op954d.f

Gets the OPAL95 opacity from the full set of tables (MHP 7/98): a four-dimensional interpolation in $Z$, $X$, $\log T$ and $\log R$ ($R = \rho/T_6^3$). It is the general case, used when the point's composition matches neither pre-interpolated table.

• Inputs: the full set in LLOT95A_MOD: the grids TGR, XXG, R0GR, ZZG, the opacities CAPPA2(126,70,19), and for each $Z$ the offset of its first table NZ (its number of $X$ tables NUMXM is in the module too, but unused? by this or any routine) (loaded by LL95TBL); the indices, weights and extrapolation data from GETOPAL95 in OP95INTERP_MOD (formerly /OP95INDX/, /OP95FACT/ and /OP95EXT/; BZ, BX(K,·), and $\log R$ as RHO_T63L).
• Who calls it: GETOPAL95, when the point's $Z$ differs from the table's ZTAB, and always when the acoustic depth is computed (LADON and LACOUT; GETOPAL95's comment, JVS 04/11).
• Outputs: $\kappa$, $\log\kappa$, QOD, QOT, as in OP952D.

For each of the four nearest $Z$ (table offset NZ(JZ+K-1)) and, within it, the four nearest $X$ (JX(K,J), which may differ between $Z$ values since high-$Z$ sets lack some $X$ tables), the $\log R$ and $\log T$ interpolation of OP952D; then four-point interpolation in $X$ (weights BX(K,·)) and in $Z$ (weights BZ), and the conversion of $\partial\log\kappa/\partial\log T$ to constant $\rho$.

Table layout. The header explains it: "tables are incremented in sets of NZ*NX," so the tables of the third metal abundance ($3\times10^{-4}$) are numbers 21–30; at high $Z$ some $X$ tables are missing ($X < 0.9$ for $Z = 0.1$). The offsets NZ (0, 10, …, 100, 109, 118, in LL95TBL) give $Z = 0.06$ and $0.08$ nine slots each and $Z = 0.1$ eight. LL95TBL stores the tables at $X = 1 - Z$ for $Z$ = 0.06 and 0.08 at NZ(IZ)+10, over the $X = 0$ tables of the next $Z$; this routine reads them with the same indexing, so it finds $X = 1 - Z$ where it was stored, but takes $X = 0$ at $Z$ = 0.08 and 0.1 from the wrong tables (lines 45–48↗; a suspected bug, listed on LL95TBL's page, with a fix).