The characterizations of monotone functions which generate associative functions

Basic Information

• Paper ID: 2409.02941
• Title: The characterizations of monotone functions which generate associative functions
• Authors: Meng Chen, Yun-Mao Zhang, Xue-ping Wang (School of Mathematical Sciences, Sichuan Normal University)
• Classification: math.GM (General Mathematics)
• Publication Date: September 2024 (arXiv v3: November 3, 2025)
• Paper Link: https://arxiv.org/abs/2409.02941

Abstract

This paper investigates the characterization problem of associative functions generated by monotone functions. Specifically, it considers a binary function $T: [0,1]^2 \rightarrow [0,1]$ defined as $T(x,y)=f^{(-1)}(F(f(x),f(y)))$, where $F:[0,\infty]^2\rightarrow[0,\infty]$ is an associative function, $f: [0,1]\rightarrow [0,\infty]$ is a monotone function satisfying specific conditions, and $f^{(-1)}$ is the pseudo-inverse of $f$. The paper proves that the associativity of $T$ depends solely on the properties of the range of $f$, and provides necessary and sufficient conditions for $T$ to be associative.

Research Background and Motivation

1. Problem to be Solved

The paper investigates how to construct new associative functions $T$ from a monotone function $f$ and an associative function $F$, with particular focus on what types of monotone functions $f$ guarantee that the constructed $T$ possesses associativity.

2. Importance of the Problem

• Theoretical Significance: Construction methods for associative functions play an indispensable role in the theory of solving associative equations
• Historical Origins: This idea traces back to the classical work of Abel (1826)
• Applied Value: Triangular norms (t-norms) and their generator theory have wide applications in fuzzy logic, probabilistic metric spaces, and other fields

3. Limitations of Existing Methods

• Schweizer & Sklar (1961), Ling (1965): Studied continuous strictly decreasing additive generators
• Klement, Mesiar & Pap (2000): Required generators $f$ to be strictly decreasing and satisfy $f(x) + f(y) \in \text{Ran}(f) \cup [f(0),\infty]$
• Viceńık (2005): Considered strictly monotone functions
• Zhang & Wang (2024): Studied monotone right-continuous functions
• Limitations: These works imposed strong restrictions on $f$ (strict monotonicity, right-continuity, etc.) and required specific range conditions

4. Research Motivation

This paper aims to generalize the above results to more general classes of monotone functions, relaxing the strictness and continuity requirements on $f$, requiring only that $f$ satisfies weaker conditions: when $f(x^+) \in \text{Ran}(f)$ then $f(x) = f(x^+)$, or when $f(x^+) \notin \text{Ran}(f)$ then $f(x) \neq f(y)$ for all $y \neq x$.

Core Contributions

1. Introduction of New Function Class $\mathcal{F}$: Defines a broader class of monotone functions than strict monotonicity and right-continuity, containing both as special cases
2. Complete Characterization of Range: Proves that the range of functions in class $\mathcal{F}$ can be uniquely represented by a pair $(S,C)$, where $S$ is a system of closed intervals and $C$ is a countable set of points (Lemma 3.1)
3. Key Equivalence Theorem: Proves that the associativity of $T$ is equivalent to the associativity of the operation $\otimes$ defined on the range $M = \text{Ran}(f)$ (Theorem 4.1), showing that associativity depends solely on range properties
4. Necessary and Sufficient Condition I: When $F$ is cancellative and satisfies $F(M, M\backslash C) \subseteq M\backslash C$, provides necessary and sufficient conditions for $T$ to be associative: $F(\cup_{k\in K}I_k, M) \cap M = \emptyset$ and $F(M, \cup_{k\in K}I_k) \cap M = \emptyset$ (Theorem 5.1)
5. Necessary and Sufficient Condition II: When $F(C,M) \cup F(M,C) \subseteq M\backslash C$, $T$ is associative if and only if the $F$-condition holds (Theorem 5.2)
6. Generalization of Existing Results: Unifies and generalizes the work of Viceńık, Zhang & Wang, Yao Ouyang, and others

Detailed Methodology

Task Definition

Input:

• Associative function $F: [0,\infty]^2 \rightarrow [0,\infty]$
• Monotone function $f: [0,1] \rightarrow [0,\infty]$ belonging to function class $\mathcal{F}$

