Home /Modules/stanford-math/EQUATIONS-MATH.md

EQUATIONS-MATH.md

stanford-math/EQUATIONS-MATH.md · 402 line(s) · markdown

Mathematics Canon

126 equations, sourced from SMMathCanon.sleela -- the single source of truth. This document is a human-readable mirror of that record; if the two ever disagree, the .sleela canon wins and this file is stale.

The foundational mathematics every graduate curriculum rests on -- identities, theorems, and inequalities from geometry through analysis, number theory, probability, and set theory. Each row is a faithful record: the result under the name history gives it, with the rightful discoverer and year, plus a one-line note on why it matters. Where attribution is genuinely diffuse (e.g. the quadratic formula, the 2x2 determinant) the discoverer field says so, rather than crediting a single name.

The one hard rule governs every row: values are faithful to the subject's real origin where publicly known; diffuse or unknown attributions are marked honestly (unknown, classical, attrib.), never invented to look legitimate. Symbolic forms use plain-ASCII notation (^ power, * product, sqrt, integral, sum, product, laplacian, Greek names spelled out) so they stay greppable and diff-friendly. The Why it matters column states each result's significance in one faithful line.

The 126 equations

#EquationFormFieldDiscoverer(s)YearWhy it matters
1Pythagorean theorema^2 + b^2 = c^2geometryPythagoras (attrib.)c. 530 BCThe defining relation of Euclidean distance; the seed of trigonometry and metric geometry.
2Euler's identitye^(i*pi) + 1 = 0analysisLeonhard Euler1748Ties the five fundamental constants in one equation; the canonical example of mathematical beauty.
3Euler's formulae^(i*x) = cos(x) + i*sin(x)analysisLeonhard Euler1748Unifies exponential and trigonometric functions; the backbone of Fourier and signal analysis.
4Fundamental theorem of calculusintegral_a^b f'(x) dx = f(b) - f(a)calculusNewton & Leibnizc. 1670Links differentiation and integration as inverse operations -- the pivot of all of calculus.
5Binomial theorem(x+y)^n = sum_{k=0}^n C(n,k) x^(n-k) y^kalgebrageneralized by Newton1665Expands any power of a sum; foundational to combinatorics, series, and probability.
6Quadratic formulax = (-b +/- sqrt(b^2 - 4ac)) / (2a)algebraancient; modern form diffuseunknownSolves every quadratic in closed form; the first general equation-solver students meet.
7Taylor seriesf(x) = sum_{n>=0} f^(n)(a)/n! * (x-a)^nanalysisBrook Taylor1715Approximates smooth functions by polynomials; the engine behind numerical analysis.
8Fourier seriesf(x) = a0/2 + sum (a_n cos(n x) + b_n sin(n x))analysisJoseph Fourier1807Decomposes periodic signals into sinusoids; the basis of harmonic analysis and DSP.
9Cauchy-Schwarz inequality|<u,v>| <= ||u|| * ||v||analysisCauchy; Bunyakovsky; Schwarz1821The most-used inequality in analysis; underpins inner-product geometry and probability bounds.
10Triangle inequality||u + v|| <= ||u|| + ||v||analysisclassicalunknownDefines what it means to be a distance (a metric); foundational to all of analysis and topology.
11Gauss (divergence) theoremintegral_V (div F) dV = surface_integral F . dSvector calculusGauss; Ostrogradsky1813Converts volume integrals of divergence to surface flux; central to electromagnetism and fluids.
12Green's theoremclosed_integral (P dx + Q dy) = integral_D (dQ/dx - dP/dy) dAvector calculusGeorge Green1828Relates a plane region integral to its boundary; the 2D ancestor of Stokes and divergence.
13Stokes' theoremintegral_S (curl F) . dS = closed_integral_C F . drvector calculusKelvin; Stokes1854The general 'boundary' theorem unifying Green, divergence, and FTC on manifolds.
14De Moivre's theorem(cos x + i sin x)^n = cos(n x) + i sin(n x)complex analysisAbraham de Moivre1707Powers of complex numbers via angles; the route to roots of unity and trig identities.
15Cauchy integral formulaf(a) = (1/(2*pi*i)) closed_integral f(z)/(z-a) dzcomplex analysisAugustin-Louis Cauchy1831A holomorphic function is fixed by its boundary values; the heart of complex analysis.
16Residue theoremclosed_integral f(z) dz = 2*pi*i * sum Res(f, a_k)complex analysisAugustin-Louis Cauchy1831Evaluates contour integrals from isolated singularities; a powerful tool for real integrals too.
17Riemann zeta functional equationzeta(s) = 2^s pi^(s-1) sin(pi s/2) Gamma(1-s) zeta(1-s)number theoryBernhard Riemann1859The symmetry s <-> 1-s of zeta; the stage on which the Riemann Hypothesis plays out.
18Euler productzeta(s) = product_p 1/(1 - p^(-s))number theoryLeonhard Euler1737Rewrites zeta as a product over primes; the analytic bridge into number theory.
19Prime number theorempi(x) ~ x / ln(x)number theoryHadamard; de la Vallee Poussin1896Pins the density of primes to x/ln x; the crowning result of analytic number theory.
20Bayes' theoremP(A|B) = P(B|A) P(A) / P(B)probabilityThomas Bayes1763How evidence updates belief; the foundation of Bayesian inference and modern machine learning.
