Fig. 1: Illustration of the pulse inverse used in the KIK method. | npj Quantum Information

Fig. 1: Illustration of the pulse inverse used in the KIK method.

From: Adaptive quantum error mitigation using pulse-based inverse evolutions

Fig. 1

a Quantum gates are executed via classical control signals, or pulses. The left panel shows a pulse schedule used for a CNOT gate in the IBM quantum computing platform. The pulse schedule in the right panel performs the inverse of the CNOT through the inverse driving \({{{{\mathcal{H}}}}}_{{{{\rm{I}}}}}(t)\). It is constructed from the original pulse schedule \({{{\mathcal{H}}}}(t)\), by inverting the amplitudes of the pulses (black curved arrow) and their time ordering (red curved arrow). b Instead of the pulse inverse, circuit folding and other variants of unitary folding13,34 use the CNOT as its own inverse. Therefore, the pulse schedule for the inverse evolution is not modified. c Noisy implementations of \({{{\mathcal{K}}}}\) and \({{{{\mathcal{K}}}}}_{{{{\rm{I}}}}}\). We assume that during the executions of \({{{\mathcal{K}}}}\) and \({{{{\mathcal{K}}}}}_{{{{\rm{I}}}}}\) temporal variations of the noise due to external factors (e.g. temperature variations) are negligible. Thus, any time dependence in \({{{\mathcal{L}}}}(t)\) is induced by the time dependence of \({{{\mathcal{H}}}}(t)\). (Top) This leads to gate dependent noise depicted by different border colors in the gates Ua, Ub, and Uc. (Bottom) Since \({{{{\mathcal{H}}}}}_{{{{\rm{I}}}}}(t)\) reverses the time ordering of \({{{\mathcal{H}}}}(t)\), the time ordering of \({{{\mathcal{L}}}}(t)\) is also reversed. However, the sign of \({{{\mathcal{L}}}}(t)\) does not change because otherwise the inverse evolution would undo the noise.

Back to article page