SYMBOLIC REGRESSION · APACHE-2.0
MilesCranmer/PySR
PySR & SymbolicRegression.jl: Evolutionary Pareto Equation Discovery
f* = argmin [ (N/2) ln(MSE) + λ·C_tree(f) ]
Multi-population evolutionary symbolic regression that discovers closed-form physical laws and constitutive fluid closures from noisy empirical trajectories.
Scite Verified (2026-11-08): 422 citing · 2 supporting · 426 mentioning · 0 retractions
github.com/MilesCranmer/PySR ·
DOI 10.48550/arXiv.2305.01582 ·
[Scite ↗] ·
Folio P.03 →
SPARSE DYNAMICS · MIT
dynamicslab/pysindy
PySINDy: Sparse Identification of Nonlinear Dynamical PDEs
u_t = Θ(u, ∇u, ∇²u) ξ, min ||u_t - Θξ||₂² + λ||ξ||₀
Recovers parsimonious governing ODEs and PDEs (Navier–Stokes, Kuramoto, reaction-diffusion) via Sequentially Thresholded Least Squares (STLSQ) and weak-form integration.
Scite Verified (2026-11-08): 307 citing · 1 supporting · 204 mentioning · 0 retractions
github.com/dynamicslab/pysindy ·
DOI 10.21105/joss.02104 ·
[Scite ↗] ·
Folio P.12 →
QUANTUM POTENTIALS · MIT
facebookresearch/fairchem
FAIRChem & UMA: Universal Equivariant Interatomic Neural Potentials
F_i = -∇_{r_i} E_UMA(r_1,…,r_N), MAE(E) ≤ 0.02 eV
Equivariant graph neural force fields trained on 100M+ DFT calculations, delivering ab-initio quantum forces at 100,000x lower computational cost.
Scite Verified (2026-11-08): 1,009 citing · 5 supporting · 1,035 mentioning · 0 retractions
github.com/facebookresearch/fairchem ·
DOI 10.1021/acscatal.0c04525 ·
[Scite ↗] ·
Folio P.01 →
SPECTRAL PDE SOLVER · GPL-3.0
DedalusProject/dedalus
Dedalus: Flexible Spectral PDE Solver for Fluid & Continuum Physics
∂_t M·X + L·X = F(X) (Chebyshev–Fourier–Spherical)
Solves symbolically specified initial, boundary, and eigenvalue PDEs using sparse Chebyshev, Fourier, and spherical harmonic spectral methods.
Scite Verified (2026-11-08): 717 citing · 2 supporting · 519 mentioning · 0 retractions
github.com/DedalusProject/dedalus ·
DOI 10.1103/PhysRevResearch.2.023068 ·
[Scite ↗] ·
Folio P.12 →
DIFFERENTIABLE MD · APACHE-2.0
google/jax-md
JAX-MD: Accelerated Differentiable Molecular Dynamics & Statistical Physics
dR = -∇_R U_θ(R) dt + √(2 k_B T dt) dW_t
End-to-end automatic differentiation through Langevin, Nose–Hoover, and Brownian molecular dynamics trajectories for inverse material design.
Scite Verified (2026-11-08): 195 citing · 0 supporting · 151 mentioning · 0 retractions
github.com/google/jax-md ·
DOI 10.1088/1742-5468/ac3ae9 ·
[Scite ↗] ·
Folio P.03 →
FINITE ELEMENT · LGPL-3.0
FEniCS/dolfinx
FEniCSx / DOLFINx: Variational PDE & Topology Optimization Engine
a(u, v) = L(v) ∀v ∈ V_h, E(ρ) = E_min + ρ^p (E_0 - E_min)
Compiles Unified Form Language (UFL) weak variational forms into high-performance parallel C++ finite-element kernels for elasticity and SIMP topology optimization.
Scite Verified (2026-11-08): 285 citing · 0 supporting · 107 mentioning · 0 retractions
github.com/FEniCS/dolfinx ·
DOI 10.1145/3524456 ·
[Scite ↗] ·
Folio E.11 →
TENSOR NETWORKS · APACHE-2.0
itensor/ITensors.jl
ITensors.jl: DMRG, Matrix Product States & Quantum Tensor Networks
|Ψ_MPS⟩ = ∑ A^{s_1}_1 … A^{s_N}_N |s⟩, S_vN ≤ ln(χ)
Automatic index-contracting tensor networks for Density Matrix Renormalization Group (DMRG), time-evolving block decimation (TEBD), and strongly correlated quantum matter.
Scite Verified (2026-11-08): 1,506 citing · 5 supporting · 531 mentioning · 0 retractions
github.com/itensor/ITensors.jl ·
DOI 10.21468/SciPostPhysCodeb.4 ·
[Scite ↗] ·
Folio P.11 →
NEURAL QUANTUM · APACHE-2.0
netket/netket
NetKet & QuTiP: Neural Quantum States & Lindblad Open Quantum Dynamics
E_0 = min_θ ⟨ψ_θ|H|ψ_θ⟩ / ⟨ψ_θ|ψ_θ⟩, dρ/dt = -i[H,ρ] + D[L](ρ)
Variational Monte Carlo (VMC) with neural-network quantum wavefunctions on JAX paired with QuTiP master-equation solvers for open quantum systems.