21Gaussian (normal) densityf(x) = (1/(sigma sqrt(2 pi))) e^(-(x-mu)^2/(2 sigma^2))probabilityGauss; de Moivre1809The bell curve that sums of many small effects converge to; ubiquitous in statistics.
22Central limit theoremsqrt(n)(Xbar - mu) -> N(0, sigma^2) in distributionprobabilityLaplace; Lyapunov; Lindeberg1810Explains why the normal distribution is everywhere; the reason averages are predictable.
23Stirling's approximationn! ~ sqrt(2 pi n) (n/e)^nanalysisJames Stirling; de Moivre1730A fast, accurate estimate of n!; essential in statistical mechanics and combinatorics.
24Gamma functionGamma(n) = integral_0^inf t^(n-1) e^(-t) dtanalysisLeonhard Euler1729Extends the factorial to all complex numbers; appears throughout analysis and probability.
25Basel problemsum_{n>=1} 1/n^2 = pi^2 / 6analysisLeonhard Euler1734Euler's stunning sum 1/n^2 = pi^2/6; the opening move of zeta-function theory.
26Harmonic/log limit (Euler-Mascheroni)lim (sum_{k=1}^n 1/k - ln n) = gammaanalysisEuler; Mascheroni1734Defines gamma, the constant gap between the harmonic series and ln n -- still deeply mysterious.
27Fibonacci / golden ratiophi = (1 + sqrt(5)) / 2number theoryclassical; Fibonacci1202The limiting ratio of Fibonacci numbers; recurs across growth, art, and number theory.
28Law of cosinesc^2 = a^2 + b^2 - 2ab cos(C)geometryclassical (Euclid II)c. 300 BCGeneralizes Pythagoras to any triangle; the workhorse of trigonometric computation.
29Heron's formulaArea = sqrt(s(s-a)(s-b)(s-c)), s=(a+b+c)/2geometryHeron of Alexandriac. 60 ADArea of a triangle from its three sides alone; a classic of computational geometry.
30Euler's polyhedron formulaV - E + F = 2topologyLeonhard Euler1752V - E + F = 2 launched topology; the first glimpse of a shape-invariant (the Euler characteristic).
31Determinant of 2x2det([[a,b],[c,d]]) = ad - bclinear algebraclassicalunknownMeasures signed area/volume scaling; detects invertibility of a linear map.
32Eigenvalue equationA v = lambda vlinear algebraCauchy; Hilbert (spectral)1829Finds the directions a transformation merely stretches; the core of spectral theory and quantum mechanics.
33Green-Tao theoremprimes contain arbitrarily long arithmetic progressionsnumber theoryBen Green & Terence Tao2004Primes contain arbitrarily long arithmetic progressions; a landmark of modern number theory.
34Bayes-rule odds formposterior odds = likelihood ratio * prior oddsprobabilityThomas Bayes1763Recasts Bayes as prior odds times a likelihood ratio -- the most practical way to update beliefs.
35L'Hopital's rulelim f/g = lim f'/g' (0/0 or inf/inf)calculusl'Hopital; Bernoulli1696Resolves indeterminate limits via derivatives; a staple technique of first-year calculus.
36Integration by partsintegral u dv = u v - integral v ducalculusBrook Taylor (attrib.)1715The integral mirror of the product rule; indispensable across analysis and physics.
37Chain ruled/dx f(g(x)) = f'(g(x)) g'(x)calculusLeibniz (notation)1676Differentiates compositions; the rule that makes backpropagation and all of deep learning possible.
38Jensen's inequalityf(E[X]) <= E[f(X)] (f convex)analysisJohan Jensen1906Convexity controls averages; the engine behind countless bounds in probability and information theory.
39AM-GM inequality(x1+...+xn)/n >= (x1...xn)^(1/n)analysisCauchy (proof)1821The arithmetic mean never undershoots the geometric mean; a foundational optimization inequality.
40Sum of first n integerssum_{k=1}^n k = n(n+1)/2algebraclassical; Gauss (attrib.)unknownThe triangular-number formula; the archetypal closed form for a finite sum.
41Geometric seriessum_{k=0}^inf r^k = 1/(1-r), |r| < 1analysisclassicalunknownA convergent sum with a clean closed form; the basis of compound interest, decay, and generating functions.
42Euler's totient productphi(n) = n product_{p|n} (1 - 1/p)number theoryLeonhard Euler1763Counts integers coprime to n; central to modular arithmetic and RSA cryptography.
43Fermat's little theorema^p = a (mod p), p primenumber theoryPierre de Fermat1640A prime's fingerprint in modular arithmetic; the basis of primality tests and public-key crypto.
44Cantor's diagonal cardinality|P(S)| > |S| (2^aleph0 > aleph0)set theoryGeorg Cantor1891Proves there are strictly larger infinities; the birth of set theory and the uncountable.
45Laplace's equationlaplacian(f) = 0partial differential equationsPierre-Simon Laplace1784The defining equation of harmonic functions and potential theory; governs steady-state heat, gravity, and electrostatics.
46Euler's theorem (totient)a^phi(n) = 1 (mod n), gcd(a,n)=1number theoryLeonhard Euler1763Generalizes Fermat's little theorem to any modulus; the arithmetic that makes RSA decryption work.
47Quadratic reciprocity(p/q)(q/p) = (-1)^(((p-1)/2)((q-1)/2))number theoryCarl Friedrich Gauss1801Gauss's 'golden theorem' linking when two primes are squares mod each other; the seed of class field theory.
48Cramer's rulex_i = det(A_i) / det(A)linear algebraGabriel Cramer1750Solves a linear system by determinant ratios; the closed-form expression of a unique solution.
49Wallis productpi/2 = product_{n>=1} (2n)(2n)/((2n-1)(2n+1))analysisJohn Wallis1656An infinite product for pi predating calculus; an early landmark in the analysis of limits and products.
