14
$\begingroup$

Google recently announced the Bristlecone 72 qubit quantum computer. However, D-Wave already announced quantum computers featuring more than $2000$ qubits.

Why is Google's new device newsworthy then? Is it better than D-Wave's machine in some respects? If so, how?

$\endgroup$

marked as duplicate by dtell, Sanchayan Dutta, MEE was the missing bracket, Discrete lizard, dippas Mar 31 '18 at 11:53

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 10
    $\begingroup$ A record for most qubits is like a record for a pie eating contest. It encourages sheer quantity, without any emphasis on either the pie being particularly delicious or the eating being particularly elegant. Perhaps ironically, there is more to computing hardware than sheer numbers. $\endgroup$ – Niel de Beaudrap Mar 28 '18 at 23:50
  • $\begingroup$ Related: How should different quantum computing devices be compared? $\endgroup$ – glS Mar 29 '18 at 0:03
  • 4
    $\begingroup$ I think it is worth noting that there is no public data from Google's 72 qubit device, so it's effectiveness cannot be commented upon directly. $\endgroup$ – James Wootton Mar 29 '18 at 11:17
11
$\begingroup$

Short explanation:

D-Wave implements quantum annealing, while Google has digitized adiabatic quantum computation.


Lengthy Explanation:

D-Wave advertises their line of quantum computers as having thousands of qubits, though these systems are designed specifically for quadratic unconstrained binary optimization. More information about D-Wave's manufacturing process.

It is D-Wave's claim that: "It is best suited to tackling complex optimization problems that exist across many domains such as":

  • Optimization

  • Machine learning

  • Sampling / Monte Carlo

  • Pattern recognition and anomaly detection

  • Cyber security

  • Image analysis

  • Financial analysis

  • Software / hardware verification and validation

  • Bioinformatics / cancer research

D-Wave's QPU uses quantum annealing (QA), a metaheuristic for finding the global minimum of a given objective function over a given set of candidate solutions (candidate states), by a process using quantum fluctuations. Quantum annealing is used mainly for problems where the search space is discrete (combinatorial optimization problems) with many local minima; such as finding the ground state of a spin glass.

D-Wave's architecture differs from traditional quantum computers. It is not known to be polynomially equivalent to a universal quantum computer and, in particular, cannot execute Shor's algorithm because Shor's Algorithm is not a hillclimbing process. Shor's Algorithm requires a universal quantum computer. D-wave claims only to do quantum annealing.[citation needed]

Papers:

Experimental quantum annealing: case study involving the graph isomorphism problem

Defects in Quantum Computers


Google's claim is: "The goal of the Google Quantum AI lab is to build a quantum computer that can be used to solve real-world problems. Our strategy is to explore near-term applications using systems that are forward compatible to a large-scale universal error-corrected quantum computer using linear array technology".

Papers:

"State preservation by repetitive error detection in a superconducting quantum circuit"

"Digitized adiabatic quantum computing with a superconducting circuit"


Inaccurate layperson' explanation:

A Graphic Card has more Cores than a CPU.

GPUs are optimized for taking huge batches of data and performing the same operation over and over very quickly, unlike PC microprocessors, which tend to skip all over the place.

Architecturally, the CPU is composed of just few cores with lots of cache memory that can handle a few software threads at a time. In contrast, a GPU is composed of hundreds of cores that can handle thousands of threads simultaneously.


Technical, but not overly complicated, layperson's explanation:

Why is Google's new device newsworthy then? Is it better than D-Wave's machine in some respects? If so, how?

There are "Annealing QPUs" and "Universal QPUs" as explained above, an incomplete list is offered on Wikipedia's page: "List of Quantum Processors".

In quantum annealing, the strength of transverse field determines the quantum-mechanical probability to change the amplitudes of all states in parallel.

In the case of annealing a purely mathematical objective function, one may consider the variables in the problem to be classical degrees of freedom, and the cost functions to be the potential energy function (classical Hamiltonian). Moreover, it may be able to do this without the tight error controls needed to harness the quantum entanglement used in more traditional quantum algorithms.

That makes it easier to provide more qubits, but the kinds of problems they are able to solve is more limited than the qubits provided in a universal QPU.

In general the ground state of a Hamiltonian can be used to encode a wider variety of problems than NP (know QMA-complete problems), and so decision to focus on NP optimization problems has led to restrictions which prevent the device from being used for general purpose quantum computing (even if noise was not an issue).

There is an interesting subtlety as regards noise: If you add noise to the adiabatic algorithm, it degrades gracefully into one of the best classical algorithms for the same problem.

The adiabatic model can encode universal quantum computation, however the limitations of DWave's implementation means that specific machine cannot.

