Conditionally Cancellative Triangular Subