Back of the Envelope

Old Post: An introduction to quantum computation is here, while an explanation of a qubit is here.

I was going to comment on President Bush's Bioethics Council, but then I thought I should start closer to home.

I have a Ph.D. in Electrical Engineering, which generally means that I am either a professor or a researcher. In my case, I am a researcher who runs experiments and analyzes the data. My area of research is quantum computation, as you may have figured out from other things I've said around here. Quantum computation has its own ethical dilemma. To date, we've discovered that it's very good at two useful applications, performing unordered searches and factoring large numbers. The first may be useful, while the second is definitely useful. It's much easier to find a prime number and multiply it by another large prime number than to factor a large product of primes. When I say much easier, I'm talking about it taking the same computer a few seconds to do the finding and multiplying, versus a few million years to do the factoring. This sort of one-way problem forms the basis for public key encryption (although it is of course more complicated than that), such as that used in RSA, the encryption protocol used to transmit information on the Internet. For more information on RSA, check this FAQ from the sci.crypt newsgroup. A quantum computer with a sufficient number of qubits could factor a large product of primes faster than a classical computer create it in the first place. If someone were to produce such a computer today, all Internet transactions would suddenly be vulnerable.

You can tell what use people are planning for a quantum computer by looking at where the funding is coming from. Right now, the people giving out the funds are the Army Research Office, the Defense Department, and, oh yes, the NSA. It's clear that the main objective is decryption (or, perhaps, to prove that a quantum computer is so far from realizable that public key encryption is secure).

In my experience, very few scientists working on quantum computation projects think about the ethical implications at all. For the most part, they console themselves with the fact that a quantum computer capable of factoring a decent sized key is so far in the future that by the time it gets here (~25 years or so), we'll have better encryption (hopefully, quantum encryption). That may or may not be the case. I've heard that the federal government may be pushing for a five-year program to develop a quantum computer that can factor 128-bit encryption. (I've been looking for confirmation but I haven't found it yet.) This is wildly ambitious--I don't think it will happen--but how many scientists, who previously considered quantum computation safe because it was decades away, would jump at the chance to partake of this funding?

For the record, I have thought a bit more in-depth about quantum computation, partially because I took part in a discussion group with MIT's Graduate Christian Fellowship based on the book Responsible Technology. To a large degree, the questions about developing a quantum computer revolve around who would get it. Quantum computers aren't going to be available on the open market anytime soon, and they'll probably be as regulated as nuclear power. So, assuming I helped create a quantum computer, would I trust the NSA to use it wisely? I certainly don't mind them cracking a terrorist's e-mail, but I wouldn't want them reading mine. So, I don't think that the technology itself is wrong, but I am concerned over how it would be used.

New Post: I explain quantum cryptography above.

Old Post: I first discussed quantum computation in this blog here.

I'd like to take a moment to expand on the subject of quantum computation, and that means starting with the term "qubit." A qubit is a quantum bit, and it refers to any quantum system that has two states which can serve as zero and one. Examples include spin states of nuclei in a molecule, an electron's orbital states in an atom, photon polarization, or, in a system such as superconductors which exhibit macroscopic quantum coherence, current circulating in a loop. If it's quantum, it's been proposed as a qubit.

Not every quantum system makes a good qubit, however. The criteria necessary for a quantum system which can be used in a quantum computer were formalized by DiVincenzo. The requirements are the following:

First the two states, called the |0> and |1> states, must be measurable. It must be possible to tell the difference between them. This may seem trivial, but quite a few quantum states are difficult to differentiate.

Second, the states must be controllable. This means that one can first place the system in the |0> state accurately. Then, one must be able to rotate the qubit in order to achieve every possible state. The possible states are not just |0> and |1>. If a and b are complex numbers, such that |a|^2+|b|^2=1, a|0>+b|1> describes all the possible states of the qubit. Thus it is possible for the qubit to be in state |0> and |1> at the same time (called a superposition), as long as the proportions add up to 1. Since a and b are complex, it's not simply a matter of achieving the right proportions, but also the correct phase--the correct complex values.

Third, the qubit must be addressable. One needs to be able to decide which qubit to control and measure. If there's a solution filled with millions of identical molecules, and there's a way to rotate the oxygen atom nucleus in all the molecules at the same time, that's still only one useful qubit (more precisely, it's an ensemble of that qubit). Now if it's possible to address the two carbon atom nuclei and the oxygen atom nucleus on each molecule separately, that's three qubits, and an ensemble of those three qubits. This is what is done in nuclear magnetic resonance (NMR) quantum computing.

Fourth, one needs to be able to couple qubits together so that they affect one another. Thus one qubit will rotate only if another qubit is in a certain state. This is how one makes quantum gates.

Finally, one needs to be able to isolate the qubits from the environment. The environment, which means everything other than the qubits themselves, causes the qubits to decohere. Information is lost as the qubits dephase, drift from the intended phase, and relax, fall to the lowest energy states. If one waits long enough, qubits in just about any system eventually fall to |0>. Only if they remain in the desired states long enough to perform a useful calculation can you make a quantum computer.

And that's the second episode of quantum computation for the layman. Tune in next time as I tell you how you can build your very own quantum computer (with a $1 million grant and a Ph.D., of course).

New Post: More on quantum computation above. First, ethical considerations, then quantum cryptography.