{
  "id": "characteristic-function",
  "title": "Characteristic Function",
  "lang": "en",
  "source_lang": "en",
  "summary": "A characteristic function is a complex-valued transform that uniquely represents a probability distribution and converts sums of independent random variables into products.",
  "category": "mathematics",
  "subcategory": "probability-statistics",
  "type": "concept",
  "aliases": [
    "Characteristic function of a probability distribution",
    "Fourier–Stieltjes transform of a probability measure"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/characteristic-function/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/characteristic-function/",
      "json": "https://aiwiki.page/zh/characteristic-function.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "In probability theory, the characteristic function of a real-valued random variable \\(X\\) is the function \\(\\varphi_X(t)=\\mathbb{E}[e^{itX}]\\), defined for every real number \\(t\\), where \\(i^2=-1\\). It encodes the entire probability distribution of \\(X\\), always exists, and turns sums of independent random variables into products. As a version of the Fourier transform of a probability measure, it connects probability with harmonic analysis and provides a central tool for studying distributions and their limits."
    },
    {
      "heading": "Definition and conventions",
      "text": "Let \\(X\\) have distribution \\(\\mu\\) on \\(\\mathbb{R}\\). Its characteristic function is\n\n\\[\\varphi_X(t) = \\mathbb{E}\\!\\left[e^{itX}\\right] = \\int_{\\mathbb{R}}e^{itx}\\,\\mu(dx), \\qquad t\\in\\mathbb{R}.\\]\n\nHere \\(\\mathbb{E}\\) denotes expectation. By Euler’s formula,\n\n\\[\\varphi_X(t) = \\mathbb{E}[\\cos(tX)] + i\\,\\mathbb{E}[\\sin(tX)].\\]\n\nThus its real and imaginary parts describe the average cosine and sine components of the distribution at frequency \\(t\\).\n\nIf \\(X\\) has a probability density function \\(f\\), then\n\n\\[\\varphi_X(t) = \\int_{-\\infty}^{\\infty}e^{itx}f(x)\\,dx.\\]\n\nIf \\(X\\) has a discrete distribution supported on values \\(x_k\\), then\n\n\\[\\varphi_X(t) = \\sum_k e^{itx_k}\\Pr(X=x_k).\\]\n\nThe measure-based definition also covers distributions that are neither discrete nor described by a density.\n\nFourier-transform conventions differ in the sign of the exponent and in factors of \\(2\\pi\\). The standard probability convention used here has \\(e^{itx}\\) in the forward transform and \\(e^{-itx}\\) in the inverse transform.\n\nIn other areas of mathematics, “characteristic function” can instead mean an indicator function of a set. That usage is distinct from the probability-theoretic meaning."
    },
    {
      "heading": "Existence and elementary properties",
      "text": "Because\n\n\\[|e^{itX}|=1,\\]\n\nthe defining expectation exists for every probability distribution and every real \\(t\\). No finite mean, variance, or higher moment is required.\n\nEvery characteristic function satisfies:\n\n- Normalization: \\(\\varphi_X(0)=1\\).\n- Boundedness: \\(|\\varphi_X(t)|\\leq 1\\).\n- Conjugate symmetry: \\(\\varphi_X(-t)=\\overline{\\varphi_X(t)}\\).\n- Uniform continuity: \\(\\varphi_X\\) is uniformly continuous on \\(\\mathbb{R}\\).\n\nFor example, uniform continuity follows from\n\n\\[|\\varphi_X(t+h)-\\varphi_X(t)| \\leq \\mathbb{E}\\!\\left[|e^{ihX}-1|\\right].\\]\n\nThe right-hand side is independent of \\(t\\) and tends to zero as \\(h\\to0\\), by the dominated convergence theorem.\n\nIf \\(X\\) is symmetric about zero, its characteristic function is real-valued and even. Conversely, a real-valued characteristic function represents a distribution symmetric about zero.\n\nFor real constants \\(a\\) and \\(b\\),\n\n\\[\\varphi_{aX+b}(t) = e^{itb}\\varphi_X(at).\\]\n\nTranslation therefore adds a phase factor, while scaling changes the frequency argument."
