Elsevier Science Home
Computer Physics Communications Program Library
Full text online from Science Direct
Programs in Physics & Physical Chemistry
CPC Home

[Licence| Download | New Version Template] aaej_v2_0.gz(44 Kbytes)
Manuscript Title: CSDUST3: a radiation transport code for a dusty medium with 1-D planar, spherical or cylindrical geometry.
Authors: M.P. Egan, C.M. Leung, G.F. Spagna Jr
Program title: CSDUST3
Catalogue identifier: AAEJ_v2_0
Distribution format: gz
Journal reference: Comput. Phys. Commun. 48(1988)271
Programming language: Fortran.
Computer: RIDGE 32C.
Operating system: RIDGE OPERATING SYSTEM (ROS 3.4).
RAM: 182K words
Word size: 64
Peripherals: disc.
Keywords: Astrophysics, Radiative transfer, Interstellar matter, Circumstellar dust Shells.
Classification: 1.3, 21.2.

Nature of problem:
Infrared observations indicate the presence of dust grains in circumstellar envelopes, interstellar clouds, and galactic nuclei. Dust grains play an important role in the thermodynamics of these objects. Theoretical modeling of the transport of radiation through dust can aid in our understanding of the nature of these objects. To construct realistic models, the effects of multiple scattering, absorption, and re-emission of photons on the temperature of dust grains must be determined self-consistently, as well as the effect of the source geometry on the transport of radiation through the medium. The program solves this problem of radiation transport in a dusty medium with one-dimensional (1-D) planar, spherical or cylindrical geometry.

Solution method:
The combined moment equation of radiation transport, cast in a quasi- diffusion form, and the energy balance equation are solved simultaneously as a two-point boundary value problem using the Newton- Raphson iterative method to determine the equilibrium dust temperature. From the dust temperature structure, the angular distribution of the radiation field is obtained by solving a set of ray equations through the medium.

Reasons for new version:
(1) To include other 1-D geometries (plane-parallel and cylindrical).
(2) To optimize code for vector computers.
(3) To rewrite code in FORTRAN 77 and allow the use of PARAMETER statements so that spatial and frequency grid sizes can be changed more easily for different applications.

Only media with 1-D planar, spherical or cylindrical geometry are considered. The program as presented can handle up to 5 dust components, 60 frequency points, and 100 grid points with 9 impact parameters through the central core. These can be changed easily by modifying one P ARAMETER statement in each subroutine.

Unusual features:
The program solves the radiation transport problem in a dusty medium with 1-D planar, spherical or cylindrical geometry. It determines the equilibrium dust temperature distribution and the characteristics of the internal radiation field. The dust density distribution can be chosen to be either gaussian or a power law. The program can treat linear anisotropic scattering and multi-grain components. In addition to externally heated cases, any of the three geometries can be considered with or without a central heat source. The dust temperature distribution, flux spectrum, surface brightness at each frequency, and the observed intensities (involving a convolution with a telescope beam pattern) are all computed.

Running time:
The typical running time of the program depends on the complexity of the problem and the geometry of the model. For a model with 59 frequency points and 100 grid points, the CPU time per iteration required on the Ridge 32C computer is: 1.5 minutes for planar geometry, 4.9 minutes for spherical geometry, and 13.2 minutes for cylindrical geometry. Using the Eddington approximation (linearly anisotropic radiation field) will reduce this time by about 10% in the planar case, 38% in the spherical case, and 83% in the cylindrical case. Typically 10 iterations are required. Including multi-grain components does not significantly affect the running time. The Ridge 32C computer, a scalar machine, has a speed comparable to a VAX-11/780 and operates at about 2 million instructions per second. On an IBM 3081-D, the CPU time required is reduced by a factor of 4. On a vector machine such as Cray 1S, the difference in timing between the two cases (with and without the Eddington approximation) is minimal since the solution of the ray equations in the program is vectorized.