Output:

• Function $T: [0,1]^2 \rightarrow [0,1]$ defined as $T(x,y) = f^{(-1)}(F(f(x), f(y)))$

Objective: Characterize the properties of function $f$ that make $T$ associative

Core Theoretical Framework

1. Definition of Function Class $\mathcal{F}$

$\mathcal{F}$ contains all non-decreasing functions $f: [0,1] \rightarrow [0,\infty]$ satisfying:

• When $f(x^+) \in \text{Ran}(f)$, then $f(x) = f(x^+)$
• When $f(x^+) \notin \text{Ran}(f)$, then $f(x) \neq f(y)$ for all $y \neq x$

Key Property: This condition ensures that $f(f^{(-1)}(x)) = G_M(x)$ for all $x \in [0,\infty]$ (Lemma 4.1)

2. Range Representation Theorem (Lemma 3.1)

For $M \in \mathcal{A}$ with $M \neq [0,\infty]$, there exist uniquely determined:

• System of closed intervals $S = \{[b_k, d_k] \subseteq [0,\infty] \mid k \in K\}$ (countable), where intervals are either disjoint or intersect only at endpoints
• Countable set of points $C = \{c_k \in [0,\infty] \mid k \in K\}$, satisfying $[b_k, d_k] \cap C \in \{\{b_k\}, \{d_k\}, \{b_k, d_k\}\}$

such that: $M = \{c_k \in [0,\infty] \mid k \in K\} \cup \left([0,\infty] \backslash \bigcup_{k\in K}[b_k, d_k]\right)$

Intuitive Understanding: The range $M$ is obtained by removing some open or half-open intervals from $[0,\infty]$, then adding back certain endpoints of these intervals.

3. Function $G_M$ and Operation $\otimes$

Definition of $G_M$ (Definition 4.1): $G_M(x) = \max\{M \cap \{\sup([0,x] \cap M), \inf([x,\infty] \cap M)\}\}$

Properties (Proposition 4.1):

• $G_M(x) = x$ if and only if $x \in M$
• $G_M$ is a non-decreasing function
• When $x \in [b_k, d_k]$ and $x \notin M$, then $G_M(x) = d_k$ if $d_k \in M$, otherwise $G_M(x) = b_k$

Definition of Operation $\otimes$ (Definition 4.2): $x \otimes y = G_M(F(x,y))$

4. Core Equivalence Theorem

Theorem 4.1: $T$ is associative $\Leftrightarrow$ $\otimes$ is associative

Proof Strategy:

1. Define auxiliary function $f_*: B \rightarrow [0,\infty]$ ($B$ is the "essential domain" of $f$)
2. Define $F_0: B^2 \rightarrow B$ as $F_0(x,y) = f^{(-1)}(F(f_*(x), f_*(y)))$
3. Prove $F_0$ is associative $\Leftrightarrow$ $\otimes$ is associative (Proposition 4.3)
4. Prove $F_0$ is associative $\Leftrightarrow$ $T$ is associative (Proposition 4.4)

Technical Innovations

1. Range-Driven Analysis Framework

Unlike traditional methods focusing on properties of $f$ itself, this paper discovers that associativity depends solely on the structure of the range $M = \text{Ran}(f)$, which is a profound insight.

2. Introduction of Interval Sets $I_k$

For each $k \in K$, define: $I_k = O(\{a_k\} \cup \{z \in [b_k, d_k]\backslash\{c_k\} \mid \exists x,y \in M: F(x,y) = z\})$ where $a_k$ is the value of $x \otimes y$ when there exist $x,y \in M$ such that $F(x,y) \in [b_k, d_k]\backslash\{c_k\}$.

These sets capture how the operation $F$ maps elements of $M$ into "gap" intervals.

