Aaditya V. Rangan

Courant Institute of Mathematical Sciences, New York University,
251 Mercer Street, New York, NY 10012.
Phone: (212) 998-3303, email: rangan (at) cims (dot) nyu (dot) edu
webpage: http://www.cims.nyu.edu/~rangan/
Office Hours Fall 2015: Thursday 1:30pm-3:25pm
Room: 1123 WWH

Research Interests: Applications of numerical-analysis and scientific-computing to the biological sciences.

CV (last updated 10/02/15)

List of publications (last updated 10/02/15)

Some Recent Preprints:
This paper investigates the potential mechanisms underlying the afterhyperpolarization phase observed within the Manduca Sexta antennal lobe:
"Intrinsic and network mechanisms constrain neural synchrony in the moth antennal lobe" (Manuscript) (Supplemental)
Here is a zip-file of a directory containing an executable "lsy.exe" which simulates a simple neuronal network.

Some Recent Papers:
A.V. Rangan, A simple filter for detecting low-rank submatrices, J. Comp. Phys. 231(7): 2682-2690, (2012). (link)

Recently I've been using this approach to bicluster gene-expression data. I believe the results are rather promising.

Here is a tutorial which presents an example of this algorithm applied to a standard data set:
(tutorial in pdf),(tutorial in pptx),(appendices: still a work in progress!).
The files associated with this tutorial are bundled into the following archives:
(level-0): This "level-0" archive contains all the basic matlab files needed to generate the output shown in the presentation. Each of the matlab files is documented internally (e.g., you can run '>> help tutorial_w1;' to see what the file tutorial_w1.m does). Note that this archive does not include the output itself. Thus, some runtime will be necessary if you want to bicluster everything yourself.
(level-1): Just like the "level-0" archive, this "level-1" archive contains the matlab files needed to generate the output shown in the presentation. In addition, the "level-1" archive also contains the output for the original data. This archive does not include all the label-shuffled trials used to generate p-values for the original data. Thus, some runtime will be necessary if you want to generate p-values yourself.
(level-2): As above, this "level-2" archive contains all files needed to generate the output. In addition, the "level-2" archive also contains all the output for both the original data and the label-shuffled permutations. If you download this archive you should be able to immediately generate the summary plots by running 'tutorial_summarize' or 'tutorial_plot'.

A significantly more efficient implementation of this algorithm (written in C) is available at:
This implementation also includes several subroutines which perform binary vector-vector, matrix-vector and matrix-matrix operations.