ELSD Platform  ·  Electronic Landscape Stability Diagnostics
📄 Published audit record  ·  5 corrections on Zenodo  ·  May 2026

R&D cycles are expensive.
Unreliable computational models make them worse.
We fix that upstream.

Quantum results should not be trusted just because they converged. We test whether the energy deserves trust.

Before battery developers, drug discovery teams, and materials scientists commit resources to synthesis and testing, Quantum-Clarity audits whether the electronic model they are building on is stable, sector-clean, reference-comparable, and decision-grade — or exactly why it is not.

5
Public Zenodo
corrections (2026)
1,500+
VQE ensemble runs
completed
20 q
Reduced active-space
operating point
15–35
Independent seeds
per condition

ELSD is a diagnostic layer, not a black-box molecule solver. Its value is not that every VQE result is correct. Its value is that unstable, sector-wrong, reference-incomparable, or optimizer-pathological results are not allowed to masquerade as physical discoveries.

Why this matters before your next R&D decision

No PhD required. Here is the business problem ELSD solves, in plain language.

The problem
Computational models fail silently

Drug discovery, battery materials, and propulsion R&D all rely on quantum chemistry models to decide which molecules or materials to synthesise and test. A single model run costs seconds and a few dollars. The synthesis or test it informs can cost $50,000 to $5 million and six months.

The problem: most quantum chemistry tools report a result when they converge — but convergence does not mean the result is correct. The model may have settled into the wrong electronic state, or it may give a different answer every time it is run from a different starting point. Neither failure mode is visible in a standard output file. The expensive R&D cycle downstream is built on a foundation no one checked.

The solution
An audit layer that checks before you commit

ELSD runs the same quantum chemistry model 15–35 times from independent starting points, measures whether each run lands in the correct electronic state, quantifies how reproducible the results are, and compares every result against a mathematically exact reference computed independently.

The output is not just an energy number. It is a verdict: this model is stable and decision-grade, or this model has a specific problem — here is what it is and why. That verdict comes before synthesis, before testing, before the expensive commitment. That is the value.

The evidence
Five public corrections prove the discipline is real

Any platform can claim to audit computational models. What makes ELSD credible is that it has been turned on its own results — five times, publicly, with the corrections deposited on Zenodo and linked from this page.

In each case: we ran the audit, found problems in our own prior work, withdrew or narrowed the claims that didn't survive, and published exactly what did. The tools that found the problems are deposited alongside the results so any external reader can re-run them. That standard of evidence is rare in computational chemistry — and it is what makes the surviving results worth trusting.

The market
Every organisation running quantum chemistry has this problem

Pharmaceutical companies running quantum-chemistry-guided drug discovery. Battery manufacturers using computational screening to select cathode or electrolyte candidates. Defense contractors and propulsion engineers modeling metal–fuel interfaces. Materials companies developing catalysts or coatings.

All of them face the same upstream question: can I trust this calculation enough to act on it? None of their current tools answer that question systematically. ELSD does — with a public audit trail, an exact-reference verification step, and results that have already been stress-tested against the platform's own prior work.

The one-paragraph version

Quantum chemistry models are used to guide billion-dollar R&D decisions in drug discovery, energy storage, and advanced materials — but they routinely produce results that look correct and aren't. ELSD is the audit layer that catches that before the expensive decision gets made. It runs models repeatedly from independent starting points, verifies the electronic state of every result, and compares against an exact mathematical reference. The output is a trust verdict, not just an energy number. Five public self-corrections demonstrate the discipline is real. The platform is already deployed across drug discovery, propulsion, battery materials, and catalysis — and every result that has survived audit has survived because it was tested, not because it was assumed.

Six domains, one audit standard

The same ensemble VQE diagnostic framework is applied across chemically distinct systems. Each domain has its own named platform and its own stage in the audit pipeline. Maturity is reported honestly per domain — Audited, Corrected, or Pending audit — not as a uniform claim of validation.