    },
    {
      "heading": "Independent sums and convolution",
      "text": "If \\(X\\) and \\(Y\\) are independent, then\n\n\\[\\varphi_{X+Y}(t) = \\mathbb{E}[e^{itX}e^{itY}] = \\varphi_X(t)\\varphi_Y(t).\\]\n\nMore generally, for mutually independent random variables \\(X_1,\\ldots,X_n\\),\n\n\\[\\varphi_{X_1+\\cdots+X_n}(t) = \\prod_{j=1}^{n}\\varphi_{X_j}(t).\\]\n\nThe distribution of an independent sum is the convolution of the individual distributions. Characteristic functions replace this operation, which may involve complicated integrals or sums, with ordinary multiplication.\n\nFor independent and identically distributed random variables,\n\n\\[\\varphi_{X_1+\\cdots+X_n}(t) = \\varphi_X(t)^n.\\]\n\nThis identity is fundamental in the study of repeated sums, random walks, and limit theorems. Without independence, the factorization generally fails."
    },
    {
      "heading": "Examples",
      "text": "Constant random variable\n\nIf \\(X=c\\) with probability one, then\n\n\\[\\varphi_X(t)=e^{itc}.\\]\n\nIn particular, a point mass at zero has characteristic function identically equal to \\(1\\).\n\nBernoulli distribution\n\nFor a Bernoulli distribution with \\(\\Pr(X=1)=p\\) and \\(\\Pr(X=0)=1-p\\),\n\n\\[\\varphi_X(t)=1-p+pe^{it}.\\]\n\nConsequently, a binomial distribution with parameters \\(n\\) and \\(p\\), obtained by summing \\(n\\) independent Bernoulli variables, has characteristic function\n\n\\[\\varphi_X(t)=(1-p+pe^{it})^n.\\]\n\nNormal distribution\n\nFor a normal distribution with mean \\(\\mu\\) and variance \\(\\sigma^2\\),\n\n\\[\\varphi_X(t) = \\exp\\!\\left(i\\mu t-\\frac{\\sigma^2t^2}{2}\\right).\\]\n\nMultiplication of these functions shows directly that a sum of independent normal random variables is normal, with means and variances added.\n\nPoisson distribution\n\nFor a Poisson distribution with parameter \\(\\lambda\\),\n\n\\[\\varphi_X(t) = \\exp\\!\\left(\\lambda(e^{it}-1)\\right).\\]\n\nThe product formula immediately establishes that independent Poisson variables have a Poisson sum whose parameter is the sum of their parameters.\n\nCauchy distribution\n\nFor a Cauchy distribution with location \\(x_0\\) and scale \\(\\gamma>0\\),\n\n\\[\\varphi_X(t) = \\exp\\!\\left(ix_0t-\\gamma|t|\\right).\\]\n\nThe Cauchy distribution has neither a finite mean nor a finite variance, yet its characteristic function is well defined. Its lack of differentiability at zero illustrates the connection between smoothness of a characteristic function and existence of moments."
    },
    {
      "heading": "Uniqueness and inversion",
      "text": "Uniqueness theorem\n\nCharacteristic functions determine distributions uniquely:\n\n\\[\\varphi_X(t)=\\varphi_Y(t) \\quad\\text{for every }t\\in\\mathbb{R} \\quad\\Longleftrightarrow\\quad X\\text{ and }Y\\text{ have the same distribution}.\\]\n\nEquality in distribution does not mean that \\(X\\) and \\(Y\\) are equal as random variables on a shared probability space. It means that they assign the same probabilities to measurable subsets of \\(\\mathbb{R}\\).\n\nUniqueness makes characteristic functions useful for identifying distributions without calculating densities or distribution functions directly.\n\nRecovering a density\n\nIf the characteristic function is absolutely integrable,\n\n\\[\\int_{\\mathbb{R}}|\\varphi_X(t)|\\,dt<\\infty,\\]\n\nthen the distribution has a bounded continuous density given by\n\n\\[f_X(x) = \\frac{1}{2\\pi} \\int_{-\\infty}^{\\infty} e^{-itx}\\varphi_X(t)\\,dt.\\]\n\nAbsolute integrability is a sufficient condition for this form of inversion, not a necessary condition for a distribution to possess a density.\n\nRecovering interval probabilities\n\nA more general inversion formula applies without assuming a density. If \\(a<b\\) are continuity points of the cumulative distribution function \\(F\\), then\n\n\\[F(b)-F(a) = \\lim_{T\\to\\infty} \\frac{1}{2\\pi} \\int_{-T}^{T} \\frac{e^{-ita}-e^{-itb}}{it} \\varphi_X(t)\\,dt.\\]\n\nAt \\(t=0\\), the quotient is interpreted by its continuous extension. When endpoints carry positive probability, the corresponding general formula gives half weight to the endpoint masses."
