120 likes | 254 Vues
This summer, I had the invaluable opportunity to intern at The MathWorks, where I focused on enhancing algorithms for sparse matrix computations. My work involved the implementation of ILU(0) and MILU preconditioners, testing their effectiveness on complex cases, and ensuring optimal performance with MATLAB functions. I also participated in developing a comprehensive timeline of our project outcomes. My experience included coding, debugging, and collaborating on product demos, which enriched my understanding of numerical methods and practical software application development.
E N D
A Summer experienceat The MathWorks Malena I. Español Tufts University
Outlines • The company • Timeline • Theory • Implementation • Results
The Company Products
Timeline June July August End Start MEX Files Tests More Tests ILUC Complex Case Build in ILU(0) MILU Final Presentation Bugs Code Code Code CSC Talk at IPTM Product Demos + Help
ILU(0) - R = * A U L Preconditioner: M=L*U
U L ILU(droptol) Droptol=0 Droptol=0.01 A Droptol=0.1
Data Structure for Sparse Matrices:Compressed sparse by columns (CSC)
ALGORITHMS • ILU(0) • JKI format • Data Structure • Trick to go through the nonzero elements • ILUC • Crout format • Data Structure • MILU
Results ...coming!