Domain Packs

Formulations & CASE

Mixture design

Constrained mixture design with simplex-lattice rows (sum to 1) plus a seeded bounded-simplex sampler respecting component bounds — reproducible per seed.

Hansen solubility

HSP group-contribution estimation (d, p, h) from SMILES via RDKit substructure counts — a screening approximation (±2–3 MPa^0.5). Ra distance and Relative Energy Difference (RED = Ra/cutoff; RED ≤ 1 = good match) drive solvent screening.

BOM mixing rules

Mixture density (harmonic rule), viscosity and properties from weight fractions with a ±5% uncertainty band; the reformulator searches alternatives meeting property targets. Engine cards label results honestly: mixture/HSP are COMPUTED, reformulation is PREDICTED.

Go from reading to running

The quickstart tutorial walks the first synthesis, evaluation and prediction on a demo dataset — each step matches a real platform page.