A Characterization of the Cumulants as Continuous Moment-Based Statistics
Sofia de la Cerda · 2026
Read A Characterization of the Cumulants as Continuous Moment-Based Statistics by Sofia de la Cerda (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-29615.txt.
Introduction
de la Cerda characterizes cumulants, classical quantities built from the moments of a random variable. Moments are expected values of powers of the variable. Cumulants are polynomial functions of those moments, set apart by additivity: the cumulant of a sum of independent variables is the sum of their cumulants.
She proves the converse. Any statistic that depends smoothly on finitely many moments, and is additive for independent sums, must be expressible as a linear combination of cumulants. Her proof recasts the problem through the Hurwitz product and a change of coordinates. A more general theorem by Mattner already implies the result, as de la Cerda notes openly. Her contribution is a basic, self-contained route showing why cumulants are forced.
Catalog record
| CREATOR | Sofia de la Cerda |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.29615
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Characterization of the Cumulants as Continuous Moment-Based Statistics Sofia de la Cerda Abstract Cumulants are classical statistics associated with a random variable, defined as polynomial functions of its moments and distinguished by their additivity under convolution of distributions. A statistic is the name given to a function of a random variable, and a moment-based statistic is one that depends only on the moments ( 𝔼 [ X n ] ) n ∈ ℕ (\mathbb{E}[X^{n}])_{n\in\mathbb{N}} . We prove a converse: any statistic depending continuously on finitely many moments and additive for independent sums must be a linear combination of cumulants. The proof uses an algebraic reformulation of the problem via the Hurwitz product and a linearizing change of coordinates. This result also follows from the more general theorem of Mattner [ 4 ] , but our approach is elementary and self-contained. 1 Introduction The cumulants (\kappa_{n})_{n\in\mathbb{N}} are real functions defined in the space of random variables with finite n -th moment, i.e., \kappa_{n}:L^{n}\to\mathbb{R} . They are so named because they are additive in the sense that \kappa_{n}(X+Y)=\kappa_{n}(X)+\kappa_{n}(Y) for independent random variables X,Y . They are defined by the relation \log\left(\mathbb{E}\left[e^{zX}\right]\right)=\sum_{\ell=1}^{n}\frac{z^{\ell}}{\ell!}\kappa_{\ell}(X)+o(z^{n})\quad\textup{as }z\to 0 for z\in i\mathbb{R} . Note that this does not necessarily hold for z\in\mathbb{C} . For concreteness, we note that \kappa_{1}(X)=\mathbb{E}[X] coincides with the expectation, \kappa_{2}(X)=\mathbb{E}[X^{2}]-\mathbb{E}[X]^{2} is the variance, and \kappa_{3}(X)=\mathbb{E}[X^{3}]-3\mathbb{E}[X^{2}]\mathbb{E}[X]+2\mathbb{E}[X]^{3} is finally a new function, and \kappa_{4}(X)=\mathbb{E}[X^{4}]-4\mathbb{E}[X^{3}]\mathbb{E}[X]-3\mathbb{E}[X^{2}]^{2}+12\mathbb{E}[X^{2}]\mathbb{E}[X]^{2}-6\mathbb{E}[X]^{4}. For a normal distribution, the cumulants of order 3 and higher all vanish, a property that characterizes normality [ 3 ] . This vanishing property motivates thinking of higher-order cumulants as measures of deviation from normality. In particular \kappa_{3} is a measure of asymmetry, and \kappa_{4} is a measure of tailedness. The dilation-invariant quotient \kappa_{3}/\kappa_{2}^{3/2} is known as skewness, and \kappa_{4}/\kappa_{2}^{2} is known as kurtosis (in some texts excess kurtosis), from the Greek κυρτ´ς meaning convex or curved. Additivity follows from X,Y independent implies e^{zX},e^{zY} independent implies \displaystyle\sum_{\ell=1}^{n}\frac{z^{\ell}}{\ell!