Search all 707 registered Python modules by subject, filename, claim or source section. This catalogue records source presence and links. It does not report a new execution or a passing result for every module.
707 matching modules · showing 361–380
Mmst seam collar
v366_mmst_seam_collar.py · core
Read the registered purpose, outcome and evidence links
Direction 7: the seam collar is IN MMST's free-lattice-fermion scaling-limit class, so the chiral scaling limit is a chiral CFT of central charge 8 pinned to (E8)_1 -- reducing SEAM.EQUIV.01 to the single S3 lattice-realisation input. Osborne-Stottmeister (arXiv:2107.13834, CMP 398 (2023) 219; legacy internal token MMST, on the MMST/Morinelli-Morsella-Stottmeister-Tanimoto OAR framework): free lattice fermions (CAR quasi-free) -> chiral-CFT massless scaling limit with correct c, range rank<=c<=D. [E] class arithmetic: D=dim S^+=2^(g_car-1)=16 Majoranas, c=g_car+N_fam=8, rank E8=8 => rank<=c<=D is 8<=8<=16 (in MMST range). [E] gapped (Delta=6 ln(3/2)>0, derived v302). [C] quasi-free CAR class (v155/v160). [E] chirality c_-=D/2=8 != 0 (non-gappable anomalous edge, v356 S4). [C] the MMST scaling-limit theorem applies (in-class & gapped & chiral & in-range) => chiral CFT c=8 (v336). [E] target pinned to (E8)_1 by c=8 + the order-4 mu4 clock (det K=1 vs SO(16) det K=4, v351). [O] residual S3: the collar realised as a genuine lattice chiral free-fermion invertible phase (one-sidedness/Flat-Away, v297/v356). Verifies MMST in-class; ADVANCES SEAM.EQUIV.01, does NOT close it.
Read the original Python source
Grav spin2 unitarity
v370_grav_spin2_unitarity.py · frontier
Read the registered purpose, outcome and evidence links
Track 2: the Barnes-Rivers spin decomposition of the graviton propagator -- perturbative graviton unitarity SECTOR-BY-SECTOR (extends v304's scalar pole algebra to the tensor spin structure + combines with v334's spin-0 conformal mode). [E] projector completeness: tr P2=(d-2)(d+1)/2, tr P1=d-1, tr P0s=tr P0w=1 sum to d(d+1)/2; the d=4 split is 5+3+1+1=10. [E] EH structure: spin-2 coefficient +1, spin-0 scalar -1/(d-2)=-1/2. [E] the Stelle ghost is a SPIN-2 state (a local 1+p^2/M^2 dressing puts the negative-residue pole at p^2=-M^2). [E] the entire KMS form factor e^{p^2/M^2} (v304/v259) makes spin-2 ghost-free (only the p^2=0 pole). [C] the wrong-sign spin-0 conformal mode (c_conf(4)=-3/2, v332) is cured by the GHP contour (v334) -- a different mechanism. [C] => perturbative graviton unitarity sector-by-sector, conditional on v304's entire-analyticity assumption. [O] perturbative-only + the open analyticity assumption; NOT the non-perturbative measure (QG.AMB.01). [O] decline: tr P2=5 is dimension counting, NOT g_car (v354/v355).
Read the original Python source
Seam s3 lattice
v367_seam_s3_lattice.py · core
Read the registered purpose, outcome and evidence links
Track 1: the EXPLICIT lattice realisation of the S3 input -- a concrete gapped chiral-Majorana (p+ip) lattice model whose edge is the 16-Majorana SO(16)_1 carrier, mu4-condensed to (E8)_1; turns the abstract S3 residual (v356/v366) into a RUNNABLE model with a numerically computed topological invariant. [E] bulk gapped (band gap 2>0 at M=1). [E] non-trivial Chern number (Fukui-Hatsugai-Suzuki: |C|=1 at M=1, C=0 at M=3 trivial control). [E] c_-=8: each chiral Majorana edge carries c_-=C/2=1/2, N_Maj=dim S^+=2^(g_car-1)=16 copies => c_-=16/2=8=g_car+N_fam. [E] edge K-matrix SO(16)_1 (det Cartan D8=4) -> (E8)_1 (det Cartan E8=1) by the order-4 mu4 clock (v351/v369). [C] the mu4 condensation SO(16)_1 -> (E8)_1 (v154/v351). [O] residual: the continuum scaling-limit existence (MMST v336). Reduces S3 to a runnable lattice model; does NOT close SEAM.EQUIV.01.
