Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror

Slashdot videos: Now with more Slashdot!

  • View

  • Discuss

  • Share

We've improved Slashdot's video section; now you can view our video interviews, product close-ups and site visits with all the usual Slashdot options to comment, share, etc. No more walled garden! It's a work in progress -- we hope you'll check it out (Learn more about the recent updates).

×

Comment: Re:So slashdotters (Score 4, Insightful) 293

by unperson (#36016524) Attached to: An IP Address Does Not Point To a Person, Judge Rules

But this probably will close the door on the 99 cases out of 100 where an IP actually does equal a bad person who needs to be caught.

I'm not sure about the 99/100 figure. However, even if that's true, I'd argue that just because something is a 99% accurate indicator of crime, it doesn't justify a forfeiture of rights for the other 1%. Is having an IP address linked to an illegal activity justification to open an investigation? Sure. Enough to break in and confiscate property of an individual who has an open WAP living in a populated area? Probably not. Keep in mind people committing internet crimes are "crafty" and know that its important to hide their own identities (often, masking them as the identities of others)

Comment: Re:Grain of salt (Score 1) 157

by unperson (#33987292) Attached to: Astonishing Speedup In Solving Linear SDD Systems

My thoughts exactly...

Symmetric, (Strictly) diagonally dominant matrices are great: Non-singular, real spectrum, diagonalizable...In fact, purely from the fact that the eigenvalues can be bounded away from 0, many iterative methods will have provably fast O(n^2) convergence...beating the classic O(n^3) by an order of magnitude.

I'm not up to speed in the particular tricks used for the Symmetric, DD regime, but certainly one would only "naively" try solving this using Gaussian elimination, due to the special structure. One thing I thought was interesting was that the authors mention that the "previous" fast algorithm solves in:

O(n*log(n)^25).

Well, for n 10^52 (HUUUUUUUUUUUGE!!!) n^2 is less than nlog(n)^25, so there complexity constant becomes really important!!! I can't imagine that the "previous" algorithm was useful (practically speaking!)

-unperson

Civilization, as we know it, will end sometime this evening. See SYSNOTE tomorrow for more information.

Working...