Our interest is to develop a new approach to annealing. Something that is a significantly different from existing approaches. But we also want to retain use of the D-Wave squids, so we would like to code something innovative using D-Wave.
I saw that the D-Wave implements their annealing approach using C++ code: cpu_sa.cpp along with simulated_annealing.cpp and simulated_annealing.pyx.
I've been assuming that to test a radically new annealing algorithm we need to implement similar code for our algorithm (maybe not cpu_sa.cpp, but I can't say for sure). If this has been done before, it would VERY helpful to see an example.
The approach we have envisioned is different enough such that the solver parameters and annealing schedule will not give us what we are looking for.
The simulated_annealing code that you mention above is actually its own implementation that is separate from our QPU system, and is classical rather than quantum.
The SimulatedAnnealingSampler documentation references links to Simulated Annealing where you can read more about it.
Take a look at our Quantum Annealing documentation to learn more about our D-Wave Systems.
Much of our technology is proprietary, so it is unclear what kind of collaboration options are available. You can reach out to us on our contact page regarding inquiries such as these, and we can point you towards the appropriate team member.
Comments
Hello,
Are you referring to adjusting the annealing schedule?
There is documentation on this in a few places.
This notebook is great and goes over specifically how to adjust the annealing schedule:
https://github.com/dwave-examples/anneal-schedule-notebook
The Solver Parameters page has more information about what options are available when using a QPU:
https://docs.dwavesys.com/docs/latest/c_solver_parameters.html
Our documentation also has a section on annealing implementation and controls:
https://docs.dwavesys.com/docs/latest/c_qpu_annealing.html
I hope this is helpful. Please let us know if you have any more questions or if you meant something slightly different.
Our interest is to develop a new approach to annealing. Something that is a significantly different from existing approaches. But we also want to retain use of the D-Wave squids, so we would like to code something innovative using D-Wave.
I saw that the D-Wave implements their annealing approach using C++ code: cpu_sa.cpp along with simulated_annealing.cpp and simulated_annealing.pyx.
I've been assuming that to test a radically new annealing algorithm we need to implement similar code for our algorithm (maybe not cpu_sa.cpp, but I can't say for sure). If this has been done before, it would VERY helpful to see an example.
The approach we have envisioned is different enough such that the solver parameters and annealing schedule will not give us what we are looking for.
Thank you for such a fast response!
Is D-Wave reticent to help with research in this area, given that their current algorithms appear to be a market differentiator?
Hello,
The simulated_annealing code that you mention above is actually its own implementation that is separate from our QPU system, and is classical rather than quantum.
The SimulatedAnnealingSampler documentation references links to Simulated Annealing where you can read more about it.
Take a look at our Quantum Annealing documentation to learn more about our D-Wave Systems.
Much of our technology is proprietary, so it is unclear what kind of collaboration options are available. You can reach out to us on our contact page regarding inquiries such as these, and we can point you towards the appropriate team member.
Please sign in to leave a comment.