Calculation of SSVQE including shotnoise 2.
In this article, we will continue to discuss the results of SSVQE in the system with shot noise.
As a result of the previous calculation with shot noise, the accuracy was significantly reduced regardless of the interatomic distance. This time, we will investigate the cause of this problem by looking at the convergence. The conditions are the same as before, but the interatomic distance is set to r=0.7414.
The convergence and noise for each level are shown in Figure 1. The noise is small in the Hamiltonian, but the energy does not converge and fluctuates wildly. The calculation was continued until about 250 iterations later, but the results did not change. The conditional terms are shown in Figure 2. This term also fluctuates wildly, but it periodically goes to zero. Therefore, the SWAP-test, which is not shown here, was found to be the cause. This can also be seen from the results of the calculation without the SWAP-test in the VQE here.
It may be a part of our ability to create an evaluation function that contains as little inner product as possible.
Figure 1 Variation of the values of each energy level and shot noise against the number of calculation iterations.
Figure 2 Conditional terms with shot noise for the energy levels for the number of calculation iterations.