Equations the universe
was hiding.
Logic Engine derives provable scientific laws from raw data — no hypotheses, no neural networks. Pure symbolic discovery that extrapolates where black-box models fail.
Published Research
Discoveries
Results the Logic Engine produced that were not previously known — derived from raw public data with no human hypothesis.
Scaffold-Dependent Quadratic Laws Govern FAAH/MAGL Selectivity: A First-Principles Derivation of the Non-Addictive Social Anxiety Treatment Window
Social anxiety disorder affects 12% of the population yet available treatments remain inadequate. Using symbolic equation discovery on 6,964 binding measurements, we derive the structural rules governing FAAH/MAGL selectivity and identify the logP 3.0–4.2 therapeutic window for non-addictive anxiolysis — validated with 87% error reduction over neural network baselines.
Tissue-Specific Decay Laws Govern Mammalian Aging: A First-Principles Derivation of Seven Distinct ODE Regimes from Single-Cell Transcriptomics
Applying an automated symmetry-breaking pipeline to the Tabula Muris Senis atlas (110,824 cells, 23 tissues), we show that aging is not governed by a universal clock. Seven distinct ODE families govern tissue-specific decay — with SIRT6 identified as the single molecule that shifts tissues on both axes of the aging phase diagram.
Validations
Known results the Logic Engine re-derived independently — confirming it finds the right answer when the answer is already known.
How It Works
Science, derived.
The Logic Engine doesn't fit curves. It finds the underlying differential equation whose integral describes the data — producing laws that extrapolate correctly where neural networks fail.
Ingests Data
Raw IC₅₀ measurements, binding assays, and molecular descriptors from public databases like ChEMBL.
Derives Laws
Computes derivative invariants to identify which ODE the data satisfies, then integrates it into a closed-form symbolic law.
Validates OOD
Tests the derived law on out-of-distribution data it never saw. ΔG = 0.869 means 87% error reduction over MLP baselines.
Produces Manuscripts
Generates peer-reviewed quality papers with full methodology, figures, and reproducible evidence reports.
Early Access
Be first to know
Logic Engine is developing new discoveries across pharmacology, materials science, and beyond. Join the waitlist for early access and research updates.
You're on the list. We'll be in touch.