When is the operation τ T , L \tau_{{}_{T,L}} a triangle function on Δ + \Delta^{+} ?

Hongliang Lai, Mengyu Luo, Jie Zhang · 2025

Read When is the operation τ T , L \tau_{{}_{T,L}} a triangle function on Δ + \Delta^{+} ? by Hongliang Lai, Mengyu Luo, Jie Zhang (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-2510-26439.txt.

Introduction

Hongliang Lai, Mengyu Luo, and Jie Zhang settle a question left open since 1983 by Schweizer and Sklar. The question concerns an operation, written tau with subscripts T and L, built from two simpler operations on distribution functions. The trio ask when that operation counts as a triangle function, a kind of generalised addition for such functions, on a set written Delta-plus.

Lai and his two co-authors prove the operation qualifies exactly when three conditions hold at once. First, L must be a continuous t-conorm, a generalised or, with an extra stated trait. Second, T must be a t-norm, a generalised and. Third, T must be weakly left continuous, with full left continuity when L is of a certain kind. Closing a forty-year-old gap gives this corner of probability theory a firm answer.

Catalog record

CREATORHongliang Lai, Mengyu Luo, Jie Zhang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-26439

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

When is the operation τ T , L \tau_{{}_{T,L}} a triangle function on Δ + \Delta^{+} ? Hongliang Lai, Mengyu Luo, Jie Zhang Abstract This paper resolves an open problem posed by Schweizer and Sklar in 1983. We establish that the binary operation τ T , L \tau_{{}_{T,L}} is a triangle function on Δ + \Delta^{+} if and only if the following three conditions hold: (a) L L is a continuous t-conorm on [ 0 , ∞ ] [0,\infty] satisfying ( L ​ C ​ S ) (LCS) ; (b) T T is a t-norm on [ 0 , 1 ] [0,1] ; and (c) T T is weakly left continuous, with left continuity required when L L is non-Archimedean. 1 Introduction The notion of probabilistic metric spaces was introduced by Menger [ 3 ] in 1942 as a generalization of classical metric spaces. In this framework, the distance between two points is described by a distribution function rather than a deterministic real number. To model the “addition” of distances in the interval [0,\infty] , Šerstnev [ 6 ] later introduced the concept of triangle functions, refining the triangle inequality originally proposed by Menger and Wald [ 7 ] . Motivated by Šerstnev’s formulation, Schweizer and Sklar [ 5 ] studied a family of operations denoted by \tau_{T,L} , constructed from a binary operation T on [0,1] and a binary operation L on [0,\infty] . They established sufficient conditions under which \tau_{T,L} becomes a (continuous) triangle function. However, by presenting a counterexample—the triangle function \tau_{{}_{D}} —they also demonstrated that these conditions are not necessary. This led them to pose the following open problem (see 7.9.6 in [ 5 ] ): characterize all pairs (T,L) for which \tau_{T,L} is a (continuous) triangle function. In 1992, Ying [ 8 ] derived a set of necessary conditions that closely approximate the known sufficient ones, yet the sufficiency of those conditions remained open. In this paper, we completely resolve this long-standing problem. We prove that \tau_{T,L} is a triangle function on \Delta^{+} if and only if the following three conditions are satisfied: (a) L is a continuous t-conorm on [0,\infty] satisfying the (LCS) condition; (b) T is a t-norm on [0,1] ; (c) T is weakly left continuous, and left continuous whenever L is non-Archimedean. While the necessity of conditions (a)–(c) was already established by Ying [ 8 ] , the sufficiency part—especially the associativity of \tau_{T,L} —presents a nontrivial difficulty. Specifically, when L is Archimedean and T is only weakly left continuous, the argument used in Theorem 7.2.4 of [ 5 ] no longer applies. By introducing a subtle technical argument, we successfully establish the associativity of \tau_{T,L} under the weakened continuity condition (see Lemma 3.18 and Proposition 3.19 ), thereby completing the sufficiency proof. The rest of this paper is organized as follows. In Section 2, we review basic concepts and results on t-norms, t-conorms, and triangle functions, mainly following [ 2 , 5 ] , and give a precise statement of Schweizer and Sklar’s open problem. Section 3 is devoted to the proof of our main results. 2 Preliminaries The terminology and notation used in this paper are primarily drawn from the literature [ 5 ] . Definition 2.1 . ( [ 5 ] ) A distance distribution function is a map \varphi\colon[0,\infty]\>\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 0.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces{\hbox{\kern 4.49588pt\raise 0.