Read the original Python source
Seam s3 inflow
v368_seam_s3_inflow.py · core
Read the registered purpose, outcome and evidence links
Track 1: the bulk-edge / anomaly-inflow leg (S4 of the v356 chain) made CONCRETE on the v367 lattice model -- a finite STRIP of the gapped p+ip model has a realised, non-gappable chiral edge, so edge EXISTENCE follows from the invertible bulk (anomaly inflow), NOT a separate assumption. [E] topo in-gap edge states (M=1 strip min_kx|E|~0) vs trivial clean gap (M=3, neg control); [E] the near-zero state is edge-localised (weight within 4 sites of a boundary, edge>>bulk); [E] bulk-edge correspondence (edge branch count = bulk Chern |C|=1) => 16-Majorana edge, c_-=8. [C] anomaly inflow: a c_-!=0 invertible bulk cannot have a gappable edge => discharges the v336/v351 'does the edge exist?' residual on the lattice. [O] residual: the continuum scaling limit (MMST v336). Establishes edge EXISTENCE on the lattice; does NOT close SEAM.EQUIV.01.
Read the original Python source
Ftransfer pole
v371_ftransfer_pole.py · frontier
Read the registered purpose, outcome and evidence links
Track 3 (F_transfer): the Koide source->pole transfer as a typed solver -- promotes the experiments/ftransfer solve into the ledger with a kill test (does NOT re-prove 53/54 v183 or Q*=2/3 v82/v101). [E] Q_pole=Koide(PDG poles)=2/3 to 3e-6, Q_src(phi0-ladder)=0.66446 (0.33% below). [E] QED runs the wrong way: dQ/deps>0 pushes Q ABOVE 2/3; the eps mapping Q_pole->Q_src is NEGATIVE while QED gives eps=+(3/2)(alpha/pi)>0 -- wrong sign for any alpha in [1/137,1/128]. [C] verdict: F_pole is NOT standard QED running; the transfer is the non-QED 53/54 operator / (2/3)^6 Moebius generator (v183/v82). Stays [C], plain running excluded as the kill.
Read the original Python source
Ftransfer boltzmann
v372_ftransfer_boltzmann.py · frontier
Read the registered purpose, outcome and evidence links
Track 3 (F_transfer): eta_B from the INTEGRATED Buchmueller-Di Bari-Pluemacher Boltzmann network at the TFPT-frozen heavy scale -- promotes the experiments/ftransfer solve, replacing the v326 toy washout. Frozen M1=M_scal phi0^2/A_Lambda (~8.65e9 GeV, v212), m~1=m3/A_Lambda=5 meV, delta_CP=4pi/3. [E] integrated efficiency kappa_f~0.09 (strong washout K~4.6) agrees with the BDP fit to ~25%. [C] eta_B(ODE)~6.5e-10 at the frozen M1 lands within a factor ~1.1 of the observed 6.1e-10, NO M_R dial; the M1 reproducing eta_B_obs is within a factor ~1.1 of the frozen one. [X] kill: eta_B outside [obs/3,3obs] for all seesaw-consistent M_R would drop the leptogenesis branch. Stays [C] (a transfer target, not a compiler power).
Read the original Python source
Ftransfer relic
v373_ftransfer_relic.py · frontier
Read the registered purpose, outcome and evidence links
Track 3 (F_transfer): the axion relic from the FINITE-T misalignment ODE -- promotes the experiments/ftransfer solve and DECIDES the two TFPT angle branches, replacing the v326 adiabatic toy. f_a=M_scal/128~2.4e11 GeV, m_a~23.8 ueV, lattice chi(T) n=8.16. [I] candidate scales fixed by the determinant-line atoms (128=2 dim S^+ |mu4|). [C] the frozen SPINE angle 3pi/5=108deg=pi N_fam/g_car gives Omega_a h^2~0.12 in [0.08,0.16] (lands on Omega_DM untuned) -- the robust dominant-DM branch. [X] the HILLTOP 170deg over-produces (~5.4x) -- disfavoured/fine-tuned. [X] kill: a finite-T solve robustly outside [0.08,0.16] for the spine across chi(T)/g_* drops the dominant axion-DM branch. Stays [C]/[X].