Prometheus Platform Audited

Drug discovery — metalloenzyme active sites

Independent statevector audit of the deposited ELSD metalloenzyme registry found that 12 of 13 systems pass the platform's own locked-gate criteria. The Cu²⁺ SOD apo/bound 9.2× σ ratio is numerically reproduced from deposited histories. One control (Cu_SOD_minimal_CuI) failed sector verification and the corresponding d⁹→d¹⁰ mechanistic claim was withdrawn in the corrected version.

Four published-state Zn²⁺ warhead chemotypes pass; six v1.0 Cu²⁺ SOD baseline systems pass with σ reproducing exactly from deposited histories. The corrected record documents what survived audit, what was withdrawn, and what remains to be independently verified.
12 of 13 systems pass deposited-statevector audit (May 2026)
View on Zenodo →
Catalysis Corrected

Nitrogen fixation — Fe₄N₂ reduced-model benchmark

A prior Fe₄N₂ redox-collapse / SRDS chemistry claim was withdrawn after sector enforcement, reference comparability, and active-space limitations were identified. The corrected record now serves as a sector-aware reduced-model VQE benchmark with corrected Sẑ penalties, sector-aware checkpointing, same-active-space exact references, and physical-root ROHF selection.

The original physical mechanism is not restored. What survives is a reduced-model methods recovery against an independent same-Hamiltonian exact reference. The correction also documents an anion ROHF multi-root failure mode as a citable methods finding.
v2.0 corrected record · methods benchmark, not chemistry discovery
View on Zenodo →
HELIOS Platform Pending audit

Energy storage — NMC811 battery cathodes

Ni-rich NMC811 cathodes degrade fastest near 50% state of charge. ELSD has been applied to map the electronic energy landscape under symmetry-breaking perturbation, with ensemble-VQE diagnostics reporting bifurcation and multi-basin structure across Ni, Co, Mn, Al, and Co+Al systems.

755 ensemble runs across the screening grid. Results are pending the same forensic audit standard applied to the five published Zenodo correction records. Application-page content will be brought into alignment as this audit completes.
755 runs ensemble screening · forensic audit in progress
Domain detail →
Propulsion & Defense Audited

Rocket propulsion — methane–nickel interface & hydrogen storage

Five-seed ensemble VQE audit of the Ni–CH₄ cooling-channel interface and AlH₃ hydrogen-storage model, with independent statevector sector verification and exact-reference Lanczos gap analysis. All 20 wavefunctions (4 systems × 5 seeds) land in the correct electronic sector. The 2.51× activation-vs-physisorption reproducibility ratio survives independent recomputation. AlH₃, previously scoped out due to a software coverage gap, has been rehabilitated by the v5 engine patch.

Lanczos exact-reference verification confirms the σ values are mathematically anchored to the in-sector ground state. All four systems classified ABOVE_ANSATZ_REACH (gaps 10–31 kcal/mol), correctly characterising the HEA ansatz limit. The 2.51× ratio is a reproducibility signal, not a physical-landscape claim — stated explicitly on the domain page.
20 / 20 sector-clean · Lanczos verified · DOI 10.5281/zenodo.20348697
Domain detail →
Electrocatalysis Pending audit

CO₂ reduction — iron porphyrin redox ladder

ELSD has been applied to the iron porphyrin redox ladder relevant to molecular CO₂ reduction electrocatalysis, from resting state through CO₂ binding. A Zenodo deposit documents the ensemble-VQE results across the four-phase campaign.

Three Rigid-Stability classifications and one Fe⁺ spin-escalation engineering constraint reported in the original record. Results are pending the same forensic audit standard applied to the five published Zenodo correction records.
Dataset live on Zenodo · forensic audit in progress
Dataset on Zenodo →
Electrocatalysis Pending audit

Green hydrogen — PEM electrolysis & storage materials

ELSD has been applied to PEM electrolysis cathode/anode materials (Pt HER, Ir OER) and to hydride-storage systems (MgH₂, TiH₂, NaAlH₄). The diagnostic is intended as an upstream quality check before catalyst or storage-material investment.