Google's universal QPU can solve a wider range of problems than D-Wave's QPU (in it's current implementation) if they can solve their decoherence problem.

In the case of Google's Bristlecone caution is warranted. Bristlecone is a scaled up version of a 9-qubit Google design that has failed to yield acceptable error rates for a commercially viable quantum system. In real-world settings, quantum processors must have a two-qubit error rate of less than 0.5 percent. According to Google, its best result has been a 0.6 percent error rate using its much smaller 9-qubit hardware.

The commercial success of quantum computing will require more than high qubit numbers. It will depend on quality qubits with low error rates and long-lasting circuit connectivity in a system with the ability to outperform classic computers in complex problem solving, i.e., “quantum supremacy”.

Google will use it's record number of more useful qubits to correct the error rate of those error prone qubits.

More qubits are needed to solve bigger problems and longer living (coherent) qubits to are needed to hold the information long enough for the quantum algorithm to run. IBM describes the problem as: "Quantum Volume: preferring fewer errors per qubit over more qubits", see also: What is the leading edge technology for creating a quantum computer with the fewest errors? .

Google plans to use Surface Codes to resolve this problem, for more info and a comparison to spin glass models see: "Quantum Computation with Topological Codes: from qubit to topological fault-tolerance".

IBM has a video titled: "A Beginner’s Guide to Quantum Computing" which explains quantum computing for laypersons in under 20 minutes.

Microsoft intends to take the wind from everyone's sails with the integration of Q# (Q sharp) into Visual Studio and some information about their Majorana fermion based qubits, and a great reduction in the error rate, in the months to come. See: "Majorana-based fermionic quantum computation". The will enable a system that uses less than 25% as many better qubits to accomplish the same amount of work as Google's qubits.

The website "The Next Platform" describes the current situation as: "Quantum Computing Enters 2018 Like it's 1968".

$\endgroup$
  • $\begingroup$ That doesn't look so much like an "inaccurate layperson explanation" as an (accurate?) analogy, except that you didn't quite explain how the analogy applies (presumably the D-Wave machine is like a GPU, but in what way?). $\endgroup$ – Kyle Strand Mar 29 '18 at 1:04
  • $\begingroup$ Well, that's the point--you said it was inaccurate. Importantly, it seems that GPUs can execute the full class of algorithms that CPUs can execute, but that doesn't seem to be the case for annealing vs adiabatic quantum computers ("cannot execute Shor's algorithm"). So I'm just suggesting that you change your final heading to simply say "Analogy" and then add a final paragraph bringing the analogy full circle. $\endgroup$ – Kyle Strand Mar 29 '18 at 15:13
  • $\begingroup$ GPU hardware has a Turing-complete set of operations (arithmetic, memory writes, and branches), so yes, GPUs can perform the same set of operations that CPUs can, they're just slower when the operations aren't highly parallelized. I understand that you contributed this answer in your free time, and I'm glad that you're going to continue to work on it; I just wanted to make a suggestion that I think is important for how to improve it. $\endgroup$ – Kyle Strand Mar 29 '18 at 17:59
  • $\begingroup$ The summary at the beginning of this answer is a bit misleading. The Google 72 qubit device is not an adiabatic quantum computer. It's a gate model device. $\endgroup$ – DanielSank Apr 5 '18 at 5:21
  • $\begingroup$ @DanielSank - That's exactly what the title of the paper you are credited on is called: "Digitized adiabatic quantum computing with a superconducting circuit", and those words are linked to that paper. $\endgroup$ – Rob Apr 5 '18 at 8:09
11
$\begingroup$

There are two points I'd make here.

D-Wave's computer and Google's computer are fundamentally different.

D-Wave's computer is a quantum annealer. Imagine a landscape with some grassy hills. If you put a ball at the top of the hill, it will roll to a local minima, or even the minimum - in this case, a valley. Similarly, a quantum annealer has the qubits as the ball and a polynomial as the landscape. It has the advantage that effects like quantum tunneling help make the process more efficient.

On the other hand, Google's computer is gate based. Much like a digital classical computer, it has qubits and gates that are then applied to those qubits. This means that they are optimized, in some senses, for different types of problems. It's like comparing apples and oranges - not totally equivalent.

Just because you have a lot of qubits doesn't mean they are good qubits.

A lot of the controversy surrounding D-Wave has been because it has been called into question whether or not D-Wave's qubits actually exhibit quantum effects. (It hasn't really helped that D-Wave has...overhyped their successes a bit along the way.) You can have a ton of qubits, but if they aren't actually coherent for a long enough length of time, it's not really useful.

Google's qubits are definitely coherent for a reasonable period of time. We don't 100% know about D-Wave's.

Tl;dr: both are different in how they work and in their problems.

Both are notable in their own ways.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.