Selective state space model for photon energy estimation in microwave kinetic inductance detectors

Microwave Kinetic Inductance Detectors (MKIDs) are superconducting photon-counting detectors that measure the energy and arrival time of individual photons on large-format arrays. The standard energy estimator applies a coordinate transform followed by a Wiener optimal filter, optimal only for stationary noise, linear response, and energy-independent pulse shapes, assumptions that MKIDs structurally violate. We present Venom (Very Efficient Neural Optimal-filter for MKIDs), a selective state space model based on Mamba that replaces the coordinate transform and optimal filter with a single learned sequence model operating directly on raw in-phase/quadrature (IQ) timestream data. The model carries 3,314 trainable parameters and targets deployment on the MKIDGen3 RFSoC readout platform. We train on synthetic pulses drawn from a streaming-PCA plus shape-preserving (PCHIP) covariance interpolator built from calibration data, allowing continuous sampling of photon energy between calibration wavelengths. We validate Venom on two published MKID data sets: an InHf bilayer resonator at ten wavelengths from 254 to 1310 nm (Zobrist et al. 2022) and a PtSi resonator at five wavelengths from 808 to 1310 nm (Zobrist et al. 2019). All numbers are reported on a stratified 15% held-out validation set unseen by the model and synthetic-pulse generator during training. On InHf, Venom reaches a mean KDE resolving power of 26.1 versus 24.8 for the per-wavelength optimal-filter baseline, a 5% improvement. The R of 42.4 at 254 nm is the highest R ever recorded for a UVOIR MKID suitable for use in an array. On PtSi, Venom reaches a mean KDE R = 19.5 vs. R = 11.8 for the shared 920 nm template filter on the same held-out subset, a 66% improvement. This increase may reflect a fault with our previous optimal filtering approach on PtSi, or new physics related to pulse shape variations with photon absorption depth.