CHOQUET INTEGRATION WITH SUBMODULAR FUNCTION ON MEASURABLE SPACE WITH SIGMA-ALGEBRA GENERATING CHAIN
2025
Read CHOQUET INTEGRATION WITH SUBMODULAR FUNCTION ON MEASURABLE SPACE WITH SIGMA-ALGEBRA GENERATING CHAIN (2025) 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-2509-11033.txt.
Introduction
Tetsuya Hattori is the sole author of this 2025 mathematics paper on Choquet integration, a way of averaging with respect to set functions that need not behave like ordinary probability measures. He works on spaces whose collection of measurable sets is generated by a single chain of nested sets.
Building on a formula that writes a submodular set function as the largest value over a family of measures it dominates, Hattori derives the matching formula for the Choquet integral itself. As an application, he recovers a basic formula used in mathematical finance for risk measures that are law invariant and coherent, meaning they depend only on the distribution of losses. He also studies a recursion for set functions whose fixed point the new formula describes.
Catalog record
| YEAR | 2025 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-11033
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
CHOQUET INTEGRATION WITH SUBMODULAR FUNCTION ON MEASURABLE SPACE WITH SIGMA-ALGEBRA GENERATING CHAIN Abstract Based on a study of a formula representing submodular set function as a supremum of measures dominated by the set function, we present a corresponding formula for a Choquet integration with respect to the set function, on a measurable space which has a chain of measurable set generating the sigma-algebra. As an application we reproduce a basic formula in mathematical finance on law invariant coherent risk measures. We also study a recursion relation of set functions for which the representation formula characterizes the fixed point. † † 2020 Mathematics Subject Classification . Primary 60A10; Secondary 60Axx. † † Key words and phrases . Choquet integration, submodular function, convex game, risk measure, measurable space † † This work was supported by JSPS KAKENHI Grant Number 22K03358. 1 Introduction Let (\Omega,{\cal F}) be a measurable space, namely, a \sigma -algebra {\cal F} is a class of subsets of \Omega and is closed under complements and countable unions. For a measurable set A\in{\cal F} denote by {\cal F}|_{A} , the class of measurable sets restricted to A , and denote the set of finite measures on the measurable space (A,{\cal F}|_{A}) by {\cal M}(A) . For a real valued set function v:\ {\cal F}\to{\mathbb{R}} and a measurable set A\in{\cal F} , let {\cal C}_{-,v}(A) be a class of measures dominated by v on A ; (1) {\cal C}_{-,v}(A)=\{\mu\in{\cal M}(A)\mid\mu(A)=v(A),\ \mu(B)\leqq v(B),\ B\in{\cal F}|_{A}\}. If (2) v(B)=\sup_{\mu\in{\cal C}_{-,v}(A)}\mu(B), holds for all A,B\in{\cal F} satisfying B\subset A , then it is easy to see that (3) v(A)+v(B)\geqq v(A\cup B)+v(A\cap B),\ \ A,B\in{\cal F}, holds [ 5 , Proposition 1] . A set function which satisfies ( 3 ) is called a submodular function. In [ 5 , Theorem 3] we gave a proof of a converse that ( 3 ) implies ( 2 ), for a non-decreasing and continuous set function v , when {\cal F} is generated by a chain. Here, by non-decreasing we mean (4) v(A)\leqq v(B),\ A\subset B,\ A,B\in{\cal F}. We will later define the continuity of set functions which we adopt in this paper, and first make the assumption on {\cal F} precise. We consider, as in [ 5 ] , a following set of conditions for a class of measurable sets {\cal I}\subset{\cal F} ; (5) \left\{\begin{array}[]{l}\displaystyle i)\ \ \sigma[{\cal I}]={\cal F},\mbox{ where $\sigma[{\cal I}]$ denotes the smallest $\sigma$-algebra containing ${\cal I}$,}\\ \displaystyle ii)\ \ \emptyset\in{\cal I},\ \Omega\in{\cal I},\\ \displaystyle iii)\ \ {\cal I}\mbox{ is a chain, i.e., totally ordered with respect to inclusion,}\\ \displaystyle\phantom{iii)}\mbox{i.e., for all $I_{1},I_{2}\in{\cal I}$ either $I_{1}\subset I_{2}$ or $I_{2}\subset I_{1}$\,.