D-Wave Sells First Quantum Computer

[wuliheron]

http://physicsworld.com/blog/2011/05/d-wave_sells_its_first_quantum.html

Some have suggested D-Wave's 128 qubit computer is nothing more then a scam, but if so it appears to be one the US government supports! Whatever the case may be progress is being made with quantum computing and the implications are mind boggling. Such a system could theoretically provide profound new insights into quantum mechanics, rapidly model countless processes, and even break every conventional encryption code used worldwide. Which leads me to wonder if we are now facing the next technological revolution and Manhattan Project.

 

[pw38]

"D-Wave Sells First Quantum Computer"

Kind of.

 

[William Gaatjes]

I find it interesting. Not easy to grasp. If i understand correctly, it seems to me this quantum computer is operated in an analog manner. The individual (qu)bits are not 1 or 0. Before the calculation takes place, these bits are set up at a reference point. I assume half 1 and half 0. I think this is done by setting up a magnetic bias field. Then when the calculation starts by applying input, each bit is set to the resulting value. And since these bits are affecting each other based on position in the lattice(as all atoms always do in a lattice), they affect the outcome in a predictable way, only much faster then conventional electronic chips. What i do not understand is if here single electrons are used or atoms of some kind of material.

It seems to me this computer functions is similar as the very first sold home computers where you had to manually enter the program into it. Only now it is done by a "conventional computer" which also does the calibration set up. I should also add that the entire processing unit seems to be cooled to almost 0K to function and heavily shielded. Otherwise any other EM disturbance would influence the individual qubits more then the data would. Producing wrong results. Amazing engineering.

If i am wrong please correct me.

 

[Mark R]

The D-Wave computer is said to be an 'adiabatic quantum computer'.

This is a device for solving 'optimisation' problems; you have a problem which takes a bunch of parameters which interact - but what combination of parameters gives the optimal result (lowest error)?

E.g. You have a device that takes a video image and tries to classify the type of image by combining the pixel values according to a series of equations (e.g. output = P1 * Q1 + P2 * Q2...) where P are the pixel values, and Q are model parameters. Some variations of this problem need to be solved by a combination of trial-and-error and fine tuning - but even so, they may not find the best answer. If the starting condition is close to a reasonable solution (but not the best solution), the fine-tuning process may home in on the nearest solution, ignoring the totally different, but better solution.

As it is, the trivial linear algebraic problem above can be easily solved with conventional algebraic techniques (e.g. matrix algebra). However, when the problem contains non-linear terms then, often, only the trial-and-error approach is an option - and when you have 128 different variables to tweak, you are highly unlikely to find the optimal combination of parameters within a sensible period of time.

This quantum computer is operated by representing each parameter as a qubit, and setting it to 'both values'. You then add connections between the qubits, using superconducting connections/'transistors' (actually Josephson junctions), etc. to represent the structure of your mathematical problem. You then expose the quantum computer to your raw data (raw input data, and intended output), and then gradually manipulate the qubits to reduce their 'uncertainty'. Once you set the quantum uncertainty to 'zero', the qubits are representations of the optimal parameters in your model. The idea is that the combination of entagled qubits simultaneously represent all possible solutions to your model, and by then gradually removing uncertainty, you all the system to home in on the optimal solution. Unlike the conventional computational solution, the system can 'see' all possible solutions at the outset, so there is no risk of homing in on an alternative, but less good solution.

 

[William Gaatjes]

Interesting.
I have the strong feeling that simultaneously having all possible combinations is a marketing phrase. It sounds to me as this variable can have all values as for example between 0 and 100^100 (it can just never have them all at once). But if i do not know what the previous value was, i could state that the variable was all possible values. This however is an error or a limitation of the observer and not an intrinsic property of the variable.

 

[Mark R]

Interesting.
I have the strong feeling that simultaneously having all possible combinations is a marketing phrase. It sounds to me as this variable can have all values as for example between 0 and 100^100 (it can just never have them all at once). But if i do not know what the previous value was, i could state that the variable was all possible values. This however is an error or a limitation of the observer and not an intrinsic property of the variable.

No. Quantum states can be in a 'superposition' of multiple possible states - in other words, they can be in multiple states simultaneously. This is a fundamental, intrinsic property of quantum states, and is not a limitation of the observer, although observing it would have the property of altering the state. Quantum computing relies on being able manipulate quantum states, and then having quantum mechanical processes find the lowest energy state.

It is this intrinsic property of quantum states that has made a practical quantum computer such a 'holy grail' of computer technology, because it means that the system can simultaneously evaluate every possible solution to a problem and find the optimal solution.

