Computer codes

T-matrix and related computer codes

This is Arturo Quirantes' webpage in computer codes. Here you will find a compilation of the main programs I've used on my light scattering research. I´ve decided to make them available to the scientific community. You can freely use them for non-profit purpuses, acknowledging original source and author.

All programs are written in Fortran language, and have been tested on Windows (sometimes also Linux) platforms.
My codes are not static, and improvements are made from time to time. Check the Updates section -at the bottom of this page- to see if new versions have been made. And don´t forget to take a look at the Graphics, too.
All codes are signed with my PGP Key (ID 0x4E2031EC)
Feel free to send me any comments, suggestions or additions.

Fax: +34-958-243214
e-mail: aquiran at ugr dot es
Snail Mail:   Departamento de Física Aplicada, Facultad de Ciencias
                     Campus Universitario Fuentenueva
                      18071 Granada (Spain)

Main computer programs

HOMER (Fortran code and digital signature).
HOMER calculates cross-efficiencies (extinction, scattering, absorption) for homogeneous spheres under the Mie theory. It´s adapted for a whole size range (not just one single size parameter), and can even be used for magnetic particles. No more. I could also add polarization or phase function capabilities, but I decided to leave it deliberately simple. After all, simplicity is beauty, and HOMER is ... D' oh!

BART (Fortran code and digital signature).
Not so complex as his sister, BART is based on the Aden-Kerker theory to calculate light-scattering properties for coated spherical particles. Polydispersity is included for either core, coating or entire particle (your pick), but it can also be used for monodisperse calculations. It´s an old code, and I did not try hard to optimize it. You now, BART is an underachiver ... and proud of it.

MARGE (Fortran code and digital signature).
MARGE uses T-matrix methods plus analytical orientation averaging to yield light-scattering data for a collection of randomly oriented, axially-symmetric nonspherical, homogeneous particles. It´s the basic code LISA came from. It´s default-set to spheroids, but any other axisymmetrical particle shape will do. Battle-hardened, efficient, all-terrain, reliable ... reminds me of a Canyonero.

LISA (Fortran code and digital signature).
An extension of MARGE, this program calculates light-scattering parameters (Cross sections, Müller matrix elements) for a collection of randomly-oriented, axially-symmetric nonspherical, coated particles using the T-matrix method. Remarks and comments are included in the code. A remarkable offspring, don´t you think?
NEW: Get your JQSRT paper describing the code here

graphic data ...
Here´s some graphics I´ve done from my computer codes. You might want to use them to compare to your own data; plus, they look cool!

1) Scatering matriz elements for coated spheroids. In all cases q is the core/particle ratio, kr is the equivalent-volume outer size parameter, the refractive indices are 2.32+i0.36 (core) and 1.2 (coating), and eps is the axial ratio (<1 for prolate spheroids, >1 for oblate, =1 for spheres). 3-D graphics for kr=8 have been published in Arturo Quirantes: "Light scattering properties of spheroidal coated particles in random orientation" J. Quant. Spectros. Radiat. Transfer 63, 263-275 (1999) (See my Research file page for the full paper in pdf format).

All figures are jpg. Please note that their sizes range from 300 to 700 kB.

LISA	kr=1	kr=5	kr=8	kr=10
Eps=0.6	kr01e06.jpg	kr05e06.jpg	kr08e06.jpg	kr10e06.jpg
Eps=1.0	kr01e10.jpg	kr05e10.jpg	kr08e10.jpg	kr10e10.jpg
Eps=1/0.6	kr01e16.jpg	kr05e16.jpg	kr08e16.jpg	kr10e16.jpg

... and some useful subroutines

MOE (Fortran code and signature). You need a subroutine to multiply and invert matrices? MOE is your choice. You can do B * (A^-1), or (B^-1) * A, whatever you need. Pity it doesn´t come with a cold Duff... oh yeah!	WILLIE (Fortran code and signature). This subroutine does the same job as MOE, but the input and output matrices are specifically declared as real: AR+iAI, BR+iBI. Sipping Scotch while running it is optional. Ach!
NED (Fortran code and signature). If you need to calculate Clebsch-Gordan coefficients, this is what you´re looking for. Just use NED in your main program, and everything will be pretty okilly-dokilly	APU (Fortran code and signature). For a single Clebsch-Gordan coefficiente, NED will do. But if you want to do calculations for a set of am values, APU might do better. It works fine, fast, 24/7. Just like a Kwik-E-Mart
BURNS (Fortran code and signature). This subroutine gives abscissas and weights for a Gauss quadrature. Yes, I copied it from the Barber-Hill book. So what? You don´t want to mess with me, do you?

Updates:

September 20, 2007: Two bugs corrected. Seems like some Fortran compilers have trouble in accepting ABS(KR) as an integer (all right, it is not). And a note in the source code will ease those who thinks the "db(0)" elements is unacceptable. Thanks, Xie.
April 24, 2007: BART's output yielded not the Müller matrix elements Sij, but their Sij/S11 ratios. Plus, there was a 0.5 factor missing. Now he delivers full matrix element values (S11, S12, S33, S34). Thanks to Brian Barkley for the tip.
January 8, 2007: A few minor bugs corrected in BART. They acted only for the x=0 (no core) or x=y (no shell). Serves him well for invading HOMER's territory. Thanks to Marcus Sherman for spotting it.
November 22, 2005: If the particle is large and highly nonspherical, the calculations of Bessel and Hankel functions done with downward recursion might contain some errors. This is noticeable in multiple-precision calculations only. MARGE and LISA have been updated accordingly.
February 16, 2005: Greetings to WILLIE. It also does matrix multiplication-invertion, like MOE, but relies on real matrices (A=AR+iAI and so on)
January 11, 2005: LISA has how a JQSRT paper describing its usage. Please feel free to DL it here
December 20, 2004: Small tinkering in LISA and MARGE
December 14, 2004: MARGE can now be run in extended precision (-r8)
December 14, 2004. Small changes in APU to reduce the chance of errors when a C-G coefficient is zero.
December 12, 2003: Graphic data from LISA. Help yourself!
November 10, 2003: BURNS. When the abscissae are close to unity, the usual procedure for weight calculation has higher absolute errors. The new modification reduces these errors. LISA and MARGE have also been updated with the new BURNS subroutine. Excellent!
June 11, 2003: APU. A few clebsch-gordan coefficientes, which are supposed to be zero, were not. Some were in fact greater than unity, by Vishnu! Corrected now.
June 11, 2003: LISA. Fine-tuning optimization. Plus, Clebsch-Gordan coefficients are now calculated on a need-to-use basis. As a result, now LISA can do calculations for larger particles.
June 11, 2003: MARGE. First time in this page.
May 27, 2003: LISA. Amazing how just a small change can speed up a program. If you DL'd it before, I strongly recommend you get the update. It´s over 20+ times faster!