    },
    {
      "heading": "Moments and derivatives",
      "text": "If \\(\\mathbb{E}[|X|^n]<\\infty\\), differentiation under the expectation gives\n\n\\[\\varphi_X^{(k)}(t) = \\mathbb{E}\\!\\left[(iX)^k e^{itX}\\right], \\qquad 1\\leq k\\leq n.\\]\n\nAt zero,\n\n\\[\\varphi_X^{(k)}(0) = i^k\\mathbb{E}[X^k].\\]\n\nIn particular,\n\n\\[\\mathbb{E}[X]=\\frac{\\varphi_X'(0)}{i}, \\qquad \\mathbb{E}[X^2]=-\\varphi_X''(0),\\]\n\nwhenever the corresponding absolute moments are finite. The variance is therefore\n\n\\[\\operatorname{Var}(X) = -\\varphi_X''(0) - \\left(\\frac{\\varphi_X'(0)}{i}\\right)^2.\\]\n\nThe converse requires care. Existence of an even-order derivative at zero implies finiteness of the corresponding even-order moment, but existence of an odd-order derivative alone need not imply finiteness of that absolute moment.\n\nEven when all moments exist, the formal series\n\n\\[\\sum_{n=0}^{\\infty} \\frac{(it)^n}{n!}\\mathbb{E}[X^n]\\]\n\nneed not converge to the characteristic function. Smoothness does not by itself imply analyticity, and distinct distributions can have identical moments despite having different characteristic functions."
    },
    {
      "heading": "Relation to other transforms",
      "text": "The moment-generating function is\n\n\\[M_X(s)=\\mathbb{E}[e^{sX}],\\]\n\nwherever this expectation is finite. Unlike characteristic functions, moment-generating functions need not be finite in any neighborhood of zero.\n\nBoth can be viewed as restrictions of a complex exponential transform when that transform is defined on an appropriate domain. The informal identity \\(\\varphi_X(t)=M_X(it)\\) requires such a complex extension; it is not automatically justified merely by defining \\(M_X\\) for real arguments.\n\nFor a nonnegative random variable, the Laplace transform of its distribution is\n\n\\[L_X(s)=\\mathbb{E}[e^{-sX}], \\qquad s\\geq0.\\]\n\nEach transform emphasizes different analytic properties. Characteristic functions have the particular advantage of existing for every probability distribution."
    },
    {
      "heading": "Convergence and limit theorems",
      "text": "The Lévy continuity theorem connects characteristic functions with convergence in distribution.\n\nIf \\(X_n\\) converges in distribution to \\(X\\), then\n\n\\[\\varphi_{X_n}(t)\\longrightarrow\\varphi_X(t)\\]\n\nfor every real \\(t\\).\n\nConversely, if \\(\\varphi_{X_n}(t)\\) converges pointwise to a function \\(\\varphi(t)\\) that is continuous at zero, then \\(\\varphi\\) is a characteristic function and the corresponding distributions converge to its distribution. Continuity at zero is essential: a pointwise limit without this property need not represent any probability distribution.\n\nCentral limit theorem\n\nCharacteristic functions give a standard proof of the central limit theorem. Let \\(X_1,X_2,\\ldots\\) be independent and identically distributed with mean \\(\\mu\\) and finite, positive variance \\(\\sigma^2\\). Set\n\n\\[Z=\\frac{X_1-\\mu}{\\sigma}.\\]\n\nFinite second moment gives the local expansion\n\n\\[\\varphi_Z(u)=1-\\frac{u^2}{2}+o(u^2) \\qquad (u\\to0).\\]\n\nFor the standardized sum\n\n\\[S_n= \\frac{\\sum_{j=1}^{n}X_j-n\\mu}{\\sigma\\sqrt n},\\]\n\nindependence yields\n\n\\[\\varphi_{S_n}(t) = \\left[\\varphi_Z\\!\\left(\\frac{t}{\\sqrt n}\\right)\\right]^n \\longrightarrow e^{-t^2/2}.\\]\n\nThe limit is the characteristic function of the standard normal distribution. The continuity theorem then establishes convergence in distribution."
