Skip to content

Latest commit

 

History

History
16 lines (13 loc) · 1.57 KB

README.md

File metadata and controls

16 lines (13 loc) · 1.57 KB

cc_finder

cc_finder is a parallelized code that finds the critical curves and caustics of micro-lensing star fields using the parametric representation of the critical curves from Witt (A&A, Vol. 236, p. 311, 1990). The code is written using NVIDIA's CUDA computing platform. Instead of a sequential method of finding the initial roots for phi=0 as described in Witt 1990 and Witt, Kayser, and Refsdal (A&A, Vol. 268, p. 501, 1993), we use a parallelized implementation of the Aberth-Ehrlich method, a cubically convergent algorithm that allows for simultaneous approximations of the roots of a polynomial (https://en.wikipedia.org/wiki/Aberth_method). The paper that inspired this work, which provides more detailed explanations of the mathematical and computational methods, is: Kahina Ghidouche, Abderrahmane Sider, Raphael Couturier, Christophe Guyeux. Efficient high degree polynomial root finding using GPU. Journal of Computational Science, 2017, 18, pp. 46 - 56.

Once initial roots are found for phi=pi, further roots for varying phase 0<=phi<=2*pi are found in parallel as well, simultaneously traversing pi -> 2*pi and pi -> 0. We directly use all the stars in a field, avoiding complex Taylor expansions as described in WKR 1993. It is possible that further speedups could be gained through such approximations.

Values for kappa_tot, shear, theta_star, kappa_star, num_stars, star input file, num_phi (number of steps to use for phi from 0->2pi), num_branches, random seed, output file type (.bin or .txt), and output filename prefix, can all be input through the command line.