Programs in Physics & Physical Chemistry
|[Licence| Download | New Version Template] adej_v2_0.tar.gz(2576 Kbytes)|
|Manuscript Title: REOS99, a revised program for transition probability calculations including relativistic, correlation and relaxation effects.|
|Authors: S. Fritzsche, C.F. Fischer, C.Z. Dong|
|Program title: REOS99|
|Catalogue identifier: ADEJ_v2_0|
Distribution format: tar.gz
|Journal reference: Comput. Phys. Commun. 124(2000)340|
|Programming language: Fortran.|
|Computer: IBM RS 6000, PC Pentium II.|
|Operating system: IBM AIX 4.1.2+, Linux 6.1.|
|RAM: 20M words|
|Word size: 32|
|Keywords: Atomic physics, Structure, Spectra, Determinant basis, Einstein coefficients, Electron rearrangement, Lifetimes, Multiconfiguration, Dirac-fock, Relativistic, Oscillator strength.|
|Classification: 2.1, 2.2.|
Nature of problem:
Transition probability calculations for (open-shell) atoms have shown that large wave function expansions are often needed to obtain sufficient accurate results. For ab-initio calculations, these wave functions must include the effects of relativity, electron correlations, and the rearrangement of the electron density within a common framework. The REOS99 program accounts for all of these effects for the computation of Einstein coefficients, oscillator strengths, and (radiative) lifetimes. It applies relativistic wave functions from the GRASP92 package  which need subsequently to be expanded into a determinant basis  in order to apply them to REOS99.
Reasons for new version:
Wave function expansions based on several ten thousand determinants and the computation of large transition arrays made a number of major modifications in the source code mandatory. Thereby, considerable improvements have been achieved by taking advantage of the new ANSI Fortran 90/95 standard. Many powerful features of this up-dated language like modules, derived data types, dynamic allocation and deallocation of memory, and others have been exploited in order to develop a revised version which is forearmed for transition probability calculations into the next decade. Efficiency has been gained in particular due to a bitwise storage and treatment of the occupation numbers of the determinants and due to a parallel computation of all transitions and multipole components. The bitwise representation of occupation numbers is maintained by using the bit manipulation procedures of the Fortran 90/95 standard.
None. A few minor limitations arise for atoms with an open f-shell from the available list of the coefficients of fractional parentage (cfp) in GRASP92 which has fully been implemented only for subshells with j<=7/2.
REOS99 is designed for interactive use. Following the interactive dialog at the beginning of the execution, a number of optional branches can be selected for the computation of transition probabilities. Two important branches are the pre-calculation of all radial integrals (as default) and an assumed orthogonality between the orbitals of the initial and final atomic states. Of course, the latter branch neglects a large part of the rearrangement effects of the bound-state density since all "overlap" contributions are omitted from the computations; however, this branch may result in a faster execution and may therefore be attractive for certain applications. A furher acceleration has been achieved by introducing a cut-off parameter to neglect the contributions of the (many-electron) matrix elements for those pairs of determinants for which the product of the mixing coefficients is smaller than a given value; the default cut-off is set to 10**-8 but can be overwritten interactively. In addition, individual transitions which need to be specified by the level numbers of the initial and final atomic states can be selected. The REOS99 program has entirely been rewritten in line with the ANSI standard Fortran 90/95 ; in particular all language features which are considered to become obsolete by the standard have been avoided. All real variables are parametrized by a selected kind parameter and, thus, can easily be adapted to any required precision as supported by the compiler. Presently, the kind parameter is set to double precision (two 32-bit words) in the module rabs_constant. The distribution file includes the code for the program CESD99 (Comput. Phys. Commun. 124(2000)353).
78 minutes on a 266 MHz Pentium II processor
|||S. Fritzsche and C.F. Fischer, Comput. Phys. Commun. 99 (1997) 323.|
|||F.A. Parpia, C.F. Fischer and I.P. Grant, Comput. Phys. Commun. 94 (1996) 249.|
|||S. Fritzsche and I.P. Grant, Comput. Phys. Commun. 103 (1997) 277.|
|||M. Metcalf and J. Reid, Fortran 90/95 Explained (Oxford University Press, 1996).|
|Disclaimer | ScienceDirect | CPC Journal | CPC | QUB|