np-compete: a science hack day 2010 elevator pitch

36
np-compete

Upload: dan-hagon

Post on 08-Jul-2015

686 views

Category:

Education


2 download

DESCRIPTION

The central idea of the np-compete platform is to semantically mark up, with essentially "hand wavy" types of proofs, proofs from the complexity theory literature written in documents that start out as TeX or LaTeX and subsequently get converted into XHTML+MathML. The aim is to find and generate new ideas that might eventually lead to a complete solution of the "P vs. NP" problem.

TRANSCRIPT

Page 1: np-compete: a Science Hack Day 2010 elevator pitch

np-compete

Page 2: np-compete: a Science Hack Day 2010 elevator pitch

Some problems are easy...

Page 3: np-compete: a Science Hack Day 2010 elevator pitch

3 × 7 = ?

Page 4: np-compete: a Science Hack Day 2010 elevator pitch

O(n²)

Page 5: np-compete: a Science Hack Day 2010 elevator pitch

Some problems are a little harder...

Page 6: np-compete: a Science Hack Day 2010 elevator pitch

× = ?

Page 7: np-compete: a Science Hack Day 2010 elevator pitch

O(n³)

Page 8: np-compete: a Science Hack Day 2010 elevator pitch

Some problems are really hard...

Page 9: np-compete: a Science Hack Day 2010 elevator pitch

http://xkcd.com/399/ CC-BY-NC

Page 10: np-compete: a Science Hack Day 2010 elevator pitch

O(n²), O(n³), ... problems are in the complexity class P

O(n!) problems or worse are in NP

Page 11: np-compete: a Science Hack Day 2010 elevator pitch

But are P and NP the same?

Page 12: np-compete: a Science Hack Day 2010 elevator pitch

http://commons.wikimedia.org/wiki/File:P_np_np-complete_np-hard.svg CC-SA

Page 13: np-compete: a Science Hack Day 2010 elevator pitch

“Of all the Millennium Problems, the P versus NP problem is the one most likely to be solved by an “unknown amateur” – someone largely untrained in mathematics, possibly someone very young, who is unknown to the mathematical community.”

Keith Devlin, “The Millennium Problems”

Page 14: np-compete: a Science Hack Day 2010 elevator pitch

“Not only is it relatively easy to understand what the problem says, it is possible that all it will take to solve it is one good new idea.”

Keith Devlin, “The Millennium Problems”

Page 15: np-compete: a Science Hack Day 2010 elevator pitch

np-compete will be a framework to aggregate every new “N vs. NP”

solution idea that’s out there

Page 16: np-compete: a Science Hack Day 2010 elevator pitch

np-compete will filter for any known ideas to find all the

genuinely new ideas

Page 17: np-compete: a Science Hack Day 2010 elevator pitch

The filter will be a competition amongst every solution idea

based on novelty and correctness

Page 18: np-compete: a Science Hack Day 2010 elevator pitch

But how will it work?

Page 19: np-compete: a Science Hack Day 2010 elevator pitch

Take the following example...

Page 20: np-compete: a Science Hack Day 2010 elevator pitch

Can you find whole numbers

p and q so that √2 = p/q?

Page 21: np-compete: a Science Hack Day 2010 elevator pitch

http://en.wikipedia.org/wiki/Square_root_of_2

Page 22: np-compete: a Science Hack Day 2010 elevator pitch

Yeah, I’m gonna need you to summarize that in a tweet

Page 23: np-compete: a Science Hack Day 2010 elevator pitch
Page 24: np-compete: a Science Hack Day 2010 elevator pitch

How about just three words?

Page 25: np-compete: a Science Hack Day 2010 elevator pitch

PROOF BY CONTRADICTION

Page 26: np-compete: a Science Hack Day 2010 elevator pitch

Mathematicians have a phrase for that kind of proof...

Page 27: np-compete: a Science Hack Day 2010 elevator pitch

http://www.flickr.com/photos/craigporteous/3801016250/ CC-BY

Page 28: np-compete: a Science Hack Day 2010 elevator pitch

...hand waving

Page 29: np-compete: a Science Hack Day 2010 elevator pitch

Although there are many proofs that √2 ≠ p/q, they generally

follow one of just a handful of proof patterns

Page 30: np-compete: a Science Hack Day 2010 elevator pitch

The central idea of the np-compete platform is to semantically mark up complete and rigorous proofs with essentially hand wavy types of proofs with the aim of finding

new ideas

Page 31: np-compete: a Science Hack Day 2010 elevator pitch

Mathematicians talk in TeX or LaTeX so this will be our input

Anyone will be able to submit solutions

Page 32: np-compete: a Science Hack Day 2010 elevator pitch

Proof by TeX

“The proof is typeset using TeX or LaTeX, preferably using one of the AMS or ACM stylesheets. When laid out so professionally, it can't possibly have any flaws.”

http://uncyclopedia.wikia.com/wiki/Proof CC-BY-NC-SA

Page 33: np-compete: a Science Hack Day 2010 elevator pitch

np-compete will take a proposed “P vs. NP” solution written in TeX,

convert this source file into XHTML+MathML and allow users to annotate proof structures in

the document with RDFa

Page 34: np-compete: a Science Hack Day 2010 elevator pitch

Think of it as massively collaborative peer-review and

idea generation

Page 35: np-compete: a Science Hack Day 2010 elevator pitch

Get involved!

The project needs people like you:

UI Designers

Semantic Web Experts

Java Hackers

Database Gurus

...

Page 36: np-compete: a Science Hack Day 2010 elevator pitch

np-compete

See you at Science Hack Day 2010

http://sciencehackday.pbworks.com/

Me:

Blog: http://www.danhagon.me.uk/blog/

Twitter: @axiomsofchoice

With thanks to @leipie for the initial collaboration at HackCamp 2010 and Cameron Neylon for discussions and slide template