Hi, it's a petit breakthrough in bringing supercomputer tech to the PC. blueqatSDK allows us to do large scale QAOA. Until now, it was at most 10-20 qubits, but now we can do 100-1000 qubit units with some ease, although it is a simulation.
Explanation.
QAOA is an algorithm that solves a combinatorial optimization problem called quantum annealing on a quantum computer with a quantum gate method.
Basically, it decomposes a combinatorial optimization problem into a set of problems, called time evolution. This time evolution is decomposed into operations called quantum gates.
In this case, we used a calculation called QAOA to decompose the equation for calculating the minimization of the social problem into quantum gates, with a size of 100 qubits. As a result, the PC at hand is an intel i5-10300, and although it took a few minutes, the calculation was completed normally without any major obstacles.
I then used the optimized quantum circuit and sampled the solution 100 times. As you can see in the figure below, it seems to be very difficult to evaluate the solution.
In QAOA, what is minimized is the expression <psi|H|psi>, which is the expected value. Since these can be obtained without storing the full state vector, it is now possible to perform large-scale calculations that were previously said to be possible only up to 50 qubits. Also, sampling, which is the process of extracting the solution, can be calculated without problems.
Also, this is still a CPU-based calculation, and the performance can be dramatically increased by using a GPU, as in AI.
The new blueqatSDK will soon be available to customers as a beta version and will be released in the future, so please stay tuned.
That's all.