Figure 6From: Robustness of quantum reinforcement learning under hardware errorsSimulation results of evaluating the trace of the Hessian matrix for the circuit shown in Fig. 2(b) with random assignments of the parameters and \(O = Z^{\otimes 4}\). The simulations are performed by sampling 2000 random parameter vectors \(\{\boldsymbol{\theta}_{m}\}_{m=1}^{2000}\) with \(\theta _{i} \sim \operatorname{Unif}[0,2\pi [\) and then evaluating the trace of the corresponding Hessian matrix \(\operatorname{Tr} [H(\boldsymbol{\theta}_{m})]\). These values are used to build the histogram showing the frequency distribution of \(\operatorname{Tr} [H]\). The length of the arrows are, respectively: “Numerical 2σ” (black solid line) twice the numerical standard deviation, “Approximation” (dashed red) twice the square root of the approximation in Eq. (31), “Bound” (dashed-dotted green) twice the square root of the upper bound in Eq. (30)Back to article page