@vzn I have yet to see or hear anyone on a cluster using something that isn't Fortran 90+, C, or C++ for heavy computations
And I would argue that the "mainstream languages" are the ones diverging from HPC as programming languages were developed for fast calculations in the first place!
what you say may be true for scientific comp but note a lot of cluster computing is not nec sci comp nowadays. eg google. & amazon cloud. (a significant part of latter is serving up netflix movies) etc
also the sci comp is not nec leading all the innovation in cluster computing.
scicomp is cool but not nec "center of universe" wink
you may see some of the more cutting edge cluster type software innovations work their way into scicomp. although scicomp does seem conservative that way. in the ways you espouse. earlier languages etc.
A friend of mine actually works on the OrangeFS with Palmetto. I asked him once how many people use the nVidia's. His response was laughter followed by, "almost no one"
@vzn Nah, none of them work. If you rely on a third party to give credit like the BOINC project, then you essentially give them ability to print money and hacker can simply hack the system.
@vzn haha, you just remind me the gmail beta for a long time without people notice that.
@vzn The magnetization is well known fact that undergrad student learns it.
It can be easily understand that every tiny magnet prefer to align will its neighbor, if the interaction is strong enough. Then they will permanently align with each other and no one can point to opposite direction.
The phase transition does relation to P/NP transition because the physics is the hardest to solve at the phase transition point.
BTW, what do you think about the D programming language in general usage?
kyles experiment might have translation symmetry...
still not exactly following the difference...
re bitcurrency based on "proof of work", its not so simple as "relying on 3rd party to give credit". it sounds like a cryptographic problem to me. note bitcoin is a solution to a cryptographic problem called "the byzantine generals"...
yes have studied P/NP phase transition quite a bit.
a P!=NP proof might have a physical meaning as far as showing there is a definite limit on the speed of demagnetization during cooling. (still sketching this idea out...)
havent used D yet.
my latest new "language(s)" are jquery/css/html5 etc :p ...
btw heres that johnny depp movie re AI taking over
@vzn See the picture of the grid there using triangulation en.wikipedia.org/wiki/…. It is grid, but not lattice
lattice is supposed to have very regular and periodic structure, like square lattice, trianglar lattice
@vzn I don't think they work at the moment, the current trust relying on the deterministic random number which is very easy to verify so it has the least known problem
@vzn I agree that NP problem should have physics meaning, but it still won't change physics. But it will make classification easier.
@vzn I think that a hard core computer scientist use functional programming such as lisp, but not these "fancy" language :)
I want to know, why is it that it is easier to cut through a string which is tightened then a loose one?
The question arose when I was watching - "What Happened Next?". In a part, they showed two clips, each one demonstrating the crash of rotor blades of a helicopter with a cable horizontally ha...
I was reading A Briefer History of Time by Stephen Hawking and Mlodinow. I found something silly. On page 36 at the bottom, it says the following :
If, say, the sun suddenly disappeared, Maxwell's theory tells us that the earth wouldn't get dark for about another eight minutes (since that is ...
I saw a TED Talk. You can watch it here on Youtube for your convenience.
At 07:17, he introduces a toy made with a pencil on which, a few notches are present; and on rubbing them with something, a fan attached to it rotates.(See transcript at TED if you are unable to watch it.)
And it is funny ...
I skipped a proposed edit on http://physics.stackexchange.com/a/107909/42975 because I was editing myself quite a lot.
Now, I want to manually apply the useful fixes from it, but can not find the skipped edit diff.
Where to look?
yeah it is not achieving all the scientific promise that was anticipated partly due to restrictions on it... it maybe supports some problems... mainly good for high thruput matrix algebra/computations afaik
You probably used a finite difference engine for that
It's similar to the finite volume method, except finite volumes take into account particle transmission of the information by computing cell averages via linear/parabolic interpolations
I skipped a proposed edit on http://physics.stackexchange.com/a/107909/42975 because I was editing myself quite a lot.
Now, I want to manually apply the useful fixes from it, but can not find the skipped edit diff.
Where to look?
