This is a celebratory post for our new paper titled “Error Mitigation in Dynamic Circuits for Hamiltonian Simulation” which was accepted in GLS VLSI 2026 as a short paper and invited talk! This is the first peer reviewed paper of my PhD journey. Although we did put up something on the arxiv before.

We benchmarked standard error mitigation techniques like dynamical decoupling (DD) and zero-noise extrapolation (ZNE) when simulating quantum Hamiltonians, like the 1D Ising and Heisenberg chains, via Trotterization with circuits that include mid-circuit measurements (MCMs) and adaptive operations. Code and data are available here.

My awesome advisor got us premium access to IBM Quantum superconducting hardware through LBNL for performing these experiments. I had to submit custom sampler jobs to IBM Quantum because they don’t yet accept estimator jobs for circuits containing dynamic instructions like while loops or if/else statements. And I kind of understand why, it is difficult to integrate error mitigation techniques like DD/ZNE/PEC etc. when your circuit changes across shots.

There has been some amazing work in error mitigation for example [PEC] [PROM] [MCMit].

We were lucky to be dealing with “dynamic gadgets” with MCMs and feedforwards that implement a specific unitary subcircuit that could be inverted easily. But it does raise the questions: how do we systematically amplify noise in non-unitary channels for ZNE? Would it even matter in the FTQC era?