}\end{array}\right. Let {\mathbb{\cal X}} denote the collection of such classes; (6) \begin{array}[]{l}\displaystyle{\mathbb{\cal X}}=\{{\cal I}\subset{\cal F}\mid\mbox{(\ref{e:sigmaalgebrawithtotallyorderedgeneratingclass}) holds}\}\\ \displaystyle\phantom{{\mathbb{\cal X}}}=\{{\cal I}\subset{\cal F}\mid\ {\cal I}\mbox{ is a chain such that }\ \emptyset\in{\cal I},\ \Omega\in{\cal I},\ \sigma[{\cal I}]={\cal F}\},\end{array} and for {\cal I}\in{\mathbb{\cal X}} and A\in{\cal F} , define {\cal I}_{A} , the insertion of A into {\cal I} , by (7) {\cal I}_{A}=\{A\cap I\mid I\in{\cal I}\}\cup\{A\cup I\mid I\in{\cal I}\}. Note that {\mathbb{\cal X}} is closed under the insertion; if {\cal I}\in{\mathbb{\cal X}} and A\in{\cal F} then \displaystyle A\in{\cal I}_{A}\in{\mathbb{\cal X}} [ 5 , Lemma 2] . With these notations, [ 5 , Proposition 1] and [ 5 , Theorem 3] imply that if {\mathbb{\cal X}}\neq\emptyset , then a non-decreasing continuous set function v is submodular, if and only if ( 2 ) holds for all A,B\in{\cal F} satisfying B\subset A . See [ 3 , 4 , 8 ] and the references in [ 5 ] for a part of large background and long history of the basic theory of submodular functions. It is proved in [ 5 , Proposition 7] that Polish spaces (separable, complete, metric space) with Borel \sigma -algebra (the smallest \sigma -algebra containing all open balls) are examples of measurable spaces satisfying {\mathbb{\cal X}}\neq\emptyset , so that the above mentioned equivalence between submodular property ( 3 ) and the representation formula ( 2 ) hold for Polish spaces. One dimensional Borel \sigma -algebra and a finite set \displaystyle\Omega_{m}=\{1,2,\ldots,m\} with m elements and \displaystyle{\cal F}=2^{\Omega_{m}} are among examples of Polish spaces, as well as a wide class of spaces extensively used in the theory of stochastic processes. In this paper we exclusively consider (\Omega,{\cal F}) satisfying {\mathbb{\cal X}}\neq\emptyset , and advance our study in [ 5 ] in the following directions. • The definition ( 1 ) of {\cal C}_{-,v}(A) implies that ( 2 ) follows from ( 3 ) if there exists a measure \mu\in{\cal C}_{-,v}(A) such that v(B)=\mu(B) holds. Such a measure is studied by Shapley [ 8 ] as an extremal measure when \Omega is a finite set. In our formulation, the extremal measure of v is defined for each {\cal I}\in{\mathbb{\cal X}} . (In [ 5 , Theorem 3] , for the sake of conciseness of the statement of the theorem, the notion of extremal measures is not explicitly written, and is implicitly introduced in the proof of the theorem.) For a non-decreasing set function v , we define \displaystyle\mu_{v,{\cal I}}:\ {\cal I}\to{\mathbb{R}_{+}} by (8) \mu_{v,{\cal I}}(I)=v(I),\ I\in{\cal I}. To extend \displaystyle\mu_{v,{\cal I}} to a measure, we define in Definition 2.1 in § 2 , that a set function v is continuous, if for every {\cal I}\in{\mathbb{\cal X}} \displaystyle\mu_{v,{\cal I}} , defined by ( 8 ) on {\cal I} and uniquely extended to the finite algebra {\cal J} generated by {\cal I} as a finitely additive measure, is continuous on {\cal J} in the standard measure theory sense. (We will see in § 2 that this definition is a generalization of the definition of continuity given in [ 5 ] for submodular functions.) If v is continuous then \displaystyle\mu_{v,{\cal I}} is uniquely extended to a measure on {\cal F} , which we call the extremal measure of v corresponding to {\cal I}\in{\mathbb{\cal X}} . We prove in § 2 another representation formula for submodular function in terms of the extremal measures (9) v=\sup_{{\cal I}\in{\mathbb{\cal X}}}\mu_{v,{\cal I}} which generalizes a corresponding result of theory of cores for a finite set in [ 8 ] to measurable spaces satisfying {\mathbb{\cal X}}\neq\emptyset . • We can define Choquet integration v(f) with respect to non-decreasing continuous set function v for an integrable function f . If, in addition, v is submodular, the functional \rho defined by \rho(f)=v(-f) satisfies the definition of the coherent risk measure studied in the field of mathematical finance. We then obtain a representation formula for \rho , based on ( 9 ), which reproduces a basic formula for a coherent risk measure with Fatou property studied in mathematical finance [ 1 , 2 , 6 ] . See [ 7 ] for generalization and further development on Choquet integration with respect to submodular functions including its relation to coherent risk measure. In [ 5 ] we briefly announced an outline of how these results are formulated in our framework, for which we give precise statements and proofs in § 3 . As a further example in § 4 , we reproduce in our framework a formula studied in [ 6 ] for law-invariant coherent risk measures. • Suggested by the representation formula ( 9 ) for submodular functions, we