AS V30 highlights ----------------- Added the *current* FAC potential (via PPOT='FAC') so as to mimic a FAC calculation. (To this end, various AS switches are also set then to align its operation with FAC's.) See AS Working Note No. 3 (ASWN_3.pdf) in the DOCS directory for a detailed discussion. Added a non-uniform ion-sphere plasma potential which is an analytic "fit" to the results of self-consistent (Kohn-Sham) calculations - see Li et al Phys.Plasmas 26, 033301 (2019). Added a look-up table of the ground configuration of every ionization stage of every element up to Z=103, accessed via MXCONF=0 and the number of electrons, NELEC (new), set in &SALEGB (and NZION, set there or &SMINIM). Some simple promotion rules are available to build on this. AS V29 highlights ----------------- Can handle very large angular momentum (l=1000) and true bound orbitals (n=1000). (&SMINIM) KUTDSK controls the write of e-vectors to SCRATCH disk - for very large problems/small memory machines. Re-activated Burgess unique STO potential option which applies scaling parameters to sub-shells. Uniqueness of the potential is useful for (near- and) neutral heavy soecies. See the AS Working Note No. 1 (ASWN_1.pdf) in the DOCS directory for a detailed discussion. (&SMINIM) DDIAG enables the user to select the LAPACK diagonalizer at runtime. The default remains DDIAG='DSYEVD'. However, for matrices of rank greater than 32766 the user should consult the AS Working Note No. 2 (ASWN_2.pdf) in the DOCS directory for compiler dependent advice and multi-threading with the Intel MKL, or other compilers' equivalents, in general. AS V28 highlights ----------------- v27 PARAM MODULE still specified a number of dimensions for non-allocated vectors, which could be exceeded for very large problems. They are now fully (re-)allocatable. All features declared obsolescent by Fortran 2018, or earlier standards, have been removed. Nevertheless, only requires a compiler which (fully) implements the Fortran95 standard. Some Fortran 2003 constructs are present, but not currently utilized (or seen by the compiler.) In additition, all COMMON blocks have been replaced by MODULEs. This completes the Fortran modernization program accelerated in earnest by V27. AS V27 highlights ----------------- No PARAM file required, or used! Large arrays are fully (re-)allocatable. The WRITEUP has been re-written by Roger Dufresne to try and make it more accessible for new users. Simplified relaxed orbital basis: groups of configurations can use their own orbital basis (as opposed to a unique one, or one per configuration). Can restrict eigenpair solutions of H for large CI problems by energy range or no. per symmetry. TBD: uses LAPACK DSYEVX, need to implement Davidson algorithm as in GRASP2K etc. else no real gain in speed as DSYEVX still reduces to tri-diagonal form. KCUT (>0) now neglects corr.-corr. interactions, for large CI problems. 2-body Debye-Huckel plasma potential (perturbative only). Perturbative effective 1-body DH (as per 2-body) optional (historic default, in SE.) AS V26 highlights ----------------- KUTLS restricts LS-mixing to within a configuration. Extensive use of BLAS to calculate autoionization and radiative rates (mixing). AS V25 highlights ----------------- Lande factors. Unitarized DW. Configuration average coupling scheme (inc. EIE). NMETAR/J & NRSLMX bundling on-the-fly (optionally) "relaxed" somewhat to give partial DR, but still highly bundled - needs to be synch'ed with ADASDR needs. AS V24 highlights ----------------- This is an enabling release to handle the half-open f-shell and, as such, is largely transparent to the user, but (see UPDATES for details): (&SALGEB) KUTDSK controls the write of VCCs to disk. NMETAR/J & NRSLMX control bundling on-the-fly of autoionization & radiative rates, respectively, for total DR. AS V23 highlights ----------------- Direct electron-impact excitation: DW Breit-Pauli intermediate-coupling. See Sec.5 of WRITEUP AS V22 highlights ----------------- Direct electron-impact excitation (currently, LS-coupling only.) See Sec.5 of WRITEUP Self-consistent configuration-average potential orbital generation. AS V21 highlights (there was a lot of code restructuring as well): ----------------- Only diagonalize S^2 and L^2 for equivalent electron sub-shells. KCUT acts on "parent" of continuum configs i.e. exactly as it would on the target configs. KCUTCC acts on correlation configs (independent of any N-electron configs) QED corrections. AS V20 highlights (there was a lot of code restructuring as well): ----------------- Generalized Breit, retardation. Interpolate 2-body fine-structure bound-free integrals. Interpolate 2-body non-fine-structure bound-free integrals Alot of stuff about iterated TECs and LECs. Minimize level energy functional. Manuel Bautista's modified TFDA potential. 2-body polarization. AS V19 highlights: ------------------ M_k>1 magnetic multiole radiation. M1+BP algebra only now symmetry grouped, i.e. "fast". Factor 10 speed-up of 2-body fine-structure algebra (there is/was a vector search which is replace by explicit indexing.) STG2-type promotion rules (e.g. NXCITE) as incorporated its CONFIG routine. No need to define continuum configs explicitly. Just define bound N and N+1 configs as in STG2.