Read the original Python source
Ftransfer qcd
v374_ftransfer_qcd.py · frontier
Read the registered purpose, outcome and evidence links
Track 3 (F_transfer): the m_p/m_e transfer as a typed uncertainty-budget + kill test -- the falsifiability companion to v262 (the precise 2-loop point), replacing the v326 toy (which hardcodes 1836.15). [E] b3=-(11-2 n_f/3)=-7 (n_f=6 carrier). [N] two declared externals (v339): alpha_s(M_Z)=0.1179+-0.0009 and C_p=2.83+-0.15, each ~5.3%. [C] m_p/m_e central ~1824 (v262) with a propagated +-7.5% band [~1690,~1960] CONTAINING the observed 1836.15 -- no free dial. [X] kill: a future lattice C_p+-0.03 + tighter alpha_s shrink the band to ~+-2%; a band then excluding 1836.15 drops the QCD-matching route (b3=-7 + the SU(9)-integer firewall v262 untouched). Distinct from v262's point; stays [C].
Read the original Python source
Observatory registry
v375_observatory_registry.py · frontier
Read the registered purpose, outcome and evidence links
Track 4: the prediction OBSERVATORY -- a status-typed CI over the frozen prediction registry (freeze_file.csv) that makes the falsifiability surface machine-checkable (no new physics). [E] falsifiability complete: every registry row carries a non-empty kill criterion (no orphan claims). [E] headline values re-derive from {phi0}: sin^2 theta12=1/3-phi0/2, sin^2 theta13=phi0 e^{-5/6}, beta_rad=phi0/(4pi). [N] live scorecard: theta12 vs JUNO first run (0.3092+-0.0087)=0.28 sigma (compatible); theta13 vs NuFIT 6.0 NO (0.02195+-0.00058)=2.0 sigma (the flagged most-tensioned core prediction); beta_rad vs ACT DR6 (0.215+-0.074)=0.37 sigma (hint); r=12/N*^2~0.004<BK18 (compatible). [E] registry integrity: the single machine-readable falsifiability surface. Tooling for the closed/open pieces.
Read the original Python source
Seam s3 centralcharge
v376_seam_s3_centralcharge.py · core
Read the registered purpose, outcome and evidence links
S3 closure stack: the central charge c=8 extracted NUMERICALLY from the lattice via Calabrese-Cardy finite-size entanglement scaling of the critical free-fermion content (the quantitative core of the MMST scaling-limit claim, on OUR model). [E] one critical complex mode gives c=1.0000 (Peschel correlation-matrix EE); [E] the collar is 16 Majoranas = 8 complex modes (v367), c additive => c=8=g_car+N_fam; [E] chiral c_-=8 (Chern |C|=1 per Majorana v367, non-gappable edge v368). [O] the specific net (E8)_1 is v377/v378; the continuum existence is the cited MMST theorem (v336). Numerical c=8 from the lattice; does NOT close SEAM.EQUIV.01.
Read the original Python source
Seam s3 e8character
v377_seam_s3_e8character.py · core
Read the registered purpose, outcome and evidence links
S3 closure stack: the chiral c=8 content is the SPECIFIC net (E8)_1, via the exact affine character q-series + the mu4/GSO promotion. [E] chi = E4/eta^8 = q^{-1/3}(1+248q+4124q^2+...): leading -c/24=-1/3 => c=8; level-1 = 248 = dim E8 (one primary). [E] mu4/GSO promotion SO(16)_1 -> (E8)_1: 248=240+8=(E8 roots)+(Cartan), 240=dim S^+*g_car*N_fam=16*5*3; 248=120+128=dim SO(16)+spinor. [E] discriminator: (E8)_1 has 1 primary/248 currents vs SO(16)_1 4 primaries/120. [O] continuum existence (v336); mu4 projection = seam one-sidedness (S3). Pins the target net; does NOT close SEAM.EQUIV.01.