50Lagrange mean value theoremf'(c) = (f(b) - f(a)) / (b - a)calculusJoseph-Louis Lagrange1797Guarantees a point where the instantaneous rate equals the average rate; the backbone of error bounds in calculus.
51Curl of a gradientcurl(grad f) = 0vector calculusclassicalunknownA gradient field is always irrotational; the identity behind conservative fields and scalar potentials.
52Intermediate value theoremf(a) < y < f(b) => exists c in (a,b): f(c) = yanalysisBernard Bolzano1817A continuous function hits every value between its endpoints; the rigorous root of existence proofs.
53Wilson's theorem(p-1)! = -1 (mod p), p primenumber theoryLagrange (proof); Wilson1771A factorial test that characterizes primes exactly; a classic link between combinatorics and primality.
54Chinese remainder theoremx = a_i (mod n_i) has unique soln mod product n_inumber theorySunzi (attrib.)c. 400 ADReconstructs a number from its remainders modulo coprime bases; foundational to modular arithmetic and RSA.
55Vandermonde determinantdet = product_{i<j} (x_j - x_i)linear algebraAlexandre-Theophile Vandermonde1772A clean closed form for a key determinant; central to polynomial interpolation and coding theory.
56Cayley-Hamilton theoremp(A) = 0 (A satisfies its characteristic polynomial)linear algebraCayley; Hamilton1858Every square matrix satisfies its own characteristic equation; the basis for matrix functions and minimal polynomials.
57Gram-Schmidt orthogonalizationu_k = v_k - sum_{j<k} proj_{u_j}(v_k)linear algebraGram; Schmidt1907Turns any basis into an orthonormal one; the engine behind QR decomposition and least squares.
58Parseval's identitysum |c_n|^2 = (1/(2 pi)) integral_{-pi}^{pi} |f(x)|^2 dxanalysisMarc-Antoine Parseval1799Energy is conserved between a signal and its Fourier coefficients; the cornerstone of Hilbert-space harmonic analysis.
59Chebyshev's inequalityP(|X - mu| >= k sigma) <= 1/k^2probabilityPafnuty Chebyshev1867Bounds how far a random variable strays from its mean using only variance; the first general concentration inequality.
60Fundamental theorem of algebraevery degree-n polynomial has exactly n complex roots (with multiplicity)algebraCarl Friedrich Gauss1799Guarantees the complex numbers are algebraically closed; the reason every polynomial factors completely.
61Rolle's theoremf(a) = f(b) => exists c in (a,b): f'(c) = 0calculusMichel Rolle1691The special case that powers the mean value theorem; the first rigorous link between equal values and a flat tangent.
62Cauchy-Riemann equationsu_x = v_y, u_y = -v_xcomplex analysisCauchy; Riemann1851The exact condition for complex differentiability; the gateway from calculus to analytic-function theory.
63Argument principle(1/(2*pi*i)) closed_integral f'(z)/f(z) dz = Z - Pcomplex analysisAugustin-Louis Cauchy1831Counts zeros minus poles inside a contour; the tool behind Rouche's theorem and root-locating.
64Lagrange interpolationp(x) = sum_j y_j product_{m!=j} (x - x_m)/(x_j - x_m)numerical analysisLagrange; Waring1795Builds the unique polynomial through given points; the foundation of polynomial interpolation and quadrature.
65Newton's methodx_{n+1} = x_n - f(x_n)/f'(x_n)numerical analysisNewton; Raphson1690Finds roots by successive linearization, converging quadratically; the workhorse of numerical root-finding and optimization.
66Singular value decompositionA = U Sigma V^Tlinear algebraBeltrami; Jordan1873Factors any matrix into rotations and scalings; the backbone of least squares, PCA, and low-rank approximation.
67Bolzano-Weierstrass theoremevery bounded sequence in R^n has a convergent subsequenceanalysisBolzano; Weierstrass1817Guarantees limit points of bounded sets; a foundational compactness result underpinning existence proofs.
68Banach fixed-point theorema contraction on a complete metric space has a unique fixed pointanalysisStefan Banach1922Turns iteration into existence-and-uniqueness; the engine behind ODE solutions and numerical convergence.
69Holder's inequalitysum |a_k b_k| <= (sum |a_k|^p)^(1/p) (sum |b_k|^q)^(1/q), 1/p + 1/q = 1analysisOtto Holder1889Generalizes Cauchy-Schwarz to L^p spaces; indispensable across functional analysis and probability bounds.
70Minkowski's inequality||a + b||_p <= ||a||_p + ||b||_panalysisHermann Minkowski1896The triangle inequality for L^p norms; what makes L^p a normed space.
71Weak law of large numbersXbar_n -> mu in probability as n -> infprobabilityJacob Bernoulli1713Sample averages converge to the true mean; the formal reason large samples are reliable.
72Markov's inequalityP(X >= a) <= E[X]/a, X >= 0, a > 0probabilityAndrey Markov1884Bounds a nonnegative variable's tail by its mean alone; the seed of Chebyshev and concentration inequalities.
73Mobius inversiong(n) = sum_{d|n} f(d) <=> f(n) = sum_{d|n} mu(n/d) g(d)number theoryAugust Ferdinand Mobius1832Inverts divisor sums via the Mobius function; a central identity of multiplicative number theory.
