the outputs otherwise unchanged): In testsuite/Test_m1p0feh+0p0GS98_base_HeIgnite (1 $M_\odot$ from the end of the main sequence, at 9.870 Gyr, up the red giant branch; the run stops near helium ignition, at 12.59 Gyr, when the center's $\log T$ leaves the OPAL 95 table, with STOP 1), NEUTRINO was called 33,460,185 times, at $\log T$ = 7.50 to 8.07, and the four fractions summed to 0.984 to 0.991 every time. With fix (a), the neutrino luminosity changes beyond the noise: To 12.5 Gyr it moves by up to 0.000072 $L_\odot$, against 0.0000011 for a null change (nearly every opacity, all with $3 \le \log T \le 9$ and $-6 \le \log\rho \le 9.75$, times $1 + 10^{-10}$, on the same case), and in the last model it is 0.876 $L_\odot$ instead of 0.845, against 0.846 for the null change. The star's luminosity stays within the noise, and its center moves a little beyond it: To 12.5 Gyr $\log L$ moves by up to 0.00016 dex, against 0.00084 for the null change, and central $\log T$ by 0.00020, against 0.000030; the track has 8,611 models instead of 8,571 (8,571 for the null change, too). In testsuite/Test_solarGS98_difrotmixper18p4d1p0, which also sets LNULOS1, NEUTRINO was never called: The Sun's center stays below $\log T$ = 7.5. Intent: The header says it "currently sets up just hydrogen, helium, carbon and oxygen" (lines 4–5↗).