3. Fine Characterization of $F$-Condition

Define three classes of sets:

• $J_1(M) = \bigcup_{y\in M}\bigcup_{k\in K} F(I_k^y, M^y)$
• $J_2(M) = \bigcup_{y\in M}\bigcup_{k\in K} F(M_y, I_k^y)$
• $J_3(M) = \bigcup_{y\in M}\bigcup_{k,l\in K} J_{k,l}^y$

The $F$-condition requires for all $k,l \in K, y \in M$:

• (C1) $H_k^y = \emptyset$ or $F(I_k^y, M^y) \cap (M\backslash C) = \emptyset$
• (C2) $H_k^y = \emptyset$ or $F(M_y, I_k^y) \cap (M\backslash C) = \emptyset$
• (C3) $H_{k,l}^y = \emptyset$ or $J_{k,l}^y \cap (M\backslash C) = \emptyset$

These conditions precisely characterize how the behavior of $F$ at range "gaps" affects associativity.

Experimental Setup

Dataset

This is pure theoretical mathematics research with no experimental datasets involved.

Examples and Counterexamples

The paper validates the theory through numerous carefully constructed examples:

Example 3.1: Demonstrates three different functions and their corresponding range representations $(S,C)$

Example 5.1: Shows that the converse of Proposition 5.1 does not hold

• $F(x,y) = x+y$, $f(x) = x$
• $T$ is associative, but $F(\cup_{k\in K}I_k, M) \cap M \neq \emptyset$

Example 5.2: Shows that conditions in Theorem 5.1 cannot be removed

• Two examples violating "cancellativity" and "$F(M,M\backslash C) \subseteq M\backslash C$" respectively
• Yet $T$ remains associative

Example 5.4: Shows that the converse of Proposition 5.4 does not hold

• $F$-condition is satisfied but $T$ is not associative

Example 5.5: Shows necessity of condition $F(C,M) \cup F(M,C) \subseteq M\backslash C$ in Theorem 5.2

• Four examples demonstrating different cases

Experimental Results

Summary of Main Theorems

Theorem 5.1 (Necessary and Sufficient Condition I): Let $F$ be cancellative and $F(M, M\backslash C) \subseteq M\backslash C$, then: $T \text{ is associative} \Leftrightarrow F(\cup_{k\in K}I_k, M) \cap M = \emptyset \text{ and } F(M, \cup_{k\in K}I_k) \cap M = \emptyset$

Theorem 5.2 (Necessary and Sufficient Condition II): Let $F(C,M) \cup F(M,C) \subseteq M\backslash C$, then: $T \text{ is associative} \Leftrightarrow \text{$F$-condition holds}$

Corollary 5.3: When $F$ is cancellative, the $H$ sets in the $F$-condition are non-empty, so the necessary and sufficient condition simplifies to: $T \text{ is associative} \Leftrightarrow J(M) \cap (M\backslash C) = \emptyset$

Key Lemmas

Lemma 5.1: If $[\min\{x,y\}, \max\{x,y\}] \cap M = \emptyset$, then $G_M(x) = G_M(y)$

This lemma plays a crucial role in proving both sufficiency and necessity.

Lemma 5.2: Provides equivalent conditions for determining $F(O(M_1 \cup M_2), c) \cap (M\backslash C) = \emptyset$

Theoretical Verification

The paper validates the theory through constructive proofs and counterexamples:

1. Sufficiency: Propositions 5.1, 5.3 provide sufficient conditions
2. Necessity: Propositions 5.2, 5.4 provide necessary conditions
3. Equivalence: Establishes necessary and sufficient conditions under appropriate assumptions (Theorems 5.1, 5.2)
4. Conditions Cannot be Weakened: Examples 5.2, 5.5 show that assumptions in the theorems are necessary

Related Work

1. Classical Theory

Abel (1826): First studied the problem of strictly monotone functions generating associative functions

Schweizer & Sklar (1961, 1963):

• Established connections between t-norms and additive generators
• Introduced the concept of additive generators

Ling (1965):

• Independently studied representations of associative functions
• Provided characterizations of continuous strictly decreasing generators

2. Generalizations

Klement, Mesiar & Pap (2000):

• Systematically studied triangular norm theory
• Generalized the concept of additive generators, requiring $f(x) + f(y) \in \text{Ran}(f) \cup [f(0),\infty]$

Viceńık (1998, 2005):

• Studied discontinuous generators
• Considered strictly monotone functions
• This paper generalizes his results to more general monotone functions

Yao Ouyang et al. (2007, 2008):

• Introduced binary operations $\star$ on totally ordered Abelian semigroups
• Required $f(x) \star f(y) \in \text{Ran}(f) \cup [f(0^+),\infty]$
• This paper does not require $f$ to be strictly decreasing or satisfy range conditions

