Fig 1.
Schematic diagram showing the process of solving the linear equations by Ising machines.
Fig 2.
Mean errors of the four representations, RANDOM_UNIFORM, RANDOM_NORMAL, CONSTANT, and BINARY, when linear equations are solved using the Fixstars Amplify AE with various settings.
The error bars denote the standard error of the estimated mean errors.
Fig 3.
Plots of the coefficient values for various conditions using the Fixstars Amplify AE.
(A) The basis vectors of each coefficient, where the first and second basis vectors p1,p2 represent the performance worsens as the number of bits R increases and the performance improves as R increases, respectively. (B) The coefficient values of all conditions tested. (C) The coefficient values by real number representation and by parameter. Lighter colors indicate higher parameter values.
Fig 4.
Plots of the coefficient values for various conditions using (A) D-Wave hybrid solver with timeout time T = 3000 ms. and (B) D-Wave full quantum solver with read counts TC = 103, 104 (corresponding to timeout time T).
Fig 5.
The plot is divided into subplots vertically by the number of bits, R, and horizontally by the maximum number of simultaneous bits flip, X. The errors achieved in all trials are sorted in each subplot. The left-most subplot summarizes the averages of these trials.
Fig 6.
The plot is divided to subplots vertically by the number of bits, R, and horizontally by the number of flip steps, X. The errors achieved in all trials are sorted in each subplot. The left-most subplot summarizes the averages of these trials.
Fig 7.
The figure shows panels for different number of bits, R. Each curve indicates the averaged error probability function 〈E(p)〉.