Six ensemble-VQE results across cathode, anode, and storage branches, including a repeatable Ir OER spin-escalation classification. Results are pending the same forensic audit standard applied to the five published Zenodo correction records.
6 systems applied · forensic audit in progress
Domain detail →
📄 Published audit record  ·  five Zenodo corrections, 2026
Self-correction discipline, in public, with reproducible audit tools

Five superseding Zenodo records document the platform's audit-grade self-correction posture. Each correction was produced by an independent verifier that shares no code with the proprietary engine; each names what survived audit, what was withdrawn, and which open items remain; each carries its verifier tool and primary evidence as part of the deposit so the audit is reproducible by any external reader.

Correction  ·  DOI: 10.5281/zenodo.20264767
Fe₄N₂ reduced-model VQE: sector-aware correction
Original redox-collapse / SRDS chemistry claim withdrawn. Sector-aware reduced-model methods recovery retained and benchmarked against independent exact references.
View on Zenodo →
Correction  ·  DOI: 10.5281/zenodo.20279079
LLZO solid-electrolyte: SCF root-multiplicity contamination
Seven of ten arms found SCF-reference contaminated. Three survivors retained; pressure-inversion and bifurcation interpretations withdrawn. SCF-multiplicity hazard documented as cross-record failure mode.
View on Zenodo →
Correction  ·  DOI: 10.5281/zenodo.20298089
Electron-sector integrity: scope extension to spin axis
Number-sector framework independently reproduced. Scope extended to spin axis (2Sẑ): number-only gate documented as necessary but not sufficient.
View on Zenodo →
Correction  ·  DOI: 10.5281/zenodo.20318424
Metalloenzyme registry: 12 of 13 systems pass
Independent audit of the drug-discovery metalloenzyme registry. 12/13 pass; 9.2× σ ratio reproduced exactly. One control failed sector verification; d⁹→d¹⁰ mechanistic row withdrawn.
View on Zenodo →
Correction  ·  DOI: 10.5281/zenodo.20348697
Propulsion re-audit: methane–nickel interface, five-seed verification
20/20 wavefunctions sector-clean. 2.51× ratio reproduces. Four claims narrowed or withdrawn. AlH₃ rehabilitated by v5 patch. Lanczos exact-reference confirms σ = gap_std identity across all four systems.
View on Zenodo →

Four electronic landscape regimes, one diagnostic principle

ELSD produces a regime classification, not just an energy value. The same four-class framework applies whether the system is a battery cathode, a metalloenzyme active site, a catalytic cluster, or a strongly correlated material — making the audit verdict comparable across domains.

How ELSD relates to DFT and other electronic-structure methods

DFT, wavefunction methods, and VQE answer different questions. ELSD is not a replacement for any established electronic-structure package. It is an audit layer: it tests whether a reduced-active-space VQE result is sector-clean, reference-comparable, reproducible across optimizer seeds, and stable under SCF rebuild — the dimensions on which conventional single-point convergence checks are silent.

Conventional VQE / single-point ELSD audit layer
Output Single converged energy Audit verdict + regime classification
Runs per system One 15–35 independent optimizer trajectories
Reliability signal Implicit — convergence assumed Explicit — σ, sector purity (N and 2Sẑ), dominant determinant, SCF-root stability
Multi-reference systems No built-in self-diagnostic Classifies whether multi-reference character is structured or pathological
Active-space adequacy Not assessed Flagged via controlled fragment-extension tests
Decision readiness Implicit — assumed from convergence Explicit — decision-grade classification or flagged constraint, reproducible by external reader
Rigid stability
Stable · decision-grade · reproducible

Single basin retained across all ensemble seeds, with SCF reference root-stable across rebuilds. The model is reliable enough to support downstream decision-making.