}\kappa_{\ell}(X+Y)+o(z^{n}) \displaystyle=\log\left(\mathbb{E}\left[e^{z(X+Y)}\right]\right) \displaystyle=\log\left(\mathbb{E}\left[e^{zX}\right]\mathbb{E}\left[e^{zY}\right]\right) \displaystyle=\log\left(\mathbb{E}\left[e^{zX}\right]\right)+\log\left(\mathbb{E}\left[e^{zY}\right]\right) \displaystyle=\sum_{\ell=1}^{n}\frac{z^{\ell}}{\ell!}(\kappa_{\ell}(X)+\kappa_{\ell}(Y))+o(z^{n}) thus \kappa_{\ell}(X+Y)=\kappa_{\ell}(X)+\kappa_{\ell}(Y) for every \ell\in[n]=\{1,\dots,n\} . To relate the cumulants to the moments, recall Faà di Bruno’s formula \frac{d^{n}}{dx^{n}}f(g(x))=\sum_{\ell=1}^{n}f^{(\ell)}(g(x))B_{n,\ell}(g^{\prime}(x),\dots,g^{(n-\ell+1)}(x)) where B_{n,\ell} is an incomplete Bell polynomial — defined as \displaystyle B_{n,\ell}(x_{1},\dots,x_{n-\ell+1})=\sum_{\begin{subarray}{c}j_{1},\dots,j_{n-\ell+1}\in\mathbb{N}\\ j_{1}+\dots+j_{n-\ell+1}=\ell\\ j_{1}+2j_{2}+\dots+(n-\ell+1)j_{n-\ell+1}=n\end{subarray}}\frac{n!}{j_{1}!\dots j_{n-\ell+1}!}\left(\frac{x_{1}}{1!}\right)^{j_{1}}\left(\frac{x_{2}}{2!}\right)^{j_{2}}\dots \displaystyle\left(\frac{x_{n-\ell+1}}{(n-\ell+1)!}\right)^{j_{n-\ell+1}}. An application of the formula for f=\exp and g(z)=\log(\mathbb{E}[e^{zX}]) gives \displaystyle\frac{d^{n}}{dz^{n}}\Bigg\rvert_{z=0}\mathbb{E}[e^{zX}] \displaystyle=\sum_{\ell=1}^{n}e^{g(0)}B_{n,\ell}(\kappa_{1}(X),\dots,\kappa_{n-\ell+1}(X)) \displaystyle\mathbb{E}[X^{n}] \displaystyle=\sum_{\ell=1}^{n}B_{n,\ell}(\kappa_{1}(X),\dots,\kappa_{n-\ell+1}(X)), i.e., the moments of X can be expressed as a polynomial in its cumulants. The polynomials B_{n}=\sum_{\ell=1}^{n}B_{n,\ell} thus give the moments as functions of the cumulants, and they are called exponential Bell polynomials. Another application of the formula for f=\log and g(z)=\mathbb{E}[e^{zX}] gives \displaystyle\frac{d^{n}}{dz^{n}}\Bigg\rvert_{z=0}\log\left(\mathbb{E}\left[e^{zX}\right]\right) \displaystyle=\sum_{\ell=1}^{n}(-1)^{\ell-1}(\ell-1)!\frac{1}{g(0)^{\ell}}B_{n,\ell}(\mathbb{E}[X],\dots,\mathbb{E}[X^{n-\ell+1}]) \displaystyle\kappa_{n}(X) \displaystyle=\sum_{\ell=1}^{n}(-1)^{\ell-1}(\ell-1)!B_{n,\ell}(\mathbb{E}[X],\dots,\mathbb{E}[X^{n-\ell+1}]), i.e., the cumulants are expressed as a polynomial in the moments via the logarithmic Bell polynomials \overline{B}_{n}=\sum_{\ell=1}^{n}(-1)^{\ell-1}(\ell-1)!B_{n,\ell} . This is a particular case of the relation of formal power series \log\left(1+\sum_{\ell=1}^{n}\frac{x_{\ell}}{\ell!}z^{\ell}\right)=\sum_{\ell=1}^{n}\frac{\overline{B}_{\ell}(x_{1},\dots,x_{\ell})}{\ell!}z^{\ell}+O(z^{n+1}). (1) It is remarkable that the function (x_{1},\dots,x_{n})\mapsto(B_{1}(x_{1}),\dots,B_{n}(x_{1},\dots,x_{n})) is (entry-wise) polynomial with a polynomial inverse (x_{1},\dots,x_{n})\mapsto(\overline{B}_{1}(x_{1}),\dots,\overline{B}_{n}(x_{1},\dots,x_{n})) . We are now ready to state the main theorem. Formally, it is not an original result as it follows from [ 4 , Theorem 1.11] , however, to the best of the author’s knowledge, this specific formulation and its proof are new. The historical background is further discussed in section 2 . Theorem 1.1 . Let F:L^{n}\to\mathbb{R} be such that there exists a continuous function f:\mathbb{R}^{n}\to\mathbb{R} with F(X)=f(\mathbb{E}[X],\dots,\mathbb{E}[X^{n}]) . Suppose that F is additive, i.e., for all X,Y\in L^{n} independent, F(X+Y)=F(X)+F(Y) . Then F is a linear combination of cumulants of order at most n , equivalently f is (in the appropriate domain) a linear combination of logarithmic Bell polynomials. Requiring that F depend only on the moments is not that significant a restriction, since under mild conditions a distribution is uniquely determined by its moments. Moreover, the hypotheses of the theorem are weak — continuity and additivity — and its consequences are strong — polynomial and even more rigidly a finite family given explicitly. The