tridi_mod.f

35 lines · 1 definitions · on GitHub (7be4133)

Python version: tridi_mod.py, 22 lines, on the right (switch to Python above the code).

Variables on this page (7)

a, b, c, d, gama, json, u

Purpose

TRIDI_MOD holds a tridiagonal system $a_i u_{i-1} + b_i u_i + c_i u_{i+1} = d_i$ and its solution, shared between the routines that set up the implicit steps of rotational mixing and the routines that solve them. Shared by 6 routines. It replaces COMMON/TRIDI/ (MHP via Claude, 10/14/26); JSON comes from DIMENSIONS_MOD.

What its variables hold, who sets them, who reads them.

A, B, C, D (JSON each): the sub-diagonal, diagonal, super-diagonal and right-hand side. Set by DCOEFT (angular-momentum transport, solved by TRIDIA) and CCOEFT (composition transport, solved by CTRIDI), and by GETQUA (the quadrupole moment, solved in place).
U(JSON): the solution, set by TRIDIA, CTRIDI and GETQUA; read by TRIDIA (which turns it into the change of angular momentum), MIXCOM and GETQUA (QUAD in QUADRU_MOD).
GAMA(JSON): the Thomas algorithm's work array, set and read by the solvers; DCOEFT also uses it as work space while setting up.

Design

A blanket USE TRIDI_MOD in each routine. The one-letter names A, B, C, D, U are global in every file that uses the module, so a local of the same name added to one of them later would silently be the module's: the hazard RUNLOOP_MOD's header describes (KNOWN_ISSUES.md issue 22).

Definitions

moduleTRIDI_MOD

lines 19–35 · GitHub

Uses modules DIMENSIONS_MOD

Used by (6 units)

CCOEFT, CTRIDI, DCOEFT, GETQUA, MIXCOM, TRIDIA

Change log

The changes since rp_1

Changed since rp_1. New file: COMMON/TRIDI/ as a module.

1C 2C 3C$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ 4C TRIDI_MOD 5C$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ 6C 10/14/26 MHP via Claude 7C code changed to convert COMMON/TRIDI/ into this module. 8C 9C Imported with a blanket USE, the convention for the conversion. 10C All 6 files declared this block identically, so the USE 11C imports exactly the names the declaration already made global in 12C each scoping unit and cannot change meaning. 13C 14C The types restate what the implicit rules supplied: L is 15C LOGICAL*4, I/J/K/M/N are INTEGER*4, everything else REAL*8. 16C SAVE is explicit rather than relying on module variables being 17C implicitly saved, which holds only from Fortran 2008 while ifort 18C Classic predates it. That matches COMMON's semantics. 19 MODULE TRIDI_MOD 20C 10/20/26 MHP via Claude 21C code changed to take JSON from DIMENSIONS_MOD instead of declaring 22C a PRIVATE copy here. 23 USE DIMENSIONS_MOD 24 25 IMPLICIT NONE 26 SAVE 27 28 REAL*8 :: A(JSON) 29 REAL*8 :: B(JSON) 30 REAL*8 :: C(JSON) 31 REAL*8 :: D(JSON) 32 REAL*8 :: U(JSON) 33 REAL*8 :: GAMA(JSON) 34 35 END MODULE TRIDI_MOD

1"""Python version of tridi_mod.f (YREC 7be4133): a tridiagonal system and its solution, formerly COMMON/TRIDI/. 2 3Translates: tridi_mod.f#TRIDI_MOD (7be4133) 4 5Module-level names, as in the Fortran module, in lower case. Routines that 6USE TRIDI_MOD import this module and write ``tridi_mod.<name>``. 7The system is a[i] u[i-1] + b[i] u[i] + c[i] u[i+1] = d[i]. Index 0 unused. 8""" 9 10import numpy as np 11 12from pyrec import dimensions_mod 13 14_JSON = dimensions_mod.json 15 16# Fortran lines 28–33. 17a = np.zeros(_JSON + 1) # sub-diagonal (dcoeft, ccoeft, getqua) 18b = np.zeros(_JSON + 1) # diagonal 19c = np.zeros(_JSON + 1) # super-diagonal 20d = np.zeros(_JSON + 1) # right-hand side 21u = np.zeros(_JSON + 1) # solution (tridia, ctridi, getqua) 22gama = np.zeros(_JSON + 1) # Thomas algorithm's work array