Scite Verified (2026-11-08): 211 citing · 0 supporting · 99 mentioning · 0 retractions
github.com/netket/netket ·
DOI 10.21468/SciPostPhysCodeb.7 ·
[Scite ↗] ·
Folio P.11 →
QUANTUM QEC & DFT · APACHE-2.0
quantumlib/Stim + pyscf/pyscf
Stim & PySCF: Stabilizer Surface-Code QEC & Ab-Initio Electronic Structure
p_L ≈ 0.1 (p / p_th)^{(d+1)/2}, H_el |Ψ⟩ = E |Ψ⟩
Megahertz stabilizer circuit sampling in Stim paired with PySCF coupled-cluster and CASSCF electronic structure integrals for fault-tolerant quantum chemistry.
Scite Verified (2026-11-08): 818 citing · 4 supporting · 731 mentioning · 0 retractions
quantumlib/Stim ·
DOI 10.22331/q-2021-07-06-497 ·
[Scite ↗] ·
Folio P.11 →
LEAN 4 PHYSICS · APACHE-2.0
HEPLean/PhysLean
PhysLean (HepLean): Formalized Classical, Quantum & Gauge Physics in Lean 4
∑ d_f Y_f = 0 ∧ ∑ d_f Y_f³ = 0, d ⋆ F = J
1,850+ machine-checked theorems in Lean 4 covering Lagrangian mechanics, Maxwell electrodynamics, Lorentz groups, Wick's theorem, and Standard Model anomaly cancellation.
Scite Verified (2026-11-08): 7 citing · 0 supporting · 6 mentioning · 0 retractions
github.com/HEPLean/PhysLean ·
DOI 10.1016/j.cpc.2024.109457 ·
[Scite ↗] ·
Folio M.01 →
FORMAL PROOF · APACHE-2.0
anthropics/formal-math + mathlib4
Anthropic Formal-Math, Fermat's Last Theorem & Leanprover Mathlib4
a^n + b^n ≠ c^n (n ≥ 3), N(σ,T) ≤ C·T^{(30/13)(1-σ)+ε}
Zero-sorry Lean 4 formalizations of Fermat's Last Theorem, Riemann XiPrime zero-density bounds, and sub-exponential 3-SAT refutations built atop Mathlib4.
Scite Verified (2026-11-08): 458 citing · 2 supporting · 218 mentioning · 0 retractions
anthropics/formal-math ·
DOI 10.1145/3372885.3373824 ·
[Scite ↗] ·
Folio M.11 →
GEOMETRIC STATICS & OT · MIT
compas-dev/compas + PythonOT/POT
COMPAS 3D Polyhedral Graphic Statics & Python Optimal Transport (POT)
∑ A_f n_f = 0 (Maxwell–Rankine), W_2²(μ,ν) = inf ∫ |x - T(x)|² dμ
Reciprocal polyhedral force diagrams for compression-only vaults in COMPAS paired with Wasserstein optimal transport and Sinkhorn barycenters in POT and teorth/equational_theories.
Scite Verified (2026-11-08): 93 citing · 0 supporting · 55 mentioning · 0 retractions
compas-dev/compas ·
DOI 10.1016/j.cad.2015.01.006 ·
[Scite ↗] ·
Folio A.11 → ·
Folio M.09 →
SMART CITATION POLARITY · SCITE
scite.ai · scite-batch + scite-mcp
Scite Smart Citations & Section-Polarity Engine: Supporting vs. Contrasting Evidence
P_scite = (N_supp − w_c·N_cont) / (N_supp + N_ment + N_cont + 1), R_retract = 0
Classifies in-text citation statements across Introduction, Methods, Results, and Discussion into supporting, mentioning, or contrasting evidence while screening all 48 Folios and 15 Radar engines for retractions.
Scite Verified (2026-11-08): 182 citing · 1 supporting · 82 mentioning · 0 retractions
scite.ai ·
DOI 10.1162/qss_a_00146 ·
[Scite ↗] ·
Folio M.01 →
DOI & RETRACTION WATCH · MIT
CrossRef/rest-api-doc · check_folio_citations.py
Crossref & Retraction Watch Deterministic Citation Gate
∀ d ∈ DOIs: HTTP(d) = 200 ∧ Retracted(d) = False ∧ Sim(T_label, T_CR) ≥ 0.5
Deterministic registry verifier (tools/check_folio_citations.py) that validates every primary and inline DOI, publication year, journal container, and Scite tally against Crossref and Retraction Watch.
Scite Verified (2026-11-08): 234 citing · 2 supporting · 126 mentioning · 0 retractions
CrossRef/rest-api-doc ·
DOI 10.1162/qss_a_00022 ·
[Scite ↗] ·
Folio E.01 →
OPENALEX & EVIDENCE LOOP · CC0-1.0
ourresearch/openalex + cf-evidence-loop
OpenAlex, Unpaywall & cf-evidence-loop Deterministic Paper-Trail
E(q) = {source_id, publisher_date, accessed_utc, fields, limitations ≠ ∅}
Zero-LLM catalog-governed CLI (evidence_cli.py paper-trail, cite-check --scite-bulk) uniting OpenAlex 250M+ works, Unpaywall open-access licensing, and Semantic Scholar citation graphs with mandatory limitation records.
Scite Verified (2026-11-08): 1,360 citing · 42 supporting · 1,096 mentioning · 0 retractions
senseiewok/cf-skills ·
DOI 10.7717/peerj.4375 ·
[Scite ↗] ·
Folio P.01 →