Zn CA2 inhibitor warheads (4 chemotypes) · v1 Cu SOD apo (audited) · LLZO P3/P4 survivors
Coherent open-shell
Broader · single-family · sector-clean

Multi-reference character present but well-structured. Ensemble converges within a single electronic family under sector enforcement. Results are reproducible and the model is usable with appropriate care.

Cu SOD water-bound (audited) · Cu_AOC3_minimal SSAO/VAP-1 (audited)
Multi-basin
Competing families · bifurcated · unstable

Two or more distinct electronic basins coexist, or the SCF reference jumps between roots. Results depend on starting conditions and should not be trusted without landscape diagnosis.

LLZO contaminated arms (7 of 10) · Fe₄N₂ anion ROHF spurious root (characterized)
Model pathology
Truncated · sector-escaped · not decision-grade

The active space is too truncated, the optimizer escapes the intended sector, or convergence reaches a wrong-sector state. Not safe to optimize against.

Cu_SOD_minimal_CuI (audited: escaped Cu⁺ d¹⁰ → Cu²⁺ d⁹)

How ELSD is built

Four layers connecting industry problems to validated quantum computation. The diagnostic engine has been through internal self-audit (May 2026): 34/34 regression tests passing, naming corrections applied in source, five cross-record failure modes encoded as tripwires.

Domain ELSD Platform — six application domains, one audit standard 6 domains · 2 audited, 1 corrected,
3 audit-pending
Diagnostic Ensemble VQE + ELS classification (σ, basin count, sector purity N and 2Sẑ, SCF-root stability, Lanczos exact-reference gap analysis) 15–35 seeds / condition
5 published Zenodo corrections
Engine self-audit (May 2026)
Simulation GPU-native VQE engine — HEA-style ansatz paths (legacy UCCSD flag audited & deprecated in source), Pauli-filtered, sector-penalty enforced, v5 Sz auto-registration NVIDIA L40S / RTX
20-qubit reduced-active-space
operating point
Quantum QuantaCore™ — hardware-validation architecture · Y⊗Z orthogonal protection · OrthoTiles™ IBM Quantum hardware-tested
1200+ device runs
Patent pending

Partner with us on electronically difficult targets

We work with battery developers, pharmaceutical R&D teams, catalyst designers, and materials scientists who need reliable electronic landscape classification before committing resources to synthesis, screening, or clinical-stage decisions.

01
Target and model audit studies

Submit a candidate composition, active-site model, or dopant strategy. We return a full ELSD audit report: sector purity (N and 2Sẑ), SCF-root stability, seed-ensemble σ, regime classification, and design recommendations. No source code shared.

02
Platform evaluation partnerships

A defined evaluation campaign against your internal targets, with results benchmarked against your existing computational workflows. Designed for R&D teams assessing ELSD fit before a broader deployment decision.

03
Enterprise deployment

Full platform deployed in your environment, running across your candidate pipeline on your hardware. For organisations screening many compositions or targets on an ongoing basis.

04
Strategic licensing conversations

For organisations interested in platform rights, IP integration, or long-term strategic access. Details available under mutual NDA.

⚠ Site update in progress

This homepage has been updated (May 2026) to reflect Quantum-Clarity's audit-grade diagnostic-platform identity following five published Zenodo corrections and an internal engine self-audit. The propulsion reliability page has been fully rewritten (v7) and is aligned with the audit-grade framing, including Lanczos exact-reference verification results. Several other per-domain pages are being rewritten and may temporarily contain claims at a confidence level superseded by the published corrections linked on this page. Where any discrepancy appears, the Zenodo correction records are the authoritative source. Updates to the remaining pages are in progress and will be applied as each domain's audit completes.

Contact Us

Interested in working together? Fill out some info and we will be in touch shortly. We can’t wait to hear from you!