I know some of you are curious about the new blueqat, so I will introduce the new blueqat in a little bit. The specifications will change as it is still under development.
The first and most important point is that the number of available qubits will be much higher than the previous quantum computer SDKs.
The quantum supremacy that Google is said to have achieved in 2018 has also been surpassed in computation time by supercomputers. Basically, the calculations that Google claimed would take 10,000 years were not done by properly devising existing computational methods, and by devising those methods, the simulation itself can be made faster.
With blueqat, we are bringing such state-of-the-art supercomputer simulation methods to the consumer market, so that even blueqat can perform calculations in the 100- to thousands-of-qubit range. If you want to do large calculations, you will need a large machine with performance similar to a supercomputer, so this is not an easy task, but we will gradually work on this in the future.
You may say that it is meaningless if the actual machine is not large even if the simulation is large, but IBM's 1,000-qubit class superconductors and the neutral-atom quantum computers of various companies are in charge of this volume zone.
Basically, the usage remains the same. You can still run blueqat to calculate large qubits. Here is a calculation of 200 qubits with H-gate and CX-gate applied appropriately. The calculation is fast.
However, there is a trade-off. You cannot get the full vector that stores the quantum state, called the state vector. Instead, you have to take the amplitude of a particular state, or sample it, or take the expected value of the Hamiltonian.
In fact, the simulator cheats somewhat. Using state vectors allows us to do calculations that quantum computers can't do, and we can omit some of the calculations. This new blueqat basically does not use state vectors, so we will mainly use samples and Hamiltonian expectation values.
Actually, this is also cheating somewhat, but since it will be more similar to the behavior of the actual machine, it is possible that there will be areas where the previous code will need to be rewritten somewhat. We'll definitely address this as soon as possible!