A Product of Armadillo Labs

QuantArm

Machine-learned interatomic potentials predict material properties and atomic interactions at near-DFT accuracy, paired with a multi-stage screening classifier. Screen candidate superconductors in hours, not years.

Evaluating for your research group or company? Use the contact form → to request a trial.

Member NVIDIA Inception

What It Does

Point it at a material. Get a score. Know whether it's worth taking into the lab before you burn a week of DFPT.

Capabilities

Machine-Learned Phonon Emulator

Machine-learned interatomic potentials (MLIPs) designed to predict material properties and atomic interactions with high accuracy. Generates phonon spectra in seconds at near-DFT accuracy.

Multi-Stage Screening Classifier

A four-stage screening cascade — structural, phonon, coupling, and coherence gates applied in sequence — that identifies the spectral signatures associated with high critical temperatures. Each stage filters false positives and passes promising candidates deeper.

First-Principles Tc Estimation

Converts screening cascade scores into an estimated transition temperature using validated alpha2F functionals. Grounded in the same physics you'd compute by hand — just thousands of times faster.

Materials Project Integration

Query any MP material by ID or formula. Run the screening engine without ever leaving the browser. Export per-material JSON and summary CSVs.

Family-Scale Batches

Screen entire structural families — cuprates, bismuthates, nickelates, iridates — in a single batch. Results ranked by screening cascade score and walltime.

Validated Against Experiment

The screening cascade has been calibrated against known superconductors across six families. When the engine ranks a material high, the lab data agrees.

Member of

NVIDIA
Inception

Built on NVIDIA

The machine-learned interatomic potential at the heart of Quantum Armadillo is trained and served on NVIDIA accelerated compute.

GPU-accelerated inference means single-material screens run in seconds, and full structural families finish overnight.

Calibration

Validation metrics will be published with the first production release.

Request preview access →

QuantArm Research — $250/mo

Full engine access. Submit any crystal, get ranked predictions in minutes. Download per-gate pass/fail matrices, phonon participation data, and full audit trails. Built-in AI research assistant explains your results and suggests next steps.

Evaluating for your research group or company? Use the contact form → to request a trial.

Stop waiting weeks for DFPT. Start screening candidates this afternoon.

Subscribe — $250/mo →