Read the original Python source
Seam s3 modular
v378_seam_s3_modular.py · core
Read the registered purpose, outcome and evidence links
S3 closure stack: the GENUS-1 discriminator -- torus ground-state degeneracy = 1 (holomorphic = (E8)_1), the 'hard part' v344 flagged (genus-0 unique vacuum is necessary not sufficient). [E] KLM condensation: mu(carrier)=det Cartan(D5)*det Cartan(A3)=4*4=16; order-4 mu4 glue (|L|=4) => mu(E8)=16/16=1 (torus GSD=1); index-2 => mu(SO16)=16/4=4. [E] |det Cartan(E8)|=1 vs |det Cartan(D8)|=4. [E] the single (E8)_1 character has modular T-eigenvalue e^{-2 pi i/3} (h=0,c=8), 1-dim rep => one character => torus GSD=1. [O] continuum existence (v336). Pins the net at genus-1; does NOT close SEAM.EQUIV.01.
Read the original Python source
Seam s3 rp
v379_seam_s3_rp.py · core
Read the registered purpose, outcome and evidence links
C7 / OS reconstruction: reflection positivity (the OS axiom) of the explicit gapped collar measure, verified numerically -- the remaining constructive-QFT ingredient for the OS reconstruction. [E] the gapped p+ip band gives G(tau)=mean_k e^{-eps_k tau} with eps_k>=gap/2~1>0 (v367) -- a positive spectral measure; [E] the Hankel/reflection Gram matrix M_ij=G(tau_i+tau_j) is PSD -- OS reflection positivity holds; [E] neg control: a growing ghost mode makes it indefinite (non-vacuous). [C] RP + gap (v76/v337) + GNS (v240) + c=8 (v376) + (E8)_1 (v377/v378) give the OS inputs; C7/QG.AMB.01 is then DISCHARGED by the redundancy theorem v369, not built. [O] the full constructive measure stays the cited step.
Read the original Python source
Grav kms hessian
v380_grav_kms_hessian.py · frontier
Read the registered purpose, outcome and evidence links
The KMS Entire Hessian -- the Stelle ghost is EXACTLY the Seeley-DeWitt truncation, and resummation pushes it to infinity (upgrades v304/v370 from 'assume entire' to a derived statement). [E] the seam KMS cutoff f(u)=e^{-u} (v259) gives the dressed spin-2 propagator e^{-p^2/M^2}/p^2 = 1/(p^2 a) with a(u)=e^u entire & nowhere zero => only the p^2=0 pole (no ghost). [E] each finite truncation carries the ghost: the R+R^2 order T_1=1+u has the Stelle zero u=-1, the Weyl^2 order T_2=1+u+u^2/2 the pair -1+-i; the entire a has none. [E] the ghost DECOUPLES: |nearest zero| of T_n grows monotonically with n (1,1.41,...,3.07 for n=1..8) -> infinity, so 'entire' = 'do not truncate'. [C] => perturbative graviton unitarity. [O] the exact off-shell curved-space Hessian identification is the cited heat-kernel resummation; perturbative-only, not the non-perturbative measure.
Read the original Python source
Qgamb redundancy
v369_qgamb_redundancy.py · frontier
Read the registered purpose, outcome and evidence links
Track 2 (next-plan v2): the AMBIENT REDUNDANCY statement -- the holographic route to Gravity-complete that SIDESTEPS building the general Euclidean QG measure (C7/QG.AMB.01). Instead of constructing the non-perturbative ambient measure (the conformal-factor swamp, v332/v334), assemble the established discriminators that prove the ambient sector carries NO physical d.o.f. relative to the holomorphic boundary algebra, so QG.AMB.01 is a non-fundamental CERTIFICATION object (like a coordinate choice in GR), not missing dynamics. [E] HOLOMORPHIC => DHR=Vec: #DHR primaries = |det Cartan| = 1 for E8 (ONE primary = vacuum) vs 4 for SO(16)/D8 -- no nontrivial superselection charges => no hidden boundary sector a bulk could carry invisibly. [E] NO TORUS GSD: ground-state degeneracy on T^2 = #anyons = |det K| = 1 for E8 => unique ground state => no topologically protected invisible bulk d.o.f. (the genus-1 obstruction of v344). [E] FINITE PETZ RECOVERY: deviation subspace contracts at (2/3)^6=64/729<1 (v221) => Rec_Sigma bounded. [E] GAP-DECOUPLING MARGIN: Delta_eff=6 ln(3/2)-31/(4 pi^2)~1.648>0 (v76/v337) => ambient sector energetically separated. [C] RECONSTRUCTION MAP: Bisognano-Wichmann gives the modular boost (v258/v309/v323) => Rec_Sigma: (bulk)->A_Sigma^{mu4}. [C] THE REDUNDANCY STATEMENT: under SEAM.EQUIV.01 + BW, O_phys^bulk/Diff ~= A_Sigma^{mu4} => QG.AMB.01 carries no extra physical d.o.f. [O] residual: the two premises -- SEAM.EQUIV.01 (S3 lattice input, v366) + BW intrinsicality (v329); provides the ALTERNATIVE route, does NOT construct the measure. Gravity-complete = Boundary-complete + Ambient-redundancy.
