I think this is a very important point and just wanted to expand upon it a little.
There are two rather different notions of "quantum computers" out there, and I think they are too often conflated. The first notion is the specific model under which Peter Shor (and others) have worked: quantum computers are composed of gates that sequentially apply unitary operators to two or three particles of an entangled ensemble. The second notion of quantum computer just uses "quantum" as an adjective, and might mean any computer built using quantum principles -- which incidentally probably includes any computer built since the seventies.
Theoretical computer scientists are really excited about the possibility of the first kind of quantum computer. If such computers could be built, it would fundamentally change the way we deal with computers. The most basic notions of digital computers, namely MIPS and bytes, do not have straightforward analogs in the quantum world. Imagine the consequences to the computer programmer of "deprecating" these concepts.
To date, however, only the second kind of quantum computer has been built (ignoring, for the moment, the ten million dollar machines that declare victory upon factoring 15). D-Wave Systems seems to be doing more of the same.
To briefly summarize some issues about the Traveling Salesman Problem (TSP) mentioned in the D-Wave press release: the traveling salesman problem is what is known as "NP-complete", a class of problems believed to be much harder than factoring, and presumably well out of reach for any computer, quantum or digital. We can, of course, solve the problem in small cases by random search, often with the aid of a number of heuristics. One popular heuristic is called "simulated annealing": annealing is the process of cooling molten metal or glass in a precisely controlled manner so that the random motions of the atoms end up "solving" the minimum energy problem and ending in a very stable configuration; the idea in computer science is to let these same principles "anneal" an approximate solution to the traveling salesman problem; it is "simulated annealing" because a computer simulates the cooling process.
What D-Wave Systems seems to have done is decided that, rather than just simulate annealing, they'll actually build a device, and anneal it manually with a refrigerator (speaking in loose terms). OK -- reasonable idea, but not earth-shattering.
It probably will end up solving TSP much faster than regular computers, but then again, one would expect $18 million custom-designed hardware to do that, quantum or not. What it won't be is the general-purpose quantum computer of the Shor et. al. model.
Slashdot Mirror
I think this is a very important point and just wanted to expand upon it a little.
There are two rather different notions of "quantum computers" out there, and I think they are too often conflated. The first notion is the specific model under which Peter Shor (and others) have worked: quantum computers are composed of gates that sequentially apply unitary operators to two or three particles of an entangled ensemble. The second notion of quantum computer just uses "quantum" as an adjective, and might mean any computer built using quantum principles -- which incidentally probably includes any computer built since the seventies.
Theoretical computer scientists are really excited about the possibility of the first kind of quantum computer. If such computers could be built, it would fundamentally change the way we deal with computers. The most basic notions of digital computers, namely MIPS and bytes, do not have straightforward analogs in the quantum world. Imagine the consequences to the computer programmer of "deprecating" these concepts.
To date, however, only the second kind of quantum computer has been built (ignoring, for the moment, the ten million dollar machines that declare victory upon factoring 15). D-Wave Systems seems to be doing more of the same.
To briefly summarize some issues about the Traveling Salesman Problem (TSP) mentioned in the D-Wave press release: the traveling salesman problem is what is known as "NP-complete", a class of problems believed to be much harder than factoring, and presumably well out of reach for any computer, quantum or digital. We can, of course, solve the problem in small cases by random search, often with the aid of a number of heuristics. One popular heuristic is called "simulated annealing": annealing is the process of cooling molten metal or glass in a precisely controlled manner so that the random motions of the atoms end up "solving" the minimum energy problem and ending in a very stable configuration; the idea in computer science is to let these same principles "anneal" an approximate solution to the traveling salesman problem; it is "simulated annealing" because a computer simulates the cooling process.
What D-Wave Systems seems to have done is decided that, rather than just simulate annealing, they'll actually build a device, and anneal it manually with a refrigerator (speaking in loose terms). OK -- reasonable idea, but not earth-shattering.
It probably will end up solving TSP much faster than regular computers, but then again, one would expect $18 million custom-designed hardware to do that, quantum or not. What it won't be is the general-purpose quantum computer of the Shor et. al. model.