Comments on A Neighborhood of Infinity: A Monad for Combinatorial Search with Heuristics

Oh gosh!! First I thought it was something complic...

2009-08-12T01:35:16.894-07:00

Oh gosh!! First I thought it was something complicated but it is very well explained!! thanks

Pseudonym, Neat idea. Penalty lists work better w...

2009-07-17T14:54:07.400-07:00

Pseudonym,

Neat idea. Penalty lists work better when you don't have to stray too far from penalty zero, otherwise you'll still get combinatorial explosion. I'm not sure you can ensure that here.

I've just wasted an afternoon trying to shoeho...

2009-07-17T01:06:19.780-07:00

I've just wasted an afternoon trying to shoehorn this into breaking simple substitution ciphers.

The theory I was going on is that if you sort the alphabet of the ciphertext into order of frequency, then it should closely correspond to the English (or whatever language) alphabet sorted into order of frequency. You can then try matching up permutations, trying the "closer" permutations first.

It's very hard.

dysfunctor, > Computer aided learning of forei...

2009-07-05T07:54:02.301-07:00

dysfunctor,

> Computer aided learning of foreign languages is great for drills and repetition

This is one of the subjects I've been thinking about recently and I was pretty surprised to suddenly see a connection with a monad I'd been tinkering with :-)

Sebastian, thanks for the links. The 'diagona...

2009-07-05T07:52:44.046-07:00

Sebastian, thanks for the links.

The 'diagonalisation' function was my starting point. I realised it made sense from an algebraic perspective a long time ago but it only dawned on me a couple of days ago that it orders searches. (It's a weird kind of convolution.) So I'm definitely interested in Spivey's paper as it also looks quite algebraic.

What a neat idea! I can immediately see an applic...

2009-07-05T04:16:07.074-07:00

What a neat idea! I can immediately see an application for computer aided learning.

Computer aided learning of foreign languages is great for drills and repetition, but giving constructive feedback to free-form responses is much harder. For simple grammar exercises, this parser could really help.

Very well explained! I think, the PList mo...

2009-07-05T04:05:23.108-07:00

Very well explained!

I think, the PList monad is almost the same as the implementation of breadth-first search by Mike Spivey in his chapter in "The Fun of Programming" or his paper Algebras for combinatorial search. Your join function performs diagonalisation and Spivey uses the function wrap instead of penalty wich has a slightly different interface but both are inter-definable.

Also, I think if you use a continuation monad as in Reinventing Haskell backtracking you don't need to implement the complex diagonalisation function.