## HP, MIT announce joint project for quantum computing

Hewlett-Packard Company and the Massachusetts Institute of Technology (MIT) announced on 8 August 2001 the launch of **a joint effort aimed at building quantum information systems**. According to an HP press release, the project will receive $US 2.5 million in funding over a 4.5 year period. Researchers from HP Labs in Palo Alto and Bristol, U.K., will work with Professors Neil Gershenfeld and Issac L. Chuang the MIT Media Lab, as well as Dr. Seth Lloyd, an Associate Professor at MIT's Mechanical Engineering Department and a leading theorist of quantum computing. "With HP's success in molecular electronics research and MIT's expertise in quantum computing, we have a powerful one-two punch for the advancement of quantum information systems," said HP Fellow Stan Williams, director of quantum science research at HP Labs.

HP was granted another molecular electronics patent in July 2001.

August 20th, 2001 at 12:31 AM

Enterprise or Pork?This story might easily be mistaken as a sign of private investment in quantum computing by a reputable, well-established company. In reality, although I don't know for sure, I am willing to bet a ham sandwich that the venture is going to be funded by DARPA, ARDA or NSF. My first guess is DARPA, because they like to fund collaborations, and they like to get private firms involved. But it is all being fueled by government money. No private investor in his right mind would put money into quantum computing at this stage.

For the government, supporting QC research is a risk-free proposition, because the government needs to know whether this technology is going to prove realizable, no matter whether the answer is positive or negative. A private investor, on the other hand, will almost certainly lose any money put into QC R&D.

The only near-term (within 10 years, possibly within 5) prospect for an application of quantum information technology is so-called "quantum crytography" where entangled photon pairs can be used to create a "provably secure" communication link.

The "quantum" cachet, combined with the physicists' assurance that the comm link is absolutely secure, means that the technology may have some very limited market potential to governments, banks, or others with money to burn and a paranoid concern for security.

Anyone else can get just as high an assurance of security, at much lower cost, and without needing to install dedicated hardware between fixed locations, by using strong PGP-type encryption over the internet.

In the long term, it is possible that QC will become an important technology, and that it could be used to decrypt old RSA or PGP-encoded emails. The government needs to worry about that possibility. I don't know who else does.

August 22nd, 2001 at 7:56 PM

Re:Enterprise or Pork?Private enterprise in quantum computing: crazy?

Well there is at least one crazy company: MagiQ

Dabacon

August 22nd, 2001 at 10:04 PM

How *exactly* do they take the measurement?I've read several articles on quantum computers and qubits and I am still mystified on how the computer collapses the state vector to arrive at the answer.

Basically how, out of a potential infinitude of states, does it choose the correct answer?

For example, a qubit is a superposition of 0 and 1, right? A forty qubit quantum computer is a system with a superpostion of a mindwrenchingly huge number of states. This is how the computer calculates the answer so fast. That part I get.

But I've yet to see a good explanation as to how the computer takes the measurement and arrives at the right answer. Can someone here point me to a good primer, book or web, on how this is done?

August 23rd, 2001 at 4:05 PM

Re:How *exactly* do they take the measurement?Good intros on quantum computing are at:

Centre for Quantum Computation

dabacon

August 23rd, 2001 at 11:36 PM

Re:Enterprise or Pork?I would not advise investing in MagiQ. They are trying to buy up patents and support a small research enterprise, mainly monitoring. It's for investors with lots of money to lose.

As I mentioned, there is a possibility of quantum cryptography (secure comm links) to be developed in the not so distant future, but the market potential is very limited, unless an actual quantum computing technology emerges that would threaten the security of existing encryption schemes. There is no sign of that now, although it is a distinct possibility in the long term (decades).

QC, if developed, would be an important technology, not only for crypto. So it isn't entirely crazy for a billionaire to put a few k$ into a company like this, but for anyone who can't afford to lose money, this would be a very bad bet.