    },
    {
      "heading": "Which functions are characteristic functions?",
      "text": "Normalization, boundedness, and continuity are not sufficient by themselves to characterize characteristic functions.\n\nA function \\(\\varphi:\\mathbb{R}\\to\\mathbb{C}\\) is positive definite if, for every finite selection \\(t_1,\\ldots,t_m\\in\\mathbb{R}\\) and \\(c_1,\\ldots,c_m\\in\\mathbb{C}\\),\n\n\\[\\sum_{j=1}^{m}\\sum_{k=1}^{m} c_j\\overline{c_k}\\, \\varphi(t_j-t_k) \\geq0.\\]\n\nFor a characteristic function, this follows from\n\n\\[\\sum_{j,k}c_j\\overline{c_k}\\varphi_X(t_j-t_k) = \\mathbb{E}\\!\\left[ \\left|\\sum_j c_j e^{it_jX}\\right|^2 \\right].\\]\n\nBochner’s theorem states that a continuous positive-definite function on \\(\\mathbb{R}\\) with \\(\\varphi(0)=1\\) is the characteristic function of a probability measure. This provides an intrinsic characterization without first specifying a random variable."
    },
    {
      "heading": "Multivariate characteristic functions",
      "text": "For a random vector \\(X\\in\\mathbb{R}^d\\), the characteristic function is\n\n\\[\\varphi_X(t) = \\mathbb{E}\\!\\left[e^{it^{\\mathsf T}X}\\right], \\qquad t\\in\\mathbb{R}^d.\\]\n\nIt uniquely determines the joint distribution. Every scalar projection \\(a^{\\mathsf T}X\\) has characteristic function\n\n\\[\\varphi_{a^{\\mathsf T}X}(u)=\\varphi_X(ua).\\]\n\nConsequently, the distributions of all scalar projections determine the joint distribution.\n\nThe components \\(X_1,\\ldots,X_d\\) are mutually independent exactly when\n\n\\[\\varphi_X(t_1,\\ldots,t_d) = \\prod_{j=1}^{d}\\varphi_{X_j}(t_j)\\]\n\nfor all \\(t\\).\n\nFor a multivariate normal distribution with mean vector \\(\\mu\\) and covariance matrix \\(\\Sigma\\),\n\n\\[\\varphi_X(t) = \\exp\\!\\left( it^{\\mathsf T}\\mu-\\frac12 t^{\\mathsf T}\\Sigma t \\right).\\]\n\nThis formula remains valid when \\(\\Sigma\\) is singular."
    },
    {
      "heading": "Statistical and computational uses",
      "text": "Given independent observations \\(X_1,\\ldots,X_n\\) from a common distribution, the empirical characteristic function is\n\n\\[\\widehat{\\varphi}_n(t) = \\frac1n\\sum_{j=1}^{n}e^{itX_j}.\\]\n\nFor each fixed \\(t\\), it is an unbiased estimator of \\(\\varphi_X(t)\\) and converges almost surely to it by the law of large numbers. It is also itself a characteristic function: that of the empirical distribution placing mass \\(1/n\\) at each observation.\n\nEmpirical characteristic functions are used in parameter estimation, goodness-of-fit testing, and tests of independence. They are particularly useful when a model has a simple characteristic function but a difficult or unavailable density.\n\nNumerical Fourier inversion can recover approximate densities or probabilities, sometimes using the fast Fourier transform. Practical limitations include truncation of the frequency range, discretization error, oscillatory integrals, and possible aliasing. Although a characteristic function uniquely specifies a distribution mathematically, recovering that distribution accurately from finitely many noisy transform values can be difficult."
    }
  ],
  "links": [
    "probability",
    "random-variable",
    "probability-distribution",
    "fourier-transform",
    "expected-value",
    "probability-density-function",
    "indicator-function",
    "uniform-continuity",
    "dominated-convergence-theorem",
    "statistical-independence",
    "convolution",
    "bernoulli-distribution",
    "normal-distribution",
    "poisson-distribution",
    "cauchy-distribution",
    "cumulative-distribution-function",
    "variance",
    "laplace-transform",
    "convergence-in-distribution",
    "central-limit-theorem",
    "joint-probability-distribution",
    "multivariate-normal-distribution",
    "covariance-matrix",
    "law-of-large-numbers",
    "fast-fourier-transform"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}