Please fucking learn the difference between "it's" & "its" you fuckwit. And to Taco: either edit the bloody thing, or at least let us know that you know the difference: sic.
"Face it, your individual viewing data is WORTHLESS. You're just not that important of a person. Viewing data is only worthwhile as an aggregate, despite what everyone seems to think."
Bullshit. The information is MUCH more valuable if it can be linked to a demographic profile. Advertisers want to know the habits demographic groups. Aggregated anonymous data is useful, but an ability to link each slice of data to a specific income/ethnic/whatever group is VERY useful.
This one caught my eye because I actually worked on the clique detection/maximal common subgraph & finding subgraph isomorphisms for my Ph.D. It turns out these NP-complete problems are directly applicable to the searching of chemical databases. The basic problem is this: given a query molecule A and a database molecule B, does B completely contain A? This is the problem of subgraph isomorphism. Finding a subgraph in the case of a single pair of molecules is tough, especially if the molecules are large. Now imagine running a query aginst a large database of patented molcules (CAS online- tens of millions of structures). The stakes are high since patents on pharm molcules can be worth billions; the research to find them can cost similar amounts of money and take years.
Now to take a step back. If a method was found to solve ANY NP-complete problem in polynomial time it would mean NP=P since the NP-complete problems form an equivalence class. NP=P? is one of the foremost unsolved questions of mathematics. My guess is someone has found a trick to solve certain NP problems quickly under specific conditions. This is nothing new, in fact, in my case, I showed that a massively parallel computer (Distributed Array Processor 4,096 1 bit processors with a taurus geometry) can be used to solve subgraph isomorphism quickly for specific cases.
Re: The Mean Value Theorem. This is probably urban folklore, but I'll tell it anyway. I heard of a traffic court case in which a University Math Professor was clocked speeding. The cop used the old fashioned method- measure the time to cover a known distance to calculate the average speed. The Professor argued that although his average speed may have been above the speed limit, this does not necessarily mean he had attained that speed. Of course he knew the the MVT could be used to prove this so, however, he also knew that the cop and judge had no idea what the MVT was, or how to apply it. In true urban legend fashion the prof escaped the ticket.
duh, this is exactly what Godel proved- there exist theorems for which it can be proved that a solution can never be found.
An example is the problem of tiling the plane.
This story is a complete red herring. Anything that can be expressed in digital form- algorithms, movies, music, books, photographs can be encoded as a natural number. This is the basis on which a Turing machine operates (the theoretical model on which all present day computers operate, with the exception of quantum computers). The assumption behind the ecoding of DeCSS to a prime seems to be that it is silly to copyright numbers. If this assumption is carried to its logical end, then it should be wrong to copyright anything digital (no doubt some of you agree with this).
The bottom line is this stunt is just a cute bit of number theory, it carries no logical weight whatsoever.
funkbrain is misleading when he says Newton's theory is "incorrect". The test of a theory is the extent to which its predictions match experimental data. Newton's theory is not incorrect, rather it is not as accurate as the theory of General Relativity. Indeed, it is usually impossible to prove that a theory is 100% correct since it is impossible to test the infinite number of conditions it predicts. If Newton's theory of gravitation is "incorrect", why do engineers, astronomers, astro-physicists use it every day? Newton works well except in very special circumstances, in these cases GR is more accurate.
What's your point? The TSP is intractable not unsolved, there's a big difference. If you could come up with a polynmial time algorithm for TSP this would be a remarkable breakthrough since TSP is NP-Complete and all such problems form an equivalence class. Showing that TSP is a member of P would prove that P=NP. Such a proof would have many real-world repurcussions.
Classic. I just love it when people say "I can't see the need for x amount of resource y. Who could use this?" (where y is CPU speed/mass storage/bandwidth).
When will people learn?
Slashdot Mirror
Please fucking learn the difference between "it's" & "its" you fuckwit. And to Taco: either edit the bloody thing, or at least let us know that you know the difference: sic.