August 23rd, 2001 at 11:49 PM

Re:How *exactly* do they take the measurement?The measurement process is just like any other measurement. The "collapse" is automatic when a measurement is taken.

A good quantum algorithm will produce a final quantum state which contains only one or a few classical (e.g. binary) states. That way you only have to take a few measurements (repeating the entire calculation each time) to determine which states are occupied, and how much of each. The ideal case would be if the final state of the quantum computation contained only one classical state; then only one measurement would be necessary.

In order to take advantage of the power of QC, the computation has to involve intermediate quantum states which contain a substantial fraction of the classical state space, i.e. a "superposition of a mindwrenchingly huge number of [classical] states." However, the final steps before measurement can use interference to funnel the state occupancy down to one or a few classical states. It's like a hologram which focuses light to a point on a screen, even though the photons have taken a superposition of all possible paths to get there. The coherent interference of the different paths produces a focusing.

It is very hard to think of quantum algorithms which satisfy these requirements and still do something useful. In fact, to date, only one useful true quantum algorithm is known.

August 24th, 2001 at 4:08 PM

Re:Enterprise or Pork?I would not advise investing in MagiQ. They are trying to buy up patents and support a small research enterprise, mainly monitoring. It's for investors with lots of money to lose.It is good, then, that the company is private! Saves me from myself!

On a more serious note, I think your description of the future of quantum computing is a very sound position,

given our current state of knowledge about the field. But everytime I hear this pseudo-pesimistic take on quantum computing, I can't help but think about the early early days of the modern computer. Early computers were built by governments. The main applications of these devices was for cryptography. Of couse, such an "arguement by history" is pretty much no arguement at all. On the other hand, it feels to easy to think we are so sophisticated as to say that there aren't more interesting quantum algorithms out there.dabacon

August 25th, 2001 at 12:42 PM

Re:How *exactly* do they take the measurement?Mark,

The article cited by dabacon helped me quite a bit.

It explained the superpostion of quantum states in classical computer terms of numbers in registers–every qubit added exponentially increases the number of registers the algorithm operates on.

This removed the scales from my eyes.

The machine operates on all the numbers in all the registers at once in a massively parallel computation. The calculation runs and ends in a group of results. Only one result of all those calulations will be right.

Any problem that grows exponentially harder to solve as you add terms, non-polynomial problems, are tractable to a quantum computer because you simply tack on another qubit for each new term added and then run the calculation in polynomial time.

That clears up my confusion.

Sorry for not explaining my confusion in a clearer way and thanks for providing explanations to me!

August 31st, 2001 at 6:01 AM

Re:How *exactly* do they take the measurement?It's really more confusing than you think.

Actually, every qubit only adds a qubit, but the size of the

classicalregisters that would be needed tosimulatethe quantum register grows exponentially. This is not the same as saying the quantum register is equivalent to the huge classical register that would be needed to simulate it. The huge classical register would actually be much more useful if it could be made.You are referring to a peculiarity of the Shor factoring algorithm, which is still the only useful truly quantum algorithm known. As I said, ideally a quantum algorithm would give only a single answer, i.e. at the end of the calculation the quantum state of the register would be one particular answer with unit probability and all other answers would have zero probability. However, the Shor algorithm does not do this; there is some probability of a wrong answer, and you have to run the calculation a few times to see what answers you get and then check them to see which one is right. This is not necessarily a universal characteristic of all quantum algorithms, but it's hard to say because so far the Shor algorithm is just about all we have.

This was an early hope for QC but it is now known not to be true. Actually, there is only one algorithm known that can take advantage of the power of QC to solve what is believed to be an exponentially hard problem, and other exponentially hard problems are known not to be soluble in linear time by QC.

There is another very important exponentially hard problem which is known to be soluble by QC, although little work has been done on algorithms for it. That is the simulation of quantum systems themselves. In the long run, this may be the most important application of quantum computing, if it ever becomes practical.