Zhang & Wang (2024):

• Studied monotone right-continuous functions
• This paper generalizes to more general monotone function class $\mathcal{F}$

3. Positioning of This Paper

This paper unifies and deepens the above work:

• More General Function Class: $\mathcal{F}$ contains strictly monotone and right-continuous functions as special cases
• No Range Condition Required: Does not require conditions like (1) or (3)
• Range-Driven: Reveals that associativity depends solely on range structure
• Precise Characterization: Provides multiple necessary and sufficient conditions

Conclusions and Discussion

Main Conclusions

1. Essence of Associativity: The associativity of function $T(x,y) = f^{(-1)}(F(f(x), f(y)))$ is completely determined by the structure of the range $M = \text{Ran}(f)$ of $f$, independent of other properties of $f$ (such as continuity or strictness)
2. Range Characterization: For function class $\mathcal{F}$, the range can be uniquely represented by a pair $(S,C)$, where $S$ is a system of closed intervals and $C$ is a countable set of points
3. Necessary and Sufficient Conditions: Under different assumptions ($F$ cancellative, satisfying specific range conditions, etc.), provides multiple equivalent characterizations of when $T$ is associative
4. Generality: Results apply to all strictly monotone functions and monotone right-continuous functions, generalizing multiple existing works

Limitations

1. Function Class Restrictions: Still requires $f \in \mathcal{F}$, i.e., satisfying specific conditions on right limits; has not been generalized to all monotone functions
2. Complexity of Conditions: Necessary and sufficient conditions (particularly the $F$-condition) involve multiple set definitions, making verification in practical applications potentially difficult
3. Theorem Assumptions: Theorems 5.1 and 5.2 both require additional assumptions (cancellativity, range conditions, etc.); Examples 5.2 and 5.5 show these assumptions cannot be simply removed
4. Constructive Nature: The paper primarily provides decision criteria; algorithmic methods for constructing $f$ given $F$ (or vice versa) are discussed less
5. Application-Oriented: As pure theoretical research, lacks discussion of concrete application scenarios

Future Directions

The paper explicitly proposes in its conclusions:

Core Question: Can the results be generalized to all monotone functions?

This is an interesting and challenging open problem. Possible research directions include:

1. Relaxing Conditions on $\mathcal{F}$: Study monotone functions not satisfying right-limit conditions
2. Algorithm Implementation: Develop effective algorithms to determine whether a given $(F, f)$ pair generates an associative function
3. Application Research: Apply the theory to concrete fields such as fuzzy logic, aggregation operators, and copula theory
4. Generalization to Other Operations: Study similar constructions in other algebraic structures (such as t-conorms)
5. Computational Complexity: Study the computational complexity of verifying necessary and sufficient conditions

In-Depth Evaluation

Strengths

1. Theoretical Depth

• Core Insight: The discovery that associativity depends solely on range structure is a profound theoretical contribution, simplifying the essence of the problem
• Completeness: Forms a complete theoretical system from range representation (Section 3) to operation definition (Section 4) to necessary and sufficient conditions (Section 5)
• Precision: Multiple necessary and sufficient conditions characterize the problem from different angles with rigorous theory

2. Generality

• Unifies multiple existing results (Viceńık, Zhang & Wang, Yao Ouyang, etc.)
• Function class $\mathcal{F}$ is broader than those considered in previous work
• Does not require traditional range conditions such as (1) or (3)

3. Technical Innovation

• $(S,C)$ Representation: The closed interval-point set representation of ranges is novel
• $G_M$ Function: Cleverly connects pseudo-inverse with range structure
• Interval Sets $I_k$: Captures the behavior of operations at "gaps"

4. Rigorous Argumentation

• Each theorem has detailed proofs
• Abundant examples and counterexamples verify theoretical boundaries
• Clearly indicates which conditions can be weakened and which cannot

5. Clear Writing

• Logical structure: Preliminaries → Range characterization → Operation properties → Necessary and sufficient conditions
• Clear definitions and complete symbol system
• Rich examples aid understanding of abstract concepts

