Quest for Quantum Computers Heats Up

http://www.scientificamerican.com/article/quest-for-quantum-computers-heats-up/

by jonbaer • 12 years ago

48 5 12 years ago

Xcelerate• 12 years ago

This is very exciting to me! It's a great time to be interested in quantum mechanics.

Quantum computers would pave the way for the holy grail of computational physics and materials science: perfect QM simulations. Imagine performing a simulation on a computer that matches up exactly* with experiment. Now that would be exciting. You could actually develop new materials right from the comfort of your chair at Starbucks (SSHing into a supercomputer of course). Your simulations would even be more accurate than a lot of experiments because of the absence of environmental and experimental errors.

Recalling history, it's amazing to me how quickly the idea of a nuclear weapon went from conception to implementation. Also, it only took like 9 years to get a person to the moon. You basically had a bunch of the world's greatest scientists all gathered in one location with the motivation to get the task done. I wonder what would happen if a similar situation was applied to quantum physics research.

* Leaving out plenty of details here. Quantum computers can tackle the NP hard fermion sign problem, which is the current main obstacle in high accuracy QM simulations, but there are still some lingering questions about whether the process of inputting a particular system into a quantum computer is itself an NP hard task. Not to mention that some heavier elements require relativistic treatment, the fact that I'm talking about quantum chemistry rather than particle physics, and the fact that even with a polynomial time algorithm, quantum computers may still be too slow to simulate a large system in a practical amount of time. But I can dream.

tfgg• 12 years ago

Their application to quantum chemistry/condensed matter is certainly pretty exciting. Currently the practical method for realistic system sizes, density-function theory (DFT), is barely quantitatively accurate for many important systems/properties. I've heard it suggested that current DFT codes could use a quantum computer as a co-processor to handle strongly correlated parts of systems, which is an interesting idea. Just speeding up post-Hartree Fock methods would be nice, and they're already polynomial but impractical!

Do you have any references for the hardness of putting the quantum system into the quantum computer? I haven't heard of that issue before. I wouldn't have thought relativity is much of a problem, that's just a tweak to the Hamiltonian.

mathgenius• 12 years ago

This is still very much work in progress: http://arxiv.org/abs/1406.4920

mturmon• 12 years ago

People have been searching for the applications of quantum computers -- excluding the adiabatic QC's associated with D-Wave, which are somewhat in a different class. I found this part of the article very interesting regarding applications:

"[Mathias] Troyer has spent the past three years looking for an answer—a 'killer app' for quantum computing that would make the development costs worthwhile. The two classic examples, code-cracking and searching databases, are not good enough, says Troyer. Shor's algorithm will require thousands of qubits to do any serious factorization, he says, and there are other forms of encryption that a quantum computer would do nothing to solve. And although quantum computers may search databases faster, they are still limited by the time it takes to feed the data into the circuit, which would not change.

"Troyer thinks that a much more fruitful application for the near future is the modelling of electrons in materials and molecules—something that quickly becomes too difficult for today's supercomputers. At first, this, too, seemed a long shot. His early estimates suggested that it would take a quantum computer as long as 300 years to simulate the molecular dynamics of even a small molecule—such as the iron sulphide inside the ferredoxin proteins that are involved in nitrogen fixation in plants. “Clearly, that was on the border of being science fiction,” he says. But by rewriting the software, he brought the figure down to 30 years—then to just 300 seconds. “Just like in classical computing, where one has to sit down and optimize the algorithm,” he says, “the same is needed for a quantum algorithm.”

buhrmi• 12 years ago

"A 30-year slog to develop a useful quantum computer may finally be on the verge of paying dividends"

Hahaha... Define "useful"

Quest for Quantum Computers Heats Up

Related Stories