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces{\hbox{\kern 4.49588pt\raise 0.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 14.99176pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces}}}}\>[0,1] such that \varphi(0)=0 , \varphi(\infty)=1 , and \varphi is increasing and left-continuous on ]0,\infty[ . Let \Delta^{+} denote the set of all distance distribution functions. The partial order of \Delta^{+} is defined pointwisely. Among distance distribution functions, the following will be needed frequently. • For each r<\infty , define \varepsilon_{r}(x)=\begin{cases}0,&0\leqslant x\leqslant r;\\ 1,&r<x.\end{cases} • For r=\infty , define \varepsilon_{\infty}(x)=\begin{cases}0,&0\leqslant x<\infty;\\ 1,&x=\infty.\end{cases} • For each p\in[0,1] , define V_{p}(x)=\begin{cases}0,&x=0;\\ p,&0<x<\infty;\\ 1,&x=\infty.\end{cases} The mappings r\mapsto\varepsilon_{r} and p\mapsto V_{p} define embeddings into \Delta^{+} with nice order-theoretic properties: (1) The map r\mapsto\varepsilon_{r} sends the extended interval [0,\infty] into \Delta^{+} . It converts suprema in [0,\infty] into infima in \Delta^{+} , and infima in [0,\infty] into suprema in \Delta^{+} . (2) The map p\mapsto V_{p} sends the unit interval [0,1] into \Delta^{+} . It preserves all suprema and infima. In particular, the endpoints satisfy \varepsilon_{0}=V_{1} and \varepsilon_{\infty}=V_{0} . Definition 2.2 . ( [ 5 ] ) A triangle function \tau is a binary operation on \Delta^{+} that is commutative, associative, increasing in each place and has \varepsilon_{0} as identity. Definition 2.3 . ( [ 5 ] ) The class \mathcal{T} is the set of all binary operations T on [0,1] that is increasing in each place and has 1 as identity. A triangular norm (briefly, a t-norm ) is a binary operation in \mathcal{T} that is commutative and associative. Let T be a binary operation in \mathcal{T} . T is continuous if it is a continuous binary function on [0,1]^{2} . T is left continuous if T(x,y)=\sup\{T(u,v)|u<x,v<y\} for all x,y\in]0,1] . T is weakly left continuous if T(x,y)=\sup\{T(u,v)|u\leqslant x,v<y\text{ or }u<x,v\leqslant y\} for all x,y\in]0,1] . There are some basic t-norms on the unit interval [0,1] in the following. • Minimum: M(x,y)=\min\{x,y\} ; • Product: \Pi(x,y)=x\cdot y ; • Łukasiewicz t-norm: W(x,y)=\max\{x+y-1,0\} ; • Nilpotent minimum: {{nM}}(x,y)=\begin{cases}\min\{x,y\},&1<x+y,\\ 0,&x+y\leqslant 1;\end{cases} • Drastic product: D(x,y)=\begin{cases}\min\{x,y\},&x=1\text{ or }y=1,\\ 0,&\text{otherwise}.\end{cases} The drastic product D is the minimal t-norm and the minimum M is the maximal t-norm. In fact, for any binary operation T\in\mathcal{T} , it always holds that D\leqslant T\leqslant M. The three t-norms M , \Pi and W are continuous. The t-norm nM is left continuous but not continuous. The t-norm D is weakly left continuous but not left continuous. By changing the values of the triangular norm nM on its anti-diagonal, we obtain the following t-norm: \widehat{nM}(x,y)=\begin{cases}\min\{x,y\},&1\leqslant x+y,\\ 0,&x+y<1.\end{cases} Then \widehat{nM} is not weakly left continuous at the point (0.5,0.5) . Definition 2.4 . ( [ 5 ] ) The class \mathcal{L} is the set of all binary operations L on [0,\infty] that satisfy the conditions: (L1) L is onto, (L2) L is increasing in each place, and (L3) L is continuous except possibly at the points (0,\infty) and (\infty,0) . Proposition 2.5 . If an operation L\in\mathcal{L} has 0 as identity, then it is continuous. Proof. Since L is increasing in each variable and has 0 as identity, we observe that \lim_{x\rightarrow 0\atop y\rightarrow\infty}L(x,y)=\infty=L(0,\infty),\quad\lim_{x\rightarrow\infty\atop y\rightarrow 0}L(x,y)=\infty=L(\infty,0). This shows that L is continuous at the points (0,\infty) and (\infty,0) . Since L is already continuous at other points, we conclude that L is continuous on the entire domain [0,\infty]^{2} . ∎ A binary operation L\in\mathcal{L} is said to be strictly increasing (see [ 4 , 5 ] ) if u<u^{\prime},v<v^{\prime}\implies L(u,v)<L(u^{\prime},v^{\prime}). [formula] It is said to be conditionally strictly increasing (see [ 8 ] ) if u<u^{\prime},v<v^{\prime},L(u^{\prime},v^{\prime})<\infty\implies L(u,v)<L(u^{\prime},v^{\prime}). [formula] A triangular conorm (shortly, a t-conorm ) on [0,\infty] is an associative binary operation L on [0,\infty] that is commutative, increasing in each place, and has 0 as identity. A t-conorm L is continuous if L is a continuous function with respect to the usual topology. There is a one-to-one correspondence between t-norms on [0,1] and t-conorms on [0,\infty] by an order isomorphism \varphi:([0,1],\leqslant)\longrightarrow([0,\infty],\geqslant)

Read the complete text