74Dirichlet's theorem (primes in AP)gcd(a,d)=1 => infinitely many primes of the form a + n*dnumber theoryPeter Gustav Lejeune Dirichlet1837Primes are spread through every valid arithmetic progression; a landmark that launched analytic number theory.
75Catalan number formulaC_n = (1/(n+1)) C(2n, n)combinatoricsEuler; Catalan1838Counts balanced structures (parenthesizations, trees, paths); one of combinatorics' most ubiquitous sequences.
76Lagrange multipliersgrad f = lambda grad goptimizationJoseph-Louis Lagrange1788Finds extrema subject to constraints by matching gradients; the backbone of constrained optimization.
77Euler-Lagrange equationd/dt (dL/dq') - dL/dq = 0calculus of variationsEuler; Lagrange1755The condition a path must satisfy to extremize an action; the foundation of variational mechanics.
78Divergence of a curldiv(curl F) = 0vector calculusclassicalunknownA curl field has zero divergence; a core identity of vector calculus and magnetostatics.
79Gradient theorem (line integrals)integral_C grad f . dr = f(b) - f(a)vector calculusclassicalunknownA conservative field's line integral depends only on endpoints; the FTC for gradients.
80Leibniz integral ruled/dx integral_{a(x)}^{b(x)} f dt = f(b) b' - f(a) a' + integral f_x dtcalculusGottfried Leibniz1697Differentiates an integral with variable limits; essential across physics and probability.
81Taylor's theorem with remainderf(x) = sum_{k=0}^{n} f^(k)(a)/k! (x-a)^k + R_nanalysisBrook Taylor; Lagrange (remainder)1715Bounds the error of a Taylor approximation; the rigorous basis of numerical estimates.
82Mean value theorem for integralsintegral_a^b f dx = f(c)(b - a)calculusclassical (Cauchy)1823Guarantees a point whose value equals the integral's average; a staple existence result.
83Cauchy's integral theoremclosed_integral f(z) dz = 0 (f holomorphic)complex analysisAugustin-Louis Cauchy1825A holomorphic function integrates to zero around a closed loop; the start of complex analysis.
84Liouville's theorem (bounded entire)a bounded entire function is constantcomplex analysisJoseph Liouville1847A bounded entire function must be constant; yields a one-line proof of the fundamental theorem of algebra.
85Maximum modulus principle|f| attains its max on the boundarycomplex analysisRiemann; Schwarz1851A holomorphic function reaches its maximum size only on the boundary; central to complex analysis.
86Schwarz lemma|f(z)| <= |z| and |f'(0)| <= 1 for f: disk->disk, f(0)=0complex analysisHermann Schwarz1869Constrains holomorphic self-maps of the disk; the seed of hyperbolic geometry and Pick interpolation.
87Morera's theoremclosed_integral f = 0 for all triangles => f holomorphiccomplex analysisGiacinto Morera1886The converse of Cauchy's theorem: zero loop integrals imply holomorphy.
88Rouche's theorem|g| < |f| on C => f and f+g have equally many zeros insidecomplex analysisEugene Rouche1862Counts zeros by comparing magnitudes on a contour; a practical zero-counting tool.
89Jordan curve theorema simple closed curve splits the plane into inside and outsidetopologyCamille Jordan1887A simple closed curve separates the plane into inside and outside; a deep foundational fact of topology.
90Brouwer fixed-point theorema continuous map of a closed ball to itself has a fixed pointtopologyL. E. J. Brouwer1911Every continuous self-map of a ball has a fixed point; a cornerstone of topology and game theory.
91Gauss-Bonnet theoremintegral_M K dA + integral_{dM} k_g ds = 2 pi chi(M)differential geometryGauss; Bonnet1848Ties total curvature to the Euler characteristic; the bridge between geometry and topology.
92Noether's theoremevery differentiable symmetry yields a conservation lawmathematical physicsEmmy Noether1918Every continuous symmetry yields a conservation law; the organizing principle of modern physics.
93Implicit function theoremF(x,y)=0, F_y != 0 => y = g(x) locallyanalysisDini; Cauchy1877Guarantees a relation defines a function locally; essential to manifolds and optimization.
94Inverse function theoremf' invertible at a => f invertible near aanalysisclassicalunknownA map with invertible derivative is locally invertible; a foundation of differential calculus.
95Picard-Lindelof existencey' = f(x,y), f Lipschitz => unique local solutiondifferential equationsPicard; Lindelof1894Lipschitz continuity guarantees a unique ODE solution; the existence theorem of differential equations.
96Wronskian (linear independence)W(f,g) = f g' - f' gdifferential equationsJozef Wronski1812A determinant test for independence of solutions; standard in linear ODE theory.
97Bessel's equationx^2 y'' + x y' + (x^2 - n^2) y = 0special functionsFriedrich Bessel1824Governs cylindrical waves and vibrations; its solutions appear throughout physics and engineering.
98Legendre's equation(1 - x^2) y'' - 2 x y' + l(l+1) y = 0special functionsAdrien-Marie Legendre1782Yields the Legendre polynomials central to potential theory and spherical harmonics.
99Hermite's equationy'' - 2 x y' + 2 n y = 0special functionsCharles Hermite1864Yields Hermite polynomials; the eigenfunctions of the quantum harmonic oscillator.