"Face it, your individual viewing data is WORTHLESS. You're just not that important of a person. Viewing data is only worthwhile as an aggregate, despite what everyone seems to think." Bullshit. The information is MUCH more valuable if it can be linked to a demographic profile. Advertisers want to know the habits demographic groups. Aggregated anonymous data is useful, but an ability to link each slice of data to a specific income/ethnic/whatever group is VERY useful.
This one caught my eye because I actually worked on the clique detection/maximal common subgraph & finding subgraph isomorphisms for my Ph.D. It turns out these NP-complete problems are directly applicable to the searching of chemical databases. The basic problem is this: given a query molecule A and a database molecule B, does B completely contain A? This is the problem of subgraph isomorphism. Finding a subgraph in the case of a single pair of molecules is tough, especially if the molecules are large. Now imagine running a query aginst a large database of patented molcules (CAS online- tens of millions of structures). The stakes are high since patents on pharm molcules can be worth billions; the research to find them can cost similar amounts of money and take years. Now to take a step back. If a method was found to solve ANY NP-complete problem in polynomial time it would mean NP=P since the NP-complete problems form an equivalence class. NP=P? is one of the foremost unsolved questions of mathematics. My guess is someone has found a trick to solve certain NP problems quickly under specific conditions. This is nothing new, in fact, in my case, I showed that a massively parallel computer (Distributed Array Processor 4,096 1 bit processors with a taurus geometry) can be used to solve subgraph isomorphism quickly for specific cases.
Re: The Mean Value Theorem. This is probably urban folklore, but I'll tell it anyway. I heard of a traffic court case in which a University Math Professor was clocked speeding. The cop used the old fashioned method- measure the time to cover a known distance to calculate the average speed. The Professor argued that although his average speed may have been above the speed limit, this does not necessarily mean he had attained that speed. Of course he knew the the MVT could be used to prove this so, however, he also knew that the cop and judge had no idea what the MVT was, or how to apply it. In true urban legend fashion the prof escaped the ticket.
If anyone is interested in further reading, I would highly recommend The Emperor's New Mind : Concerning Computers, Minds, and the Laws of Physics by Roger Penrose. However, be prepared to devote some serious time to this book, I have spent days pondering single paragraphs.
duh, this is exactly what Godel proved- there exist theorems for which it can be proved that a solution can never be found. An example is the problem of tiling the plane.
This story is a complete red herring. Anything that can be expressed in digital form- algorithms, movies, music, books, photographs can be encoded as a natural number. This is the basis on which a Turing machine operates (the theoretical model on which all present day computers operate, with the exception of quantum computers). The assumption behind the ecoding of DeCSS to a prime seems to be that it is silly to copyright numbers. If this assumption is carried to its logical end, then it should be wrong to copyright anything digital (no doubt some of you agree with this). The bottom line is this stunt is just a cute bit of number theory, it carries no logical weight whatsoever.
funkbrain is misleading when he says Newton's theory is "incorrect". The test of a theory is the extent to which its predictions match experimental data. Newton's theory is not incorrect, rather it is not as accurate as the theory of General Relativity. Indeed, it is usually impossible to prove that a theory is 100% correct since it is impossible to test the infinite number of conditions it predicts. If Newton's theory of gravitation is "incorrect", why do engineers, astronomers, astro-physicists use it every day? Newton works well except in very special circumstances, in these cases GR is more accurate.
That's 400 SQUARE MILLIMETRES which is less than a chip measuring 1 inch x 1 inch.
What's your point? The TSP is intractable not unsolved, there's a big difference. If you could come up with a polynmial time algorithm for TSP this would be a remarkable breakthrough since TSP is NP-Complete and all such problems form an equivalence class. Showing that TSP is a member of P would prove that P=NP. Such a proof would have many real-world repurcussions.
Classic. I just love it when people say "I can't see the need for x amount of resource y. Who could use this?" (where y is CPU speed/mass storage/bandwidth). When will people learn?