 

[William Gaatjes]

No. Quantum states can be in a 'superposition' of multiple possible states - in other words, they can be in multiple states simultaneously. This is a fundamental, intrinsic property of quantum states, and is not a limitation of the observer, although observing it would have the property of altering the state. Quantum computing relies on being able manipulate quantum states, and then having quantum mechanical processes find the lowest energy state.

It is this intrinsic property of quantum states that has made a practical quantum computer such a 'holy grail' of computer technology, because it means that the system can simultaneously evaluate every possible solution to a problem and find the optimal solution.

I know this description very well, it is just not satisfying. It reminds me to much of Werner Heisenberg uncertainty principle.

 

[William Gaatjes]

I have always seen entanglement as the following way :
It also adds a little relativity into the mix.

Take 2 tuning-forks. These can be seen as 2 atoms or as 2 photons. All EM energy is removed , it is a sound proof room. The laser pulse or photon is replaced by a sound wave made up from a sine wave from a speaker which also acts as a microphone to pick up the change. The speaker is rotating around to act as an lesly-speaker while the 2 tuning forks are rotated around as well but at a different rotation frequency.

When you want to observe the 2 tuning-forks, you have to transmit a tone by use of an single sine wave from a speaker and then use that speaker as a microphone. Since the 2 tuning forks are always spaced apart, the peaks of the sine wave from the speaker will hit the 2 tuning forks always at a different time. But that is not all, the lesly speaker is rotating around the setup as well.
But you are not only interested in the frequency returned in hope to determine the F0 or resonating frequency of each individual tuning-fork, you also want the phase delay relative to your own oscillator of each tuning-fork. The oscillator that produced the signal for the speaker. No matter how you try, you can never get the the same phase delay with your setup. The problem is that the sound you receive is always a mixture of sound produced by the 2 tuning-forks. And if you where only trying to determine the phase from 1 tuning-fork, you notice you still get every time a different result.
And of course, you cannot see the setup, thus you have no idea how to relate the results in time and towards each other. You are in another room and all you have is the picked up signal and the transmitted signal.
The classical uncertainty is born. You cannot determine the system without disturbing it.

Thus you claim it is uncertain. It cannot be determined. However, what you can do is place both tuning-forks on a single base that can resonate as well and couples the 2 tuning-forks with each other producing a new tone that is based on the characteristics from each individual tuning-fork and the base plate and the frequency of the signal you bring it into oscillation. Basically the 2 tuning-forks are entangled with each other. Now as long as you do not observe again by transmitting a signal through the use of the speaker to get a result back, you are not influencing the whole system while it is resonating. But as soon as you read out 1 tuning-fork, you change the state of the other as well. The magic in entanglement is that the base connecting the 2 tuning-forks is something we can not easily understand. Since it desires to maintain the connection between the 2 tuning-forks as long as there is nothing to disturb the combination of 2 tuning-forks and connecting base. When the system is disturbed the base will become unstable and disappear.

If super position means two states at once, i can understand it as 2 or more frequencies where the super position is a modulation of the 2 or more individual frequencies. As long as you do not disturb the modulation, you could state the modulation describes both frequencies while being different from both.

 

[Farmer]

I have always seen entanglement as the following way :
It also adds a little relativity into the mix.

Take 2 tuning-forks. These can be seen as 2 atoms or as 2 photons. All EM energy is removed , it is a sound proof room. The laser pulse or photon is replaced by a sound wave made up from a sine wave from a speaker which also acts as a microphone to pick up the change. The speaker is rotating around to act as an lesly-speaker while the 2 tuning forks are rotated around as well but at a different rotation frequency.

When you want to observe the 2 tuning-forks, you have to transmit a tone by use of an single sine wave from a speaker and then use that speaker as a microphone. Since the 2 tuning forks are always spaced apart, the peaks of the sine wave from the speaker will hit the 2 tuning forks always at a different time. But that is not all, the lesly speaker is rotating around the setup as well.
But you are not only interested in the frequency returned in hope to determine the F0 or resonating frequency of each individual tuning-fork, you also want the phase delay relative to your own oscillator of each tuning-fork. The oscillator that produced the signal for the speaker. No matter how you try, you can never get the the same phase delay with your setup. The problem is that the sound you receive is always a mixture of sound produced by the 2 tuning-forks. And if you where only trying to determine the phase from 1 tuning-fork, you notice you still get every time a different result.
And of course, you cannot see the setup, thus you have no idea how to relate the results in time and towards each other. You are in another room and all you have is the picked up signal and the transmitted signal.
The classical uncertainty is born. You cannot determine the system without disturbing it.