Weaknesses

1. Insufficient Application Orientation

• As a math.GM classified paper, lacks connections with concrete application fields
• Does not discuss practical significance of theoretical results in fuzzy logic, aggregation theory, etc.
• Lacks consideration of computational or algorithmic aspects

2. Representative Nature of Examples

• Although numerous, examples are mostly artificially constructed piecewise functions
• Lacks natural examples from practical problems
• Insufficient explanation of why these specific examples are important

3. Verifiability of Conditions

• The $F$-condition involves multiple set definitions and operations, potentially difficult to verify in practice
• Does not discuss how to effectively determine whether a given $(F,f)$ pair satisfies the conditions
• Lacks analysis of condition complexity

4. Open Problems

• Although proposing the problem of generalization to all monotone functions, does not discuss the difficulties involved
• Provides no possible solution approaches or partial results
• Remark 4.1 indicates certain conditions cannot be removed, but lacks deep analysis of why

5. Technical Details

• Some proofs (such as Part II of Lemma 3.1) are quite technical with room for improved readability
• The symbol system, while complete, is complex (such as $M_k^y, I_k^y, H_k^y$, etc.), requiring frequent reference to definitions

Impact

1. Contribution to the Field

• Theoretical Value: Establishes a new analytical framework in the theory of generating associative functions
• Unification: Provides a unified perspective for multiple existing results
• Deepened Understanding: Reveals the essential connection between associativity and range structure

2. Practical Value

• Decision Tool: Provides theoretical tools for determining function associativity
• Construction Guidance: Although not constructive, theory can guide selection of generator functions
• Limitation: Direct application may be limited due to condition complexity

3. Reproducibility

• Theoretical Verifiability: All theorems have detailed proofs that can be verified
• Example Reproducibility: Given examples have explicit function definitions, easy to verify
• No Experiments: As pure theoretical research, no experimental reproducibility issues

4. Potential for Follow-up Research

May inspire research directions such as:

• Generalization to all monotone functions
• Generation of multivariate associative functions
• Similar theory for other algebraic structures (such as quasi-groups)
• Algorithmic and computational complexity research
• Applications in aggregation theory, fuzzy systems

Applicable Scenarios

1. Theoretical Research

• Associative Equation Theory: Study structure of solutions to associative equations
• Triangular Norm Theory: Construction and classification of t-norms and t-conorms
• Semigroup Theory: Study representations of ordered semigroups

2. Potential Application Fields

• Fuzzy Logic: T-norms are fundamental operations in fuzzy logic
• Probabilistic Metric Spaces: Generalization of triangle inequality
• Aggregation Operators: Design of aggregation functions for information fusion
• Copula Theory: Construction of multivariate distributions

3. Inapplicable Scenarios

• Applications requiring explicit construction algorithms
• Scenarios with strict computational efficiency requirements
• Cases involving non-monotone generator functions

References

The paper cites 17 references, primarily including:

Classical Literature:

• 1 Abel (1826): Pioneering work on associative functions
• 9 Schweizer & Sklar (1961): Triangular norms and generators
• 5 Ling (1965): Representations of associative functions

Important Monographs:

• 2 Alsina, Frank, Schweizer (2006): Associative Functions: Triangular Norms and Copulas
• 4 Klement, Mesiar, Pap (2000): Triangular Norms

Recent Related Work:

• 13 Viceńık (2005): Additive generators of associative functions
• 8 Yao Ouyang et al. (2008): Generalization of generators
• 16 Zhang & Wang (2024): Characterization of monotone right-continuous functions

Overall Assessment

This is a high-quality pure theoretical mathematics paper that makes substantial contributions to the theory of generating associative functions. The paper's core insight—that associativity depends solely on range structure—is profound, unifying and generalizing multiple existing results. The theoretical system is complete, proofs are rigorous, and examples are abundant.

Main strengths are theoretical depth and generality; main weaknesses are insufficient application orientation and complexity of condition verification. For researchers working in associative function theory, triangular norms, fuzzy logic, and related fields, this is a paper worth careful study.

Recommendation Index: ⭐⭐⭐⭐ (4/5)

• Theoretical Innovation: ⭐⭐⭐⭐⭐
• Practical Value: ⭐⭐⭐
• Readability: ⭐⭐⭐⭐
• Completeness: ⭐⭐⭐⭐⭐