htimer.f

HTIMER sets the time step of the next model. Several criteria each propose a step, and it takes the most restrictive one (its comment): structure changes (PTIME: the change of $T$, $P$, $r$ and $L$ at any point since the previous model), angular-velocity changes for rotating models (WTIME: the largest fractional change of $\Omega$), hydrogen burning (XTIME: the hydrogen burnt in the core, or in the shell once the core is exhausted), helium burning (YTIME: the helium burnt at the temperature maximum) and, since 04/14, the motion of the model in the envelope-fitting "triangle" (ENTIME: the change of $\log T_\mathrm{eff}$ and $\log L$); and the step may not grow by more than a factor ATIME(13) from the previous one (default 1.5, "originally = 1.5" by PARMIN's comment; under its newer name TIME_MAX_DT_FRAC, also 1.5, when REMAP applies the newer names).

• Who calls it: the main program, main.f: at the start of each kind card, and after each converged model to set the next step. main.f always passes $|\Delta t|$ after a model (DELTS = DABS(DELTS)), and makes the step negative itself before a relaxation step.
• Inputs: the previous step DELTS (s); the model (HD, HL, HS1, HS2, HT, HP, HR, HCOMP, OMEGA, M, TEFFL); JCORE, JXMID, JXBEG (core and burning-shell indices); TLUMX (the luminosities by source, in solar units: elements 1–4 hydrogen burning, 5 helium burning, 7 the gravitational term); the age DAGE (Gyr); the run number NK.
• Outputs: the new DELTS (s), DELTSH (= DELTS, except on the relaxation branch, where it is left as passed unless the final-age cut applies), DDAGE (the step in years), and DWMAX (from WTIME: the largest change in $\Omega$, used by the rotational-mixing routines to set their number of substeps).
• Modules: NMLPAR_MOD holds the settings: ATIME (the step controls, formerly /CTLIM/); per run, SETDT for a fixed step and ENDAGE for a final age, and the stop values for the central abundances END_DCEN, END_XCEN, END_YCEN (since 10/24, which is why LENDAG alone no longer means a stop age); LPTIME (formerly /CT3/; HTIMER can switch it off; default true, set by PARMIN's namelist or by REMAP from LSTRUCT_TIME; main.f restores it at each solar-calibration cycle); LTRIST (the triangle governor, formerly /GOVS/) and LROT (formerly /ROT/). RUNLOOP_MOD holds the per-run flags LSETDT (fixed step) and LENDAG (final age); with SETDT and ENDAGE these were /SETT/, set by PARMIN and shared with main.f, CALCAD, CHKSCAL, SETCAL, SETSCAL. CONSTANTS_MOD gives CSECYR (seconds per year).

1. A negative DELTS means the model is being relaxed, not evolved: skip to the end.

2. A fixed step for this run (LSETDT(NK)): $\Delta t$ = SETDT(NK) years, skip to the end.

3. If $\log T_c > 7.1$ and the gravitational luminosity TLUMX(7) is negative, switch the structure criterion off (LPTIME = false) for the rest of the job. Its comment is cut short ("this is done when"); the purpose is in REMAP's comment, "this is used only in the pre-MS and disabled for MS stars automatically," and in the documentation of LSTRUCT_TIME, "disabled once Lgrav<0 (minimum radius, close to ZAMS)" (settled from the code and the documentation, Nov 9, 2026: lines 96–104↗; remap.f, lines 283–285↗).

4. Collect the proposals, each $10^{20}$ s when not applicable or turned off: $\Delta t_P$ from PTIME (if LPTIME), $\Delta t_\Omega$ from WTIME (if rotating), $\Delta t_H$ from XTIME (if the H-burning luminosity exceeds $10^{-34}$), $\Delta t_{He}$ from YTIME (likewise for He), $\Delta t_E$ from ENTIME (if LTRIST and not LPTIME). The structure and rotation proposals ($\Delta t_P$, $\Delta t_\Omega$) below 1 s are replaced by $10^{20}$ ("a zero timestep in the model will fritz out the code"); the others are not checked (for ENTIME, see bug 1).

5. The new step: $$\Delta t = \min\left(f\,\Delta t_\mathrm{prev},\ \Delta t_H,\ \Delta t_{He},\ \Delta t_\Omega,\ \Delta t_P,\ \Delta t_E\right),\qquad f = \texttt{ATIME(13)},$$ leaving out the first term when $\Delta t_\mathrm{prev} \le 1$ s.

6. At the end: $\Delta t \leftarrow |\Delta t|$, DDAGE $= \Delta t/$CSECYR; if the run stops at an age (LENDAG(NK) and ENDAGE(NK) > 0), shorten the step so that the age does not pass ENDAGE (in years; the age DAGE is in Gyr).

The switch-off of LPTIME changes a module flag (a namelist setting) permanently, and is announced with a bare WRITE(*,…). A block of commented-out code ("Grant added") once tightened ATIME(5) and ATIME(6) with the helium luminosity.

Physics Options (Timestep criteria): LTRIST "restricts the timestep during the Hertzsprung gap"; in the code it keeps the model from moving more than TRIDT in $\log T_\mathrm{eff}$ or TRIDL in $\log L$ in a step (the size of the envelope triangle), at any phase, when the structural criterion (LPTIME, LSTRUCT_TIME) is off. The step limits ATIME are documented under their newer names on REMAP's page; ENDAGE (Control Options) ends a run at an age.