Data for "Excising dead components in the surface code using minimally invasive alterations: A performance study"

This repo contains the circuits and corresponding data for the simulations performed in arXiv:2508.XXXXX. The circuits used for Fig. 2 (Figs. 3-5) are contained in single_dead_circuits.zip (distribution_circuits.zip); the corresponding simulation data is in single_dead_data.csv (distribution_data.csv). The dead-qubit configs used for Figs. 3-5 are in selected_configs.csv. circuit_{circuit_index}.txt is the circuit associated with column `circuit_index` in the respective data files. The circuit description is derived from one we use internally for QEC simulations at Microsoft, though it shares similarities with the Stim format. The simulated circuits only contain 1- and 2-qubit Pauli measurements and idle operations: - Measurement operations are denoted as m{Pauli operators} {qubit ids}- Idle operations are denoted as i {qubit id} For the qubit ids, we use the spatial coordinates of the qubit as a 2-tuple. Blocks of operations happening simultaneously in time begin with: -------- {time step} We also place labels on the operations as follows, starting with the @ marking: - If noise is applied to the operation according to the noise model specified in the main text, we apply the label 'noisy' (operations near the beginning and end of the simulation cell in time may be noiseless, as we buttress the noisy rounds with O(1) noiseless rounds).- If a flipped measurement outcome flips check (in the chosen basis of checks used for decoding) we add a label 'check{i}'. For example, mxx (4, 4) (4, 4.5) @noisy check205 check236 denotes a noisy XX measurement between qubits located at (4, 4) and (4, 4.5), whichcontributes to checks 205 and 236. For Auger circuits that fail to form a logical qubit, we report p_logical = 50%, and for a few data points that failed to run due to HPC issues, we report values based on linear interpolation in log-log space using successfully completed runs at nearby p_physical.