QPU concepts
at our company (Finance sector) we've got great opportunity (finally can spent working time on this matter, not just limited private time) to participate at quantum hackathon, using universal quantum computer / gate model.
do you think the knowledge gained over time with d-wave can be used here? i mean for the same optimization problem do you use the same thinking / translation (formulating problem as BQC) or the Ising/qubo formula is not used and replaced by something entirely different?
so in short, is there a synergy effect of understanding / learning annealing and gate model, or you simply must make a choice and follow only direction that pleases you more?
can you think of a comparison like, e.g. that programming dwave vs gate model is like programming python vs R?
thank you :-)
Comments
Hi Branislav,
Gate-model QPUs can run different algorithms, but most likely if you want to get good answers to an optimization problem, the gate-model algorithm you would want to use is QAOA (Quantum Approximate Optimization Algorithm). The other algorithms that can be run on gate-model QPUs are only applicable to specific problem types (e.g., quantum molecular simulation).
Quantum annealing on a D-Wave QPU and QAOA on a gate-model QPU will both require you to formulate your input in Ising/QUBO format. They also both require embedding into a QPU working graph. So in that sense, there's good overlap in learning annealing and gate-model quantum computing, since formulating your problem properly and embedding it make up a very large amount of the work.
Interesting, thank you for your answer Jamie K. appreciated. this helps me further with my search :-)
Hello Branislav,
I wanted to share a response to your question about comparing gate model and quantum annealing approaches. This comes from one of our researchers:
Solving a problem by quantum annealing is more like the declarative programming paradigm: you formulate the desired outcome and sent it to the QPU for solution without giving step-by-step instructions about how to solve it. Examples of this paradigm in everyday programming include Prolog, SQL, and yes, R. The main challenge in programming this way is figuring out how to formulate your problem in the expected framework. Solving a problem in the gate model is more like the imperative programming paradigm, where you spell out the instructions for how to do it. This is like Python programming, where the main challenge is figuring out the step-by-step instructions.
To all,
I know very very little about QC'g, (quantum computing). However I agree Mr. Branislav, that [paraphrased] D-Wave's paradigm is very declarative like PL. Looking over Microsoft's (docs.microsoft...) gate-model examples in the past, appear to be more, (or perhaps much more) imperative. They use functions of course, but moreover, operations, to define;
SignChange(){}, StateChange() {}, PhaseFlip(){}, AmplitudeChange(){}, various BellStates(){} and multi-qubit-states(){} ,
etc etc.
(I lost a tiny bit of interest in the Microsoft quantum documentation (intuitive to follow IMHO) mainly because of the Ocean-SDK && the declarativeness the coding offers.)
In checking out D-Waves nurse_scheduling example, it's clear that the anneal coding is profoundly declarative.
(PS. BTW, I don't know very many ppl that use PL (Prolog) for "everyday programming". : ) )
Cherio.
Please sign in to leave a comment.