RESEARCH INFRASTRUCTURE / EVIDENCE REGISTER
Research benchmarks.
Evidence before claims.
A transparent register of published reference results, defined experimental targets, and measurements that Gottesman Software has independently reproduced or verified.
01Published literatureSource DOI · conditions · reported metric
↗
02Gottesman targetWorkload · acceptance criteria · protocol
↗
03Measured resultPending laboratory experiment
○
04Independent verificationEvidence · environment · reproducible run
○
03Research streams
03Published LiDMaS+ studies
03Literature anchors
04Initial hardware tests
00Independently verified results
01 / VALIDATION PROGRAMME
Targets are not results.
Every target is marked by evidence level and associated with the scope we intend to test. The register starts with the electrical ESP32-S3 loopback bench.
Literature: independently published
Published: project paper or preprint
Planned: proposed criterion
Pending: no measured result
Verified: requires evidence artifact
STREAM 01 / HARDWAREPlanned test
Photon-QDrivers ESP32-S3 N16R8
Digital pulse-generation and GPIO-loopback experiments using the proposed USB-powered bench. No photon detector or FPGA timing claims are made for this phase.
Proposed Phase 1 validation criteria| Metric | Proposed criterion | Observed |
|---|
| Pulse frequency | 1 kHz ± 1%, after analyzer verification | Pending |
|---|
| Loopback event agreement | 10,000 / 10,000 at 1 kHz under defined input conditions | Pending |
|---|
| Host command integrity | 100 / 100 acknowledgements under a fixed test sequence | Pending |
|---|
| Local display | OLED count and state agree with host response | Pending |
|---|
STREAM 02 / QECPublished evidence
LiDMaS+ Decoder validation
Compare logical error rate and decoding latency with established decoder implementations on matched codes, noise, syndrome streams and host configurations.
Primary measurementsLogical error rate · threshold crossings · decoder throughput
Peer-reviewed resultMedian pc = 0.0531; bootstrap interval [0.0415, 0.0572]
Public replay result4,108 / 4,108 fixture and public-slice lines reported without parse errors
Independent verificationNot yet completed by this register
STREAM 03 / SIMULATIONBaseline design
SchroSIM Photonic numerical agreement
Define like-for-like Gaussian and selected non-Gaussian circuits; compare numerical observables, execution time and peak memory at matched precision and cutoff settings.
Primary measurementsObservable agreement · runtime · peak RAM
Reference classReproducible photonic simulator implementations
Numerical acceptance thresholdTo be set per circuit family and tolerance
Observed measurementsNot yet reported in this register
No matches in this register. Try another stream or search term.
02 / LIDMAS+ RESULTS
Published evidence from the project.
These claims are reported in peer-reviewed or public research. Their publication status is shown separately from this register's stricter independent-verification status.
PEER REVIEWED / QEC THRESHOLDS2026
Decoder dependence under hybrid CV-discrete noise
0.0531Pauli matching-style median crossing pc
bootstrap interval [0.0415, 0.0572]
Dense hybrid sweeps used d=3, 5, and 7 with 3,000 trials per point. Decoder and estimator sensitivity remain part of the claim boundary.
Fortschritte der Physik 74(6), e70124 · DOI ↗
PREPRINT / PARALLEL SAMPLING2026
Native GKP digitization and decoder comparison
1.34× / 1.94×reported threaded speedups
Pauli / native GKP
The same study reports mean absolute logical-error-rate deltas of 6.07e-3 and 5.20e-3 between serial and threaded sampling.
arXiv:2603.25757 ↗
PREPRINT / REPLAY ARCHITECTURE2026
Hardware-to-decoder evidence path
4,108 / 4,108fixture and public-slice lines
reported without parse errors
The scope is fixture input and sampled public datasets, not live photonic hardware. Rerun artifacts were reported with matching SHA-256 hashes.
arXiv:2604.15389 ↗
Public releaseLiDMaS+ v1.2.2
RuntimeC++20 + Python 3.9+
LicenseMIT
03 / PUBLISHED CONTEXT
Literature reference library.
Each cited result belongs to its original instrument, software, workload and measurement conditions. Published performance is not a claim of Gottesman performance.
HARDWARE / FPGA TIMING2020
High-count-rate coincidence counting
Arabul, Paesani, Tancock, Rarity & Dahnoun · IEEE Photonics Journal
Specialized FPGA-based coincidence system with a TDC. Not directly comparable with ESP32 GPIO sampling or a 24 MHz USB analyzer.
Read paper · DOI 10.1109/JPHOT.2020.2968724 ↗
QEC / MATCHING DECODER2023
Sparse blossom: fast MWPM decoding
Higgott & Gidney · arXiv:2303.15933
Specific syndrome-decoding workload, not a general CPU or FPGA latency bound. A valid comparison requires matched noise, code distance and experimental protocol.
Read paper · arXiv:2303.15933 ↗
PHOTONICS / SIMULATION2019
Strawberry Fields photonic platform
Killoran et al. · Quantum 3, 129
Methodological reference, not a numeric speed target. Gottesman comparisons must first define equivalent circuits, cutoff dimensions and numerical precision.
Read paper · DOI 10.22331/q-2019-03-11-129 ↗
04 / REPRODUCIBILITY
What qualifies as verified?
A result is promoted from “measured” to “verified” only when its inputs and provenance can be inspected. Publication alone does not replace this register's independent rerun requirement.
01Freeze the workload
Record circuit, noise model, timing mode, sample count, device, firmware and software versions.
02Capture raw evidence
Archive traces, data files, logs, scripts, configuration and calibration details with a run identifier.
03Reproduce & review
Publish commands, the Git commit and uncertainty or failure modes; repeat the experiment independently.
Comparison boundaryFor FPGA latency, separate hardware event-to-decision latency from host command-to-result latency. Do not compare picosecond TDC precision with a low-rate USB logic analyzer. All numbers displayed as Gottesman targets are proposals until measurement artifacts are available.