100Laguerre's equationx y'' + (1 - x) y' + n y = 0special functionsEdmond Laguerre1879Yields Laguerre polynomials; the radial part of the hydrogen-atom wavefunction.
101Beta functionB(x,y) = integral_0^1 t^(x-1)(1-t)^(y-1) dt = Gamma(x)Gamma(y)/Gamma(x+y)special functionsEuler; Legendre1730Euler's integral linking factorials and probabilities; the basis of the Beta distribution.
102Riemann sum (definite integral)integral_a^b f dx = lim sum f(x_i*) dx_icalculusBernhard Riemann1854Defines the integral as a limit of partition sums; the rigorous meaning of area under a curve.
103Weierstrass M-testsum M_n converges, |f_n| <= M_n => sum f_n converges uniformlyanalysisKarl Weierstrass1880A simple bound guaranteeing uniform convergence of a series; a workhorse of analysis.
104Abel's summation / testconvergence of sum a_n b_n via partial sumsanalysisNiels Henrik Abel1826Summation by parts giving convergence of products of sequences; foundational to series analysis.
105Dirichlet's test for seriesa_n monotone ->0, partial sums of b_n bounded => sum a_n b_n convergesanalysisPeter Gustav Lejeune Dirichlet1862Convergence from a monotone factor and bounded partial sums; a standard series test.
106Holder continuity|f(x) - f(y)| <= C |x - y|^alphaanalysisOtto Holder1882Quantifies smoothness between continuity and differentiability; central to PDE regularity.
107Poisson summation formulasum f(n) = sum F-hat(k)harmonic analysisSimeon Denis Poisson1827Equates a sum of a function to a sum of its Fourier transform; links time and frequency sampling.
108Parseval-Plancherel theoremintegral |f|^2 dx = integral |F-hat|^2 dwharmonic analysisMichel Plancherel1910Energy is preserved by the Fourier transform; the isometry of harmonic analysis.
109Convolution theoremF{f * g} = F{f} . F{g}harmonic analysisclassical (Fourier)unknownConvolution in time is multiplication in frequency; the reason transforms simplify filtering.
110Spectral theorem (symmetric matrix)A = Q Lambda Q^T, A symmetriclinear algebraCauchy; Hilbert1829A symmetric matrix diagonalizes in an orthonormal basis; the foundation of PCA and quantum observables.
111LU decompositionA = L Unumerical linear algebraAlan Turing (modern); Gauss1948Factors a matrix for fast linear solves; the standard method behind Gaussian elimination.
112QR decompositionA = Q R, Q orthogonal, R upper-triangularnumerical linear algebraclassical (Gram-Schmidt)1907Orthogonal-times-triangular factorization; the stable workhorse of least squares and eigenvalues.
113Cholesky decompositionA = L L^T, A symmetric positive-definitenumerical linear algebraAndre-Louis Cholesky1910An efficient factorization for positive-definite matrices; standard in optimization and statistics.
114Rayleigh quotientR(A,x) = (x^T A x) / (x^T x)linear algebraLord Rayleigh1877Estimates eigenvalues by a normalized quadratic form; the basis of iterative eigenvalue methods.
115Perron-Frobenius theorema positive matrix has a unique largest positive eigenvaluelinear algebraPerron; Frobenius1907A positive matrix has a dominant positive eigenvalue; underpins PageRank and Markov chains.
116Sylvester's law of inertiacongruent forms share the signature (p, q, z)linear algebraJames Joseph Sylvester1852The signature of a quadratic form is invariant under congruence; classifies conics and metrics.
117Hadamard's inequality|det A| <= product of column normslinear algebraJacques Hadamard1893Bounds a determinant by its column norms; a geometric bound on volume.
118Cramer-Rao boundVar(theta-hat) >= 1 / I(theta)mathematical statisticsCramer; Rao1945Sets the minimum variance of an unbiased estimator; the gold standard of estimation theory.
119Maximum likelihood estimatortheta-hat = argmax L(theta; x)mathematical statisticsRonald A. Fisher1922Chooses parameters that make the data most probable; the dominant estimation principle in statistics.
120Poisson distributionP(k) = lambda^k e^(-lambda) / k!probabilitySimeon Denis Poisson1837Models counts of rare independent events; ubiquitous in queues, physics, and reliability.
121Binomial distributionP(k) = C(n,k) p^k (1-p)^(n-k)probabilityJacob Bernoulli1713Counts successes in independent trials; the discrete foundation of probability.
122Markov chain stationary distributionpi P = pi, sum pi_i = 1probabilityAndrey Markov1906The long-run state frequencies of a chain; central to MCMC and equilibrium analysis.
123Borel-Cantelli lemmasum P(A_n) < inf => P(A_n i.o.) = 0probabilityBorel; Cantelli1909Links summable probabilities to almost-sure non-recurrence; a key tool of probability theory.
124Jensen's gap / convex dualityf convex => f(E[X]) <= E[f(X)] with Legendre dualconvex analysisLegendre; Fenchel1949Convex-conjugate duality underlying bounds and optimization; the Legendre-Fenchel transform.
125Simpson's rule (quadrature)integral_a^b f dx = (h/3)[f0 + 4 f1 + f2 + ...]numerical analysisThomas Simpson1743Approximates integrals with parabolic arcs; a staple of numerical integration.
126Runge-Kutta (RK4) stepy_{n+1} = y_n + (h/6)(k1 + 2 k2 + 2 k3 + k4)numerical analysisRunge; Kutta1901A fourth-order one-step ODE integrator; the default method for numerical simulation.

