Re: Category Theory of Algorithms
- From: Chris F Clark <cfc@xxxxxxxxxxxxxxxxxxxx>
- Date: Wed, 11 Apr 2007 16:56:26 -0400
"the.theorist" <the.theorist@xxxxxxxxx> writes:
To get back to my originally stated intent: to find a
means of moving us to a parallel paradigm.
Actually, there is an interesting point there. I recently was
listening to a many-core researcher, who was making "algorithms" for
Google-like huge arrays of many machines. He did develop a model that
was "physics" like and the "algorithms" self-assembled themselves,
learning the topology of the network and determining which part of the
problem they were supposed to solve.
His algorithms were in some sense both very simple and very complex
and he combined them with and/or type logic. However, it is hard to
describe his work without having you "see" his demonstrations. The
image that kept coming to my mind is that he was programming by taking
eyedroppers of the algorithmic parts and letting them diffuse through
the "solution" in the network.
There is definitely a functional programming connection here, which is
not surprising since the lack of side-effects in functional
programming tends to make it particularly suitable for parallel
replication.
.
- References:
- Category Theory of Algorithms
- From: the.theorist
- Re: Category Theory of Algorithms
- From: Jym
- Re: Category Theory of Algorithms
- From: Pummelo
- Re: Category Theory of Algorithms
- From: the.theorist
- Re: Category Theory of Algorithms
- From: the.theorist
- Category Theory of Algorithms
- Prev by Date: PSPACE closed under union
- Next by Date: Re: PSPACE closed under union
- Previous by thread: Re: Category Theory of Algorithms
- Next by thread: Re: Category Theory of Algorithms
- Index(es):
Relevant Pages
|