fundamental observation in the proof is that addition of independent random variables, at the level of moments, is an algebraic operation (the Hurwitz product) involving binomial expansions. After encoding additivity in these terms, we find a system of coordinates that transforms this product into ordinary addition in \mathbb{R}^{n} . The problem then reduces to solving a Cauchy functional equation under hypotheses that imply linearity. The cumulants appear as a basis associated to this linearization. Before proceeding, define the set of moment sequences \Omega=\big\{(\mathbb{E}[X],\dots,\mathbb{E}[X^{n}]):X\in L^{n}\big\} so we can outline the proof: 1. Prove that \Omega has non-empty interior; 2. Define the Hurwitz product * on \mathbb{R}^{n} and observe that the additivity hypothesis of the main theorem is equivalent to f(x*y)=f(x)+f(y) ; 3. The Hurwitz product gives \mathbb{R}^{n} the structure of an abelian group which is isomorphic to the usual \mathbb{R}^{n} via the function \varphi , i.e., \varphi(x+y)=\varphi(x)*\varphi(y) ; 4. f\circ\varphi extends to a linear functional on \mathbb{R}^{n} ; 5. The compositions \overline{B}_{n}\circ\varphi span the space of linear functionals. 2 Previous Work The cumulants were introduced by Thiele, and the reader can find a translated version of his paper in [ 1 ] . Rota [ 5 ] also credits him with the following characterization of the cumulants: Sometime in the last [19th] century, the Danish statistician Thiele observed that the variance of a random variable, namely \text{Var}(X)=\mathbb{E}[X^{2}]-\mathbb{E}[X]^{2} , possesses notable properties, to wit: 1. It is invariant under translation: \text{Var}(X+c)=\text{Var}(X) for any constant c . 2. If X and Y are independent random variables, then \text{Var}(X+Y)=\text{Var}(X)+\text{Var}(Y) . 3. \text{Var}(X) is a polynomial in the moments of the random variable X . He then proceeded to determine all nonlinear functionals of a random variable which have the same properties. Thiele’s works predate modern notation, and we could not find a modern statement of the polynomial characterization in the literature. His characterization (as described by Rota) is similar to ours, with the main difference being that he considers only functions of the form f(\mathbb{E}[X],\dots,\mathbb{E}[X^{n}]) for polynomial f , where we considered any continuous f . Another far-reaching characterization was found by Mattner [ 4 ] . He gives the space of probability measures with all moments finite a topology, and proves that functions that are additive and continuous with respect to that topology are linear combinations of cumulants. He chooses a strong topology, so continuity is a weak assumption of f , and in fact our result can be obtained as a particular case of his. His proof uses functional and Fourier analytic techniques. The present work may thus be viewed as an intermediate result between Thiele’s and Mattner’s characterizations. 3 The Hamburger Moment Problem The Hamburger moment problem is: which sequences x=(x_{\ell})_{\ell\in[n]} are moment sequences, insofar as there exists a random variable X such that \mathbb{E}[X^{\ell}]=x_{\ell}\>\forall\>\ell\in[n] , equivalently x\in\Omega ? Is this random variable unique up to distribution? The first question has a rather simple answer, and the second one is beyond the scope of this paper, but very interesting. Its answer is negative in general, but positive under mild restrictions (e.g. Carleman’s condition) under which the corresponding random variables are characterized by their moments. In this context, the hypothesis of the main theorem that F may only depend on the moments appears less restrictive than at first glance. Theorem 3.1 ( [ 6 , Theorem 3.8] ) . A sequence (x_{\ell})_{\ell\in[n]} is a moment sequence if and only if