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.
corrections (2026)
completed
operating point
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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 |
Single basin retained across all ensemble seeds, with SCF reference root-stable across rebuilds. The model is reliable enough to support downstream decision-making.
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.
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.
The active space is too truncated, the optimizer escapes the intended sector, or convergence reaches a wrong-sector state. Not safe to optimize against.
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.
3 audit-pending
5 published Zenodo corrections
Engine self-audit (May 2026)
20-qubit reduced-active-space
operating point
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.
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.
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.
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.
For organisations interested in platform rights, IP integration, or long-term strategic access. Details available under mutual NDA.
Bring us your electronically difficult target.
Whether it is a cathode composition, a metalloenzyme active site, a catalytic cluster, or a strongly correlated material — if conventional workflows are not giving you a reliable picture, ELSD is designed for exactly that problem.
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!