Thus you claim it is uncertain. It cannot be determined. However, what you can do is place both tuning-forks on a single base that can resonate as well and couples the 2 tuning-forks with each other producing a new tone that is based on the characteristics from each individual tuning-fork and the base plate and the frequency of the signal you bring it into oscillation. Basically the 2 tuning-forks are entangled with each other. Now as long as you do not observe again by transmitting a signal through the use of the speaker to get a result back, you are not influencing the whole system while it is resonating. But as soon as you read out 1 tuning-fork, you change the state of the other as well. The magic in entanglement is that the base connecting the 2 tuning-forks is something we can not easily understand. Since it desires to maintain the connection between the 2 tuning-forks as long as there is nothing to disturb the combination of 2 tuning-forks and connecting base. When the system is disturbed the base will become unstable and disappear.

Is it really necessary to use such an extended and confusing analogy to describe superposition and wavefunction collapse? I feel entanglement is mathematically obvious, but as with so many things in QM, finding an "intuitive" understanding leads to confusion and rampant blabbering.

If super position means two states at once, i can understand it as 2 or more frequencies where the super position is a modulation of the 2 or more individual frequencies. As long as you do not disturb the modulation, you could state the modulation describes both frequencies while being different from both.

Certainly. In your soundbox, any function can be written as a composition of orthogonal basis functions. Since your box is limited in space, these basis functions are countably infinite. These are your eigenstates, and your function is your wavefunction. You're right, the analogy here is obvious because the original Q.M. wave mechanics was founded on this "Fourier" type wave equation solution.

The discussion of entanglement leads to discussion of mixed states, which is related to my question on density matrices. Please answer!

AFAIK D-Wave is held in disdain by most research institutions in quantum information because of their unnecessarily aggressive IP strategy.

 

[wuliheron]

It also doesn't take into account that recent experiments have demonstrated that entanglement is contextual and subject to Indeterminacy. As useful as it often is to describe quantum mechanics in terms of wave-functions it is increasingly obvious this is an approximation of something deeper occurring, which is exactly why theorists are so eager to get their hands on a full scale quantum computer. Sometimes it is just best to say "we don't know how it works, but it does" and leave it at that.

 

[pw38]

Gotta love quantum theory.

 

[William Gaatjes]

Is it really necessary to use such an extended and confusing analogy to describe superposition and wavefunction collapse? I feel entanglement is mathematically obvious, but as with so many things in QM, finding an "intuitive" understanding leads to confusion and rampant blabbering.

It is easier to follow then you think. The magic is in the "base".

Certainly. In your soundbox, any function can be written as a composition of orthogonal basis functions. Since your box is limited in space, these basis functions are countably infinite. These are your eigenstates, and your function is your wavefunction. You're right, the analogy here is obvious because the original Q.M. wave mechanics was founded on this "Fourier" type wave equation solution.

The discussion of entanglement leads to discussion of mixed states, which is related to my question on density matrices. Please answer!

AFAIK D-Wave is held in disdain by most research institutions in quantum information because of their unnecessarily aggressive IP strategy.

What you write is interesting, because i never dug into the subject. I just accepted how in nature everything is a repetition and an analogy of one and other. From that basis i started to think how i could reverse and dig into the realm of the atom. For the last few years, i have had my share of hits with proven experiments and accepted existing theories without being aware of either. But of course i also have had my misinterpretations. I do know nothing of your matrices. I do not look at the world that way. It is to abstract. Knowing how the mind works, moving in the realm of abstraction is a recipe for misinterpretation. ^_^

IMHO :

Nothing is truly infinite. Infinity is a concept to be able to limit your math to a usable way. The amount of atoms found on this and in this planet is a very large number. Big enough, to say it is infinite when only doing a calculation on the amount of atoms in a single euro or dollar coin. It is also of no use depending of the calculation you are doing having to much variables that have no effect on the result in a given fixed situation. It is all about scales.

When you are working with weights of 10 kilograms, and you can only measure and understand in grams, micrograms are part of infinity and uncertainty and indeterminacy. It is the same standard issue that has always been the case. How do you look at wavelengths much shorter then you are capable of. How do you measure a subject that you can not measure within your existing realm. Science is amazing, what it has accomplished by means of indirect measurement. It is fantastic.

 

[William Gaatjes]

Although this is about entangled photons, the linked websites are quite interesting :

http://forums.anandtech.com/showthread.php?t=2078382

 

[William Gaatjes]

I just had this idea about how an atom "jiggles" on temperature in a lattice .

If i would see the entire range of position of the entire atom similar as an modulation of a frequency signal, then the thermal radiation would make for jitter. just as a frequency spectrum can grow larger and smaller. It makes me think of sidebands. It would be like matrix calculations in 3d with frequency modulation. Of course it is just a start, there is more that has an influence.
I think i have to seriously type that out.