Read the original Python source
Qft4d eg allorder
v381_qft4d_eg_allorder.py · registry
Read the registered purpose, outcome and evidence links
The all-order Epstein-Glaser / BRST contract for the perturbative 4D S-matrix S_pert -- the 'all-order closing statement' for the 4D perturbative QFT leg (review Point 3). Types the standard causal-perturbation-theory closure for the TFPT finite interaction class; does NOT build a path integral and is NOT a new mechanism. [E] POWER-COUNTING: every gauge+matter vertex (Yang-Mills + Dirac-Yukawa + Higgs + ghost) is a mass-dimension-4 marginal operator => the EG singular order omega<=4 => a FINITE counterterm space at every order (power-counting renormalisable). [E] BRST NILPOTENCY s^2=0 verified via the Jacobi identity for su(2) (residual 0) and su(3) (1e-16), the carrier weak+colour gauge content. [E] the seam gap Delta=6 ln(3/2)>0 (v302) gives the IR-safe adiabatic limit g->1. [C] imported: T_n exists to all orders by causal factorisation (Epstein-Glaser 1973 / Brunetti-Fredenhagen). [C] imported: all-order BRST invariance (Piguet-Sorella / Kugo-Ojima) => Slavnov-Taylor to all orders. [E] scope fence: matter+gauge ONLY -- the R^2/Weyl^2 gravity sub-sector is v304/v370/v380, not a local EG vertex; S_pert is not the ambient measure QG.AMB.01 (a [C] redundancy v369). NET [C] (imported EG+BRST on the TFPT dim-4 class + gap), prerequisites [E]; NOT [E], NOT a path integral. Extends v269/v271/v273/v278 from 1-loop to all-order.
Read the original Python source
Alpha quillen exact
v382_alpha_quillen_exact.py · em
Read the registered purpose, outcome and evidence links
Names the Quillen determinant-line VARIATION delta_tau(log det_zeta Delta_U(1) + 8 b1 c3^6 log phi_seam)=0 as a tracked target (review Point 10), elevating EM.WARD.01's residual to a citable [O] so the one EM obligation is trackable separately from the seam-net identification. [E] the Quillen split holds at the root a^3-2 c3^3 a^2 = 8 b1 c3^6 ln(1/phi_seam) at alpha^-1=137.0359992 (F_U(1)=0 IS the stationarity of the U(1) determinant line, v341). [E] every coefficient a named atom: 8 b1=(4/5)*41 a U(1) index, c3=1/(8 pi) the Gauss-Bonnet boundary coeff (c3^3,c3^6 heat-kernel powers), q(D5)+q(A3)=2 discriminant forms (v341/v48/v216). [E] the value alpha^-1=137.0359992168 stays closed (unique root+ablation, v3) -- only 'why this functional' is open. [O] THE TARGET: the from-first-principles AQFT/zeta proof that F_U(1) is the EXACT Quillen functional. [C] a FACE of SEAM.EQUIV.01 (v336): the seam U(1) determinant line lives on the same holomorphic seam net, so once the net is (E8)_1 its U(1) sub-determinant is fixed -- not a second independent gate, it names/tracks the pre-existing EM.WARD.01 residual (v341/v342). No new number (reuses v341, Wolfram-mirrored).
