News and Views

Comments about the article in Nature: Quantum Computing

Following is a discussion about this "NEWS & VIEWS" article in Nature Vol 463 28 January 2010 by Emanuel Knill Nist Boulder
In the last paragraph I explain my own opinion.

Quantum computing

The article starts with the following text:
The race is on to build a computer that exploits quantum mechanics. Such a machine could solve problems in physics, mechanics and cryptography that were once thought intractable, revolutionizing information technology etc. But when
IMO this introduction is much to optimistic.
Next we read the following question:
What is a quantum computer?
The simplests answer is: A computer that exploits quantum mechanics. Of course this answer is much too simple. You have to explain how it works. The article does not give this answer. To claim that it uses the concept of superposition is also too simple. Next we get the following question:
What are qubits made of?
One answer is: trapped atoms or ions. That being the case you should immediate ask: And how do you manipulate those trapped atoms and how do you prove that they are in a superposition state.

Next we read:

How is quantum information manipulated?
The answer is: by applying quantum gates . That is logical correct, but this does not answer the question: How do you that by using trapped atoms and ions. Physical that is a very difficult problem.

Are Quantum computers faster than classical computers?
Theoretical the answer is Yes, but you first have to build one with a reasonable sized application to demonstrate this.

When was the potential of quantum computing first recognized?
The reader should be aware that almost evrything related to Quantum Computers is theoretical.

What kick-started the field ? It was a remarably efficient quantum algoritm designed by Peter Shor in 1994 for factoring large numbers.
Again this is almost all theory. For a critical evaluation see: Shor's Algorithm

What else are quantum Computers good for? There are mathematical problems for which quantum algorithms can yield solutions drastically faster than classical qomputers.
Again all theory.

Is that all? That is a lot! etc I have no doubt that there are applications that we cannot imagine
The comment becomes repetitif: all theory.

What do we require to build a quantum computer?
This question implies that until now we did not make much progress.

Next we read:

There are notable examples of qubits, such as photon qubits, that are easy to entangle but hard to usefully quantum compute with
This answer shows that photons can not be used in Quantum Computers. I'am not amazed. I expect that it is very difficult to make logic using single photon's
Does the anaolog nature of configuration amplitudes cause problems?

Will quantum error correction work?
A certain aspect of Shor's algorithm is based on probabilities. That means there is a chance that during one run you do not get the correct answer. If there is a chance on errors and if you need error correcting and detection logic, which in turn can also cause errors than you have a serious issue.

What is the required accuracy of physical qubit manipulations?
Ordinary Digital Computers do not have this problem. In a DC it is 100%

Have sufficiently accurate quantum gates been demonstrated? No
A honnest reply, which gives a good idea about the current state of art

What can be done in the absence of such accurate gates?
Shor's algorithm was invented in 1994. As such there is not much progress.

How many qubits can be used for quantum computing today? Answer eight.
I think the first Digital Computer in 1940? was much more powerfull.

What are the leading technologies for quantum computing? Large arrays of trapped atoms or ions may soon be used for specially devised quantum simulations.
When you read the full text, than all of it is not very impressive. Read also next correspondance.

Which will be the best in the long term?
Also here there is not much meat.

When will quantum computers outperform classical computers?I am optimistic that I (Knill) will be able to perform interesting computations etc
I expect never.


Correspondance

Comments about the correspondance: Could a boom in technologies trap Feynman's Simulator

By Dietrich Liebfried Nist Boulder, in Nature 4 Februari 2010
I believe that quantum simulation is slated for a breakthrough by 2020.
Also like the previous article much too optimistic. Both writers are from National Institute of Standards and technology.

In a groundbreaking lecture Richard Feynman pointed out that large quantum systems are impossible to simulate with a classical (digital) computer.
I completly agree with Richard Feynman. Both use completely different physical concepts. The only thing that you can do is to try to solve the same problem on both a classical computer and a quantum computer.

He proposed that they might instead be simulated with a quantum mechanical machine - in short a quantum simulator.
IMO that is not possible. The problem is that each quantum mechanical problem (For example shor's algotithm) requires its own quantum computer (its own quantum system) to solve a problem in real time. For each new problem you have to rebuild your quantum computer. To use a simulator, which is also a quantum computer, only adds problems.

Some promising technologies have emerged, but verifying a quantum simulation is not straightforward.
At least the writer identifies here a problem. In fact what he writes that verifying a quantum computer is not straightforward. That is a serious problem. I am not amazed.

Do the results demonstrate properties of the simulated model or are they due to unrelated features of the simulator.
If you are solving a problem it should not matter inprinciple if you use a simulator QC or you own QC. You should know what the answer is for test cases. If your simulator QC does not give that answer than you have a problem. However most probably the answer has nothing to do with the fact that you use a simulator.

But if the same physical models are simulated on different quantum simulators based on different technologies it is quite likely that common features of all the results will be due to the quantum physical model and not to the systematics of the simulators.
First of all you have to demonstrate that each simulator (i.e. simulator QC) works i.e. that it can solve reasonable test cases. I already expect serious problems with those simulators as with each QC.
Scientists and funders should therfore support development of as many technologies as possible.
If I was a funder than I would ask the following question: which are the practical problems that you can presently solve and what was the answer 5 years ago. My expectation is that both lists are not very large, showing almost no progress. Conclusion: If I was a funder I should spend my money somewhere else.

Reflection

If both articles are a good presentation of the present state of art in Quantum Computing than the future is gloomy.


Feedback

None


Created: 13 February 2010

Back to my home page Index