Validate

These equations are compiled, not just documented. Build the source compiler, then check and run the sibling demo (which prints every entry and the running total across all four canons):

SLEELA_LIB="$PWD/stanford-math" impl/build/sleela check stanford-math/SMArchive.sleela
SLEELA_LIB="$PWD/stanford-math" impl/build/sleela run   stanford-math/SMArchive.sleela

Sources & attribution

Names, forms, discoverers, and years follow the standard history-of-science record; the Why it matters notes summarize the result's established significance. See SMProvenance.sleela for the tiered source registry (primary papers, secondary references such as NIST DLMF / Abramowitz & Stegun, and tertiary locators including Wikipedia, cited as CC BY-SA and never treated as a primary authority) and the repo README.md for per-class integrity goals. This archive records and attributes published results; it does not originate, own, or certify them.

Definitions & relevance (second-layer additions)

The entries below are the second-layer additions to this canon, organized by type. Each gives a short definition (what the result states and what its symbols mean) and its relevance (why it is foundational to mathematics).

Optimization & the calculus of variations

Lagrange multipliers — grad f = lambda grad g. At a constrained extremum the gradient of the objective is parallel to the gradient of the constraint. It is the backbone of constrained optimization, from economics to physics to machine learning.

Euler-Lagrange equation — d/dt (dL/dq') - dL/dq = 0. The condition a path must satisfy to make an integral (an "action") stationary. It is the foundation of variational mechanics and of every principle-of-least-action formulation of physics.