Read the original Python source
Dynamics universal
v383_dynamics_universal.py · frontier
Read the registered purpose, outcome and evidence links
DYNAMICS.UNIVERSAL.01: the UNIVERSAL spectral-gap / Perron-Frobenius principle -- every TFPT sector is the SAME object (a gapped operator with a UNIQUE leading attractor = the physics, a spectral gap = the reason there is no free parameter), extending v303 from F_transfer to the whole theory. [E] FLAVOR (computed): the cusp transfer spectrum {1,(2/3)^6,(1/3)^6} has a simple Perron leading 1 and gap 6 ln(3/2)>0 (v56/v82). [E] DISCRETE COMPILER (computed): the affine-E8 adjacency has Perron eigenvalue 2 (Kac marks, v312/v313) and subleading 2cos(pi/5)=phi, gap 2-phi=1/phi^2>0. [E] DECOUPLING MARGIN: Delta_eff=6 ln(3/2)-31/(4 pi^2)~1.648>0 (v76/v337). [E] TWO NUMBER-FIELD FACETS (v314): golden phi in Q(sqrt5) (carrier-5, static) and (2/3)^6 in Q (family-3, dynamic) are the prime-5 and prime-3 facets of the order-30 Coxeter clock 30=2*3*5=h(E8) -- a factorisation, not an equality. [C] the other sectors are the same structure: gravity (v358/v359), QG saddle (v365), RG fixed point (v157/v158), all-order S-matrix (v381), recovery (v221), F_transfer (v303), tau=i (v333). [C] META-THEOREM: gapped => unique attractor => parameter-free, theory-wide. A synthesis, no new number.
Read the original Python source
Residual certification
v384_residual_certification.py · frontier
Read the registered purpose, outcome and evidence links
RESIDUAL.CERTIFICATION.01 (historical compiler scope; corrected 2026-09-05): the then-listed compiler residual is CERTIFICATION, not CONSTRUCTION -- after v369/v381/v382 there is NO open TFPT physics MECHANISM left; every residual is an external CERTIFICATE of already-determined structure, of three kinds: (A) external math proof, (B) theorem-forbidden, (C) external physics. A ledger-CI audit. [E] LEDGER CI: every residual claim id is present (no orphan). [E] NO-UNIT (B): c3=1/(8 pi) and g_car=5 are mass-dim 0, so by No-Unit (v153/v364) no dimensionless compiler outputs an absolute scale => v_geo is forbidden by theorem, not a gap. [E] GAP-DECOUPLED (A): Delta_eff~1.648>0 (v76/v337) => every readout is independent of the ambient measure (QG.AMB.01 a [C] redundancy, v369). [E] CLASSIFICATION COMPLETE: SEAM.EQUIV.01 (continuum existence=cited MMST/Adamo v336)->A, ALPHA.QUILLEN.EXACT.01 (v382)->A, QG.AMB.01 (v369)->A, v_geo->B, F_transfer (v371-v374, firewall v187)->C; ZERO open internal mechanisms within that scoped compiler list. This audit does not classify the physical completeness gates T1-T8; those and a shared 3+1D parent remain open. [C] CONVERGENCE applies only to the historical compiler residual. An audit, no new number.
Read the original Python source
Uvbranch killtest
v385_uvbranch_killtest.py · registry
Read the registered purpose, outcome and evidence links
UVBRANCH.KILLTEST.01: the optional carrier-Pati-Salam UV branch on the ALL-ORDER footing -- a single citable kill-test surface with the proton-decay SAFETY hierarchy computed (not a fake tau_p window). A consolidation (does NOT re-run the RGE v246 nor re-derive the PS scale v249/v253). [E] carrier content = SM (b=(41/10,-19/6,-7), v159/v246, no new states) and S_pert closes to all orders (v381) => the SM-non-unification is all-order-robust. [X] plain-SM 4D-GUT killed (sin^2 thW=3/8 vs 0.231; v246). [C] optional carrier-PS branch: M_PS=kappa M_scal, M_scal=c3^{7/2}Mbar~3.06e13 GeV, kappa in [1.0,1.2] (v253), content {1,10,16,45} no 126 (v247/v248), not default (v265). [E] proton-decay SAFETY hierarchy: minimal PS gauge bosons are SU(4) leptoquarks (B-L-conserving q<->l, rare LFV K_L->mu e), NOT the p->e+ pi0 dim-6 operator (needs SO(10)/SU(5) diquark, absent); binding bound M_PS>~1e6 GeV cleared by M_PS/M_bound~1e7>>1. [X] frozen kill tests (v265): kappa leaving O(1) / a forbidden rep / a new gauge-charged state / a rare-LFV signal below the bound; no fake tau_p. A consolidation, no new number.
Read the original Python source