You have to remember that these guys are chemists not computer scientists. Maybe there is a way of providing some authentication, but I can't think of one. The thing is how to tell if a returning piece of data has been processed in the correct way. If someone changed some of the parameters in the code then the results would not be right.
The multiple runs aren't identical runs they are different runs of the same problem. Basically what they are trying to do is search the multidimensional hypersurface that belongs to the protein in question to find how the protien folds, which energy minima it visits and which minimum it eventually ends up in. They start in an unfolded state and use molecular dynamics (MD) or Monte Carlo (MC) (I can't remember which) to look for minima. Instead of having one long simulation run they send out lots of different runs going in parallel on different computers in the project. When a new minimum is found all of the runs stop what they are doing and jump into that minimum. Due to some complex mathematics and science (the ergodic principle I think) these shorter runs add together in this way to produce results the same as one long run. Because all of the runs are different they cannot be checked against each other.
If there was a way to allow some sort of authentication and have the source code available that would great. After all the tinker code is available and the distributed computing code is available, so they have no reason to hide the code.
What OS are you running? If your running on unix or linux create a "folding_at_home" user with their own group and run it as them so that it can't do any harm. Just a thought.
I think that the licence is more to do with protecting their work rather than denying you rights. If you had the source code then they would have no control over the experiment, which would make it worthless and unlike seti@home the work that your computer does has an effect on the work that is done by all of the other computers in the project.(Because of the way they are using multiple short simulation runs to simulate one long run)
Any scientist who didn't control their experimental conditions, and tried to pass off there work as valid, wouldn't be a very good scientist would they? They would just have wasted everyones generously given cycles for nothing.
Slashdot Mirror
by releasing open source developers toolkits that exploit new features they add to their processors
Intel have already released Linux compilers for its chips (including the Itanium).
Tom.
You have to remember that these guys are chemists not computer scientists. Maybe there is a way of providing some authentication, but I can't think of one. The thing is how to tell if a returning piece of data has been processed in the correct way. If someone changed some of the parameters in the code then the results would not be right.
The multiple runs aren't identical runs they are different runs of the same problem. Basically what they are trying to do is search the multidimensional hypersurface that belongs to the protein in question to find how the protien folds, which energy minima it visits and which minimum it eventually ends up in. They start in an unfolded state and use molecular dynamics (MD) or Monte Carlo (MC) (I can't remember which) to look for minima. Instead of having one long simulation run they send out lots of different runs going in parallel on different computers in the project. When a new minimum is found all of the runs stop what they are doing and jump into that minimum. Due to some complex mathematics and science (the ergodic principle I think) these shorter runs add together in this way to produce results the same as one long run. Because all of the runs are different they cannot be checked against each other.
If there was a way to allow some sort of authentication and have the source code available that would great. After all the tinker code is available and the distributed computing code is available, so they have no reason to hide the code.
What OS are you running? If your running on unix or linux create a "folding_at_home" user with their own group and run it as them so that it can't do any harm. Just a thought.
Tom.
I think that the licence is more to do with protecting their work rather than denying you rights. If you had the source code then they would have no control over the experiment, which would make it worthless and unlike seti@home the work that your computer does has an effect on the work that is done by all of the other computers in the project.(Because of the way they are using multiple short simulation runs to simulate one long run)
Any scientist who didn't control their experimental conditions, and tried to pass off there work as valid, wouldn't be a very good scientist would they? They would just have wasted everyones generously given cycles for nothing.
Tom.
Plan 9 is not a new project and it isn't even *nix. There is a faq that tells all about it. Tom.