Vector & single-variable calculus

Divergence of a curl — div(curl F) = 0. The divergence of any curl field vanishes identically. It is a core vector-calculus identity and the reason magnetic fields have no sources.

Gradient theorem (line integrals) — integral_C grad f . dr = f(b) - f(a). A line integral of a gradient depends only on the endpoints. It is the fundamental theorem of calculus for gradient fields and defines what "conservative" means.

Leibniz integral rule — d/dx integral f dt = f(b)b' - f(a)a' + integral f_x dt. Differentiates an integral whose integrand and limits depend on the parameter. It is a workhorse across physics, probability, and applied analysis.

Mean value theorem for integrals — integral_a^b f dx = f(c)(b - a). A continuous function attains its average value somewhere on the interval. It is a basic existence result underpinning error estimates and the FTC.

Riemann sum (definite integral) — integral_a^b f dx = lim sum f(x_i*) dx_i. Defines the integral as the limit of partition sums. It is the rigorous meaning of "area under a curve" and the starting point of integration theory.

Real analysis

Taylor's theorem with remainder — f(x) = sum f^(k)(a)/k!(x-a)^k + R_n. Adds a quantified error term to the Taylor expansion. It is what makes polynomial approximation rigorous and bounds the accuracy of numerical methods.

Implicit function theorem — F(x,y)=0, F_y != 0 => y = g(x) locally. A relation defines one variable as a function of the others wherever a derivative is nonzero. It is essential to the theory of manifolds, constraints, and optimization.

Inverse function theorem — a map whose derivative is invertible at a point is locally invertible. It is a foundation of differential calculus on R^n and the companion to the implicit function theorem.

Weierstrass M-test — bounding each term by a convergent series sum M_n guarantees uniform convergence. It is the simplest, most-used test for the uniform convergence that lets one interchange limits, sums, and integrals.

Abel's summation / test — summation by parts yields convergence for products of sequences. It is a foundational technique for series whose terms are products of a monotone and an oscillating factor.

Dirichlet's test for series — a monotone factor tending to zero times a factor with bounded partial sums converges. It is a standard, powerful convergence test, e.g. for conditionally convergent trigonometric series.

Holder continuity — |f(x) - f(y)| <= C |x - y|^alpha. Quantifies smoothness on a scale between mere continuity (alpha -> 0) and Lipschitz (alpha = 1). It is central to the regularity theory of PDEs.

Complex analysis

Cauchy's integral theorem — closed_integral f(z) dz = 0 for a holomorphic function. The loop integral of an analytic function vanishes. It is the cornerstone from which the whole edifice of complex analysis is built.

Liouville's theorem (bounded entire) — a bounded entire function must be constant. Its relevance includes a one-line proof of the fundamental theorem of algebra and deep rigidity results for analytic functions.

Maximum modulus principle — a non-constant holomorphic function attains its maximum size only on the boundary. It is a defining rigidity property of analytic functions, used throughout complex analysis and PDE.

Schwarz lemma — a holomorphic self-map of the unit disk fixing the origin satisfies |f(z)| <= |z|. It seeds hyperbolic geometry on the disk and the theory of conformal self-maps.

Morera's theorem — vanishing integrals over all triangles imply holomorphy. It is the converse of Cauchy's theorem and a key tool for proving functions are analytic (e.g. limits of analytic functions).

Rouche's theorem — if |g| < |f| on a contour, then f and f+g have the same number of zeros inside. It is the practical instrument for counting and locating zeros, including a slick proof of the fundamental theorem of algebra.

Topology & geometry

Jordan curve theorem — a simple closed curve divides the plane into an inside and an outside. Obvious to the eye yet subtle to prove, it is a foundational fact of plane topology.

Brouwer fixed-point theorem — any continuous map of a closed ball to itself has a fixed point. It is a cornerstone of topology with profound consequences in economics (equilibrium existence) and game theory (Nash).

Gauss-Bonnet theorem — integral_M K dA + integral_{dM} k_g ds = 2 pi chi(M). Total curvature equals 2 pi times the Euler characteristic. It is the prototypical bridge between local geometry (curvature) and global topology.

Noether's theorem — every continuous symmetry of an action yields a conservation law. It is the organizing principle of modern physics: time symmetry gives energy conservation, space symmetry gives momentum, and so on.

Differential equations & special functions

Picard-Lindelof existence — a Lipschitz right-hand side guarantees a unique local solution of y' = f(x,y). It is the fundamental existence-and-uniqueness theorem for ordinary differential equations.

Wronskian (linear independence) — W(f,g) = f g' - f' g. A determinant that tests whether solutions of a linear ODE are independent. It is a standard tool in the structure theory of linear differential equations.

Bessel's equation — x^2 y'' + x y' + (x^2 - n^2) y = 0. Its solutions, Bessel functions, describe cylindrical waves and vibrations. They appear across physics and engineering wherever cylindrical symmetry arises.

