feynlag

Tree-level Feynman rules from Beyond-Standard-Model Lagrangians, in pure SymPy.

You declare particle fields with their gauge and discrete-symmetry representations, write the Lagrangian explicitly with library building blocks (Dmu, dag, Bilinear), and feynlag takes it from there:

• gauge / discrete invariance and hermiticity checks,
• electroweak symmetry breaking: VEV expansion, tadpole conditions,
• mass-matrix extraction and diagonalization (orthogonal, unitary, SVD for Dirac fermions, Takagi for Majorana),
• rotation from the weak basis to the physical (mass) basis,
• vertex extraction (SSS, SSSS, VSS, VVS, VVSS, VVV, VVVV, FFS, FFV, plus four-fermion and Majorana/Weinberg operators) with derivative couplings taken to momentum space,
• model consistency in one call (Model.validate()): invariance, hermiticity, anomaly cancellation, charge conservation,
• export: LaTeX vertex tables and UFO (MadGraph et al.),
• phenomenology (feynlag.pheno): decay widths and branching ratios (tree-level 1→2, off-shell h→WW*/ZZ*, loop-induced h→gg/γγ/Zγ) and tree-level 2→2 cross sections with forward–backward asymmetries.

Parameters are split into external (fixed by experiment, e.g. v, m_h, g) and internal (derived: tadpole solutions, mixing angles, inverted quartics), forming a dependency chain that closes the UFO parameter card.

Documentation

Full docs, including an Algorithms Manual deriving the physics and design of every pipeline stage (invariance checking, EWSB/tadpoles, mass matrices, diagonalization, vertex extraction, export), tutorial notebooks, an examples gallery, and the API reference: https://moiseszeleny.github.io/feynlag/

Install

pip install feynlag            # add [numeric] for SciPy-backed integration

For development:

pip install -e .[dev]
pytest

Quick tour

import sympy as sp
from feynlag import (ExternalParameter, InternalParameter, SU2, U1, Scalar,
                     Lagrangian, Model, Dmu, dag)

gw  = ExternalParameter("gw", 0.6535, positive=True)
g1  = ExternalParameter("g1", 0.3580, positive=True)
SU2L, U1Y = SU2("SU2L", coupling=gw), U1("U1Y", coupling=g1)

v   = ExternalParameter("v", 246.0, positive=True, unit_dim=1)
lam = ExternalParameter("lam", 0.129)
mu2 = InternalParameter("mu2", unit_dim=2)      # defined by the tadpole

H = Scalar("H", reps={SU2L: 2, U1Y: sp.Rational(1, 2)},
           component_names=["Gp", "H0"])
H.expand_vev({H.components[1]: v})              # H0 -> (v + h + i G0)/√2

HdH = (dag(H) * H.mat)[0]
DH  = Dmu(H)
L = Lagrangian()
L.add((dag(DH) * DH)[0], sector="kinetic")
L.add(mu2.s * HdH - lam.s * HdH**2, sector="potential")

m = Model("SM", gauge_groups=[SU2L, U1Y],
          fields=[H, SU2L.bosons("W"), U1Y.bosons("B")],
          parameters=[gw, g1, v, lam, mu2], lagrangian=L)

m.check_invariance()          # gauge invariance, hermiticity, dim ≤ 4
m.solve_tadpoles([mu2])       # {mu2: lam v²}, registered as internal

h = sp.Symbol("H0_r", real=True)
m.mass_matrix([h])            # Matrix([[2 lam v²]])
m.feynman_rules([h])          # {(h,h,h): -6i lam v, (h,h,h,h): -6i lam}

See examples/ for full runs: sm_scalar_gauge.py (complete SM: Higgs + electroweak gauge + leptons + quark/QCD sector), sm_vll.py (SM + a vector-like lepton doublet, biunitary mass-matrix diagonalization), sm_u1x.py (SM × U(1)_X with a Z′, symbolic charges, chained rotations), thdm.py (2HDM with the α rotation), thdm_s3.py (3HDM+S₃, where the tadpole conditions force the √3 vacuum alignment), sm_ckm.py (CKM quark mixing), fermi_theory.py (four-fermion muon decay), sm_weinberg.py and sm_seesaw.py (Majorana neutrino masses), sm_decays.py and sm_higgs_decays.py (widths and the full Higgs branching-ratio table), and ee_to_ff.py (2→2 scattering). The docs site walks these models stage by stage in ten executed tutorial notebooks.

Validation

The test suite pins the physics, not just the code (dual verification: symbolic difference and random-point numeric checks):

• SM Higgs: μ² = λv², m_h² = 2λv², h³ = −3i m_h²/v, h⁴ = −3i m_h²/v²
• SM gauge: m_W = gv/2, Weinberg rotation, hWW = i g m_W g^{μν}, γW⁺W⁻ = e, ZW⁺W⁻ = g cosθ_W, scalar-QED Goldstone vertices
• SM leptons: hℓℓ = −i m_ℓ/v, Wℓν = i g/√2 γ^μ P_L, Z couplings ∝ T³ − Q sin²θ_W
• 2HDM: tadpoles, all three mass matrices and rotation angles vs the Gunion–Haber/Branco expressions
• 3HDM+S₃: invariant potential from the library's CG products; the tadpole system forces the √3 alignment
• UFO: generated model imports cleanly; parameters resolve in dependency order; hWW coupling pinned numerically

Status / roadmap

feynlag 0.1 is a beta: the tree-level pipeline above is complete and tested (400+ tests pinning physical results), and the exported SM UFO is cross-checked against MadGraph (e+e-→μ+μ- and the gauge-cancelling e+e-→W+W- reproduce the stock sm cross sections to MC precision, and a four-fermion UFO reproduces the muon width — see docs/benchmark.md).

Known limitations (planned, see docs/roadmap.md):

• no R_ξ gauge fixing or ghosts (Goldstone bosons are kept, but no gauge-fixing terms, ghost vertices or ξ dependence);
• Majorana vertices are symbolic-only, not yet exported to UFO;
• 2→2 scattering handles single-diagram processes only (no interference yet), and VVV decays are not implemented;
• no NLO / UFO 2.0 extensions.

Unsupported cases raise NotImplementedError rather than returning a plausible-looking wrong answer.

License

MIT — see LICENSE.