Supplementary Materials for "Universal Sensor Frontend for Event Inference in Photonic Stochastic Systems"

Supplementary Material for:“Universal Edge Frontend for Heterogeneous Sensors in Photonic Stochastic Systems”OverviewThis supplementary package contains the full computational materials used tosupport the modeling study of a universal stochastic edge frontend forheterogeneous sensors and event-level inference in photonic stochastic systems.The package is organized into five main computational stages:- S1: Core frontend validation- S2: Weighting strategies study- S3: Synchronization and decorrelation study- S3b: Time-varying synchronization study- S4: Robustness and scaling studyEach stage includes:- a Jupyter notebook (.ipynb)- an exported Python script (.py)- a local stage-specific README file- a results folder containing generated data, figures, tables, and a JSON summaryRecommended reading / execution orderThe stages are intended to be read and, if desired, executed in the following order:1. S1_core_frontend_validation2. S2_weighting_strategies3. S3_synchronization_and_decorrelation4. S3b_time_varying_synchronization5. S4_robustness_and_scalingThis order follows the logical progression of the study:- first validating the universal frontend,- then comparing weighting policies,- then analyzing decorrelation,- then introducing time-varying synchronization effects,- and finally testing robustness and scaling.Folder structureSupplement S1 notebook/Baseline validation of heterogeneous sensor-to-stochastic mapping.Supplement S2 notebook/Comparison of equal, expert, reliability-aware, and data-driven weighting.Supplement S3 notebook/Static synchronization/decorrelation analysis with autonomous, shared,and grouped stochastic pulse sequence generation.Supplement S3b notebook/Time-varying synchronization study with slot-wise event inference,fixed lags, and random jitter.Supplement S4 notebook/Robustness and scaling analysis under noise, dropout, missing slots,weight perturbation, and increasing channel count.Main scientific role of each stageS1Demonstrates that heterogeneous sensor channels can be mapped into a commonevent-oriented probability space and converted into Bernoulli pulse sequencewhose stochastic event estimate converges to the float baseline.S2Shows that weighting is a critical design parameter in event-level fusion,with reliability-aware weighting outperforming naive equal weighting in thetested benchmark.S3Demonstrates that inter-channel correlation is a first-order system factor:autonomous stochastic pulse sequences outperform grouped and shared-randomnessgeneration in the static formulation.S3bThe synchronization analysis is extended to a time-varying setting, showing that temporal misalignment becomes measurable when event inference is performed slot-wise rather than after averaging the full pulse sequence.S4Demonstrates graceful degradation under moderate corruption and identifieschannel criticality, weight sensitivity, and scaling behavior of theuniversal frontend.Suggested use by readers/reviewersReaders who want a minimal technical overview may inspect:- S1 results for baseline validation,- S2 results for weighting comparison,- S3b results for time-varying synchronization,- S4 results for robustness and scaling.Readers who want full reproducibility should use the notebooks in the orderlisted above.Software environmentThese notebooks were prepared and tested in Google Colab using Python 3.The main libraries used in the package are:- numpy- pandas- matplotlib- scikit-learn- json / zipfile / dataclasses / typing (standard library)A separate file named software_environment.txt is included for quick reference.Notes on results foldersEach stage contains a results folder with:- generated data files,- figures used for interpretation,- tables used for reporting metrics,- a JSON summary of key parameters.These folders are intended to make the package self-contained and easy to audit.Final noteThe supplementary package is designed to support the article at the levels of:- architectural validation,- methodological reproducibility,- sensitivity analysis,- robustness evaluation,- and scalability assessment.