Legendre's equation — (1 - x^2) y'' - 2 x y' + l(l+1) y = 0. Yields the Legendre polynomials, central to potential theory and the spherical harmonics.

Hermite's equation — y'' - 2 x y' + 2 n y = 0. Yields the Hermite polynomials, the eigenfunctions of the quantum harmonic oscillator and a basis for Gaussian-weighted analysis.

Laguerre's equation — x y'' + (1 - x) y' + n y = 0. Yields the Laguerre polynomials, which form the radial part of the hydrogen-atom wavefunction.

Beta function — B(x,y) = integral_0^1 t^(x-1)(1-t)^(y-1) dt. Euler's integral linking factorials via the Gamma function. It underlies the Beta distribution and many integral evaluations.

Harmonic analysis

Poisson summation formula — sum f(n) = sum F-hat(k). Equates a sum of a function over a lattice to a sum of its Fourier transform over the dual lattice. It links sampling in time to periodicity in frequency and is central to analytic number theory.

Parseval-Plancherel theorem — integral |f|^2 dx = integral |F-hat|^2 dw. The Fourier transform preserves energy (the L^2 norm). It is the isometry at the heart of harmonic analysis and signal processing.

Convolution theorem — F{f * g} = F{f} . F{g}. Convolution in one domain is pointwise multiplication in the transform domain. It is the reason transforms make filtering and differential equations tractable.

Linear algebra

Spectral theorem (symmetric matrix) — A = Q Lambda Q^T for symmetric A. A symmetric matrix is diagonalized by an orthonormal basis of eigenvectors. It is the foundation of principal component analysis and of quantum observables.

Rayleigh quotient — R(A,x) = (x^T A x)/(x^T x). A normalized quadratic form whose extremes are eigenvalues. It is the basis of variational eigenvalue estimates and iterative methods.

Perron-Frobenius theorem — a positive matrix has a unique largest positive eigenvalue with a positive eigenvector. It underpins PageRank, Markov-chain equilibria, and population-growth models.

Sylvester's law of inertia — the numbers of positive, negative, and zero eigenvalues of a quadratic form are invariant under congruence. It classifies quadratic forms and the signature of metrics.

Hadamard's inequality — |det A| <= product of column norms. The determinant is bounded by the product of its column lengths, with equality for orthogonal columns. It is a sharp geometric bound on volume.

Numerical linear algebra

LU decomposition — A = L U. Factors a matrix into lower- and upper-triangular parts for efficient repeated solves. It is the matrix form of Gaussian elimination and the default direct solver.

QR decomposition — A = Q R, with Q orthogonal and R upper-triangular. A numerically stable factorization. It is the workhorse behind least squares and the QR eigenvalue algorithm.

Cholesky decomposition — A = L L^T for symmetric positive-definite A. An efficient, stable factorization using half the work of LU. It is standard in optimization, statistics (covariance), and simulation.

Mathematical statistics, probability & convexity

Cramer-Rao bound — Var(theta-hat) >= 1 / I(theta). Sets the minimum variance any unbiased estimator can achieve, via the Fisher information I. It is the gold standard for judging estimator efficiency.

Maximum likelihood estimator — theta-hat = argmax L(theta; x). Chooses the parameters that make the observed data most probable. It is the dominant estimation principle across statistics, with strong large-sample optimality.

Poisson distribution — P(k) = lambda^k e^(-lambda)/k!. The law of counts of rare, independent events at constant average rate. It is ubiquitous in queues, physics, and reliability.

Binomial distribution — P(k) = C(n,k) p^k (1-p)^(n-k). The law of the number of successes in independent trials. It is the discrete foundation of probability and the seed of the normal approximation.

Markov chain stationary distribution — pi P = pi, sum pi_i = 1. The long-run frequencies a chain settles into. It is central to Markov chain Monte Carlo, PageRank, and equilibrium analysis.

Borel-Cantelli lemma — if sum P(A_n) < inf then A_n happens only finitely often, almost surely. It is a basic but powerful tool for almost-sure convergence and the strong law of large numbers.

Jensen's gap / convex duality — convex-conjugate (Legendre-Fenchel) duality underlying many bounds and optimization problems. It formalizes the gap in Jensen's inequality and is foundational to convex optimization.

Numerical analysis

Simpson's rule (quadrature) — integral_a^b f dx = (h/3)[f0 + 4 f1 + f2 + ...]. Approximates an integral with parabolic arcs over pairs of subintervals. It is a staple of numerical integration, markedly more accurate than the trapezoidal rule.

Runge-Kutta (RK4) step — y_{n+1} = y_n + (h/6)(k1 + 2 k2 + 2 k3 + k4). A fourth-order one-step method for integrating ODEs. It is the default general-purpose integrator in scientific computing.