Slashdot Mirror
German Scientists Create 5 qubit Quantum Register
CMan0 writes "In the University of Bonn, a team of scientists has built a 5 qubit register, using cesium atoms trapped by a laser-beam grid, The Register reports. They've been able to install an empty 5 bit register(i.e. all bits 0), change two of them to 1, and later read those 1s back. The next goal is to create an interaction between 2 bits. The full scientific article can be found here in PDF format."
Actually, there was uncertainty.
;)
:-(
Upon further observation, it was known to have a probability of 1
On a serious note, this is awesome. With a 5 qubit entanglement and this, we might be able to build a primitive functional Quantum Computer, for the first time.
The team is now working to create a quantum gate in which two or more qubits of the register will interact in a controlled way.
Amazing. The beginnings of a first QC. We've memory, redundancy, processing capabilities and a lot more.
Now the only problem that remains is a suitable and reliable means of error correction - which is the biggest problem thus far in QC
Very small scale.
But it's not the scale that matters, it's the fact that it has been done. The problem with any QC related operation is the inherent difficulty -- in terms of having redundancy, storage, observation and retrieval.
That's why you keep hearing these things about quantum entanglement for 5 qubits and registers and the like. It's not the scale, it's the fact that people have been able to do them.
The problem is that a lot of things are THEORETICALLY possible in QC, but have not been practically achievable. Often times, people find that although it is theoretically possible, it's not realizeable due to some problem or the other. And then, further studies would prove that there are variables that people hadn't considered.
So, this would mean that we can store observed states -- in some way that can be copied and retrieved -- which is a big leap.
For example, one qbit setup is to use a helium superfluid, which naturally bonds electrons to the surface. The bound electrons can then be controled with a combination of microwave radiation and an electric potential from wafer posts under the fluid. Each electron (qbit) sits on top of a post, which are spaced just a few nm apart. The system is still being developed, but the nice thing is once they get it to work, they can just build a large wafer holding millions of qbits.
However, the huge problem with the above example is that it needs to run at about 50 mK, which is very close to absolute zero and requires a dilution fridge, which is a 6 foot tall cylinder. There are similar (though more complicated) limitations to the laser trapping methods.
For a commercial unit I suppose the QC wafer, microwave source, and dilution fridge could be packaged together nicely, but it is still 6 feet tall, heavy, not well suited for a house. Even if it were possible to make one small enough, there are currently no real benefits for a home user unless they really wanted to find elements in a large array or crack PGP codes... I suppose the first computers were also only suited for a lab environment and scientists probably thought the average person would never need a computer either, so who knows what will develop in the next 50 years...
"Qomputing" (qubit computing, get it?) is pursued independently across the globe, with separate teams reporting breakthrus in different pieces of the puzzle. One team has produced quantum entanglement, using "spooky action at a distance" to offer apparently instant communication between terminals. If each of these components in its distant lab were entangled in a quantum net, we'd get a qomputer built from the start to network in parallel while computing literally in parallel. Linux's unix heritage shows the compelling momentum derived from including networking from the beginning of the platform. Qomputing is born in the age of the network: entangled networks are natural midwives and gossips for a new qomputing qommunity.
--
make install -not war
Disclaimer: I'm a graduate student doing research on quantum computing in optical lattices. I'm not affiliated with the group that published this article.
This result is quite exciting, because it demonstrates the feasibility of some of the techniques necessary for an optical lattice-based quantum computer. Imagine taking their 1-D lattice and turning it into a 3-D lattice, with about 30 atoms in each direction. That's a whole lot of qubits...
So what are the next steps?
1) A new means of addressing atoms (selecting one or two atoms on which to perform operations while excluding the rest) is necessary. Their magnetic gradient technique works fine for a small 1-D lattice, but it would likely be impractical for a large 3-D lattice (Maxwell's equation div B = 0 gives one major obstacle, which would require fancy tricks to overcome).
2) One and two-qubit gates need to be demonstrated using an appropriate addressing scheme.
3) Error correction, which would likely require quantum non-demolition measurements to check to see if an atom had been lost from a lattice site. Translation: we need to be able to measure if we've lost an atom from a lattice site, without disturbing the atom's state (i.e. measuring whether it's |0> or |1>).
4) Full-blown fault-tolerant computation.
My group plans to solve (1) using an addressable optical lattice. What that means is that the lattice spacing is sufficiently large that a laser can be focused on an individual atom (in 3-D, two lasers in orthogonal directions would be used). I'm currently doing simulations of one-qubit gates in this scheme.
That brings me to my question for slashdot: Some of the simulations I'll be doing (specifically, studying decoherence in the one and two qubit gates) will be very computationally intensive. They're also embarrassingly parallel, as they're essentially quantum Monte Carlo simulations. Would people be interested in a BOINC-based distributed computing project (a la SETI@home) to help develop quantum computers? If so, what kinds of things would help you get involved? Would you be interested in helping develop the software (it's C++)?
I probably won't be at that stage for another six months to a year, but it would be helpful to me to start planning now. I have just (last night) completed the core simulation engine, and would need to add code for decoherence, as well as adapt it to BOINC. The code will be GPL'd, of course.