URGE to Compute Wiki : Seminar Fall

Thursday, Dec 2

An introduction to Python

Bobby Dygert, Matt Heavner, Michael Skvarch, Devanshu Pandey, John Ringland

Max's Idle transcript. See also this tutorial.

Prof. Carbonara's data-scraping/plotting problem: joaquin.py (now includes lines to save as png & svg), joaquin2.py (skips over the non-data), balloon_launch.txt (whole file including headers), balloon_launch.dat (just the data). Bobby suggests that "regular expressions" would be useful here: an example of pattern matching using re: reg.py.

A Matlab to NumPy translation: cheb.m, cheb.py. See also Numpy for Matlab Users.

Really nice plots with matplotlib: matplotlib_ellipses.py

. See also this gallery.

Thursday, Nov 25 Thanksgiving Break

Thursday, Nov 18

Roundtable discussion of research project status, publication and conference activities and plans.

Friday, Nov 12

Alberto Apostolico

has accepted Valentin's invitation to visit and give a talk on a high level view of pattern matching and the discovery of surprising motifs in sequences.

Thursday, Nov 11, 3:30pm

Eric Allender

CSE distinguished speaker series

Sep 30: (4:30pm start)

P, NP, and the Open-Source Model

Kenneth W. Regan

UB Computer Sci. & Eng.

Dr. Vinay Deolaikar of Hewlett Packard Labs in Palo Alto, CA, has claimed a proof of P != NP, which is one of the six remaining Millennium Prize Problems in Mathematics, as well as the great 54-year open problem in theoretical computer science. After a leak and wide news dissemination disrupted his original desire for private circulation of his 102-page draft paper, he released it on Aug. 8th and consented to have it linked and discussed on the weblog of Professor Richard J. Lipton of Gerogia Tech, rjlipton.wordpress.com, on which I am the chief assistant. The story beyond the story is that over the next 12 days an unprecedented---and notably civil---Internet discussion of the paper's merits and flaws unfolded in over 1000 comments on that blog, plus a couple hundred on other blogs, by Fields Medalists Terence Tao and Timothy Gowers and leading CS researchers down to anonymous unknowns who wrote well and had good insight into the paper. I summarized them in the technical part of posts every other day, and all this was coordinated with a wiki page under the PolyMath framework for mass research collaboration sponsored by Gowers and others. This story was picked up by the NY Times, Forbes, Science News, other media, and hailed by writers such as Stephen Landsburg, http://www.thebigquestions.com/2010/08/16/o-brave-new-world/

My talk will introduce the P versus NP problem, and explain how it's almost a red herring, obscuring the real problem of our incredible lack of knowledge about the true power of computation. We have instead discovered barriers to gaining such knowledge, including one mechanism by which a "too-easy" proof of P != NP would damage the security of the RSA cryptosystem and so imperil the world economy. Next I will explain how Deolalikar's proof strategy attempts to evade the barriers, but evidently trips up on fundamentals. All of this speaks not only to the subject matter but also the human endeavor of targeting one's research. Then I will assess what happens to this endeavor when it goes to many humans, by appeal to what computer programmers have learned about Open-Source Software for decades.

Here is the PDF of my talk. I added a slide with the links I used. I should also mention a much jazzier talk by Ron Fagin and Ryan Williams and others, with video, at http://www.almaden.ibm.com/cs/disciplines/pm/#PvNPtalk (IMHO, the objection about projections is much more telling---I tried to meet Ryan's objection at the time, and I might get back to it after other things calm down.)

Here's a Project Euclid page for the paper by Andrew Bremner which I mentioned: http://projecteuclid.org/DPubS?verb=Display&version=1.0&service=UI&handle=euclid.em/1243429952&page=record I've been attacking the equations with /Singular/, but no great success to date. But it's IMHO the one way an undergrad might stand a chance to "destroy the world economy" (quote by Alexander Razborov from the ASL meeting last March)---if you can find solutions at all sizes. The paper where my Montreal sabbatical host is involved, and with the link to Blum-Shub-Smale theory, is http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.154.1832 Smale's "problem" survey is linked from here: http://en.wikipedia.org/wiki/List_of_unsolved_problems_in_mathematics His book on B-S-S theory is reviewed at http://portal.acm.org/citation.cfm?id=504192.1005765

Sep 23: Gonsalves conclusion. Cusick on cryptography.

Sep 16: Gonsalves on models of superconductivity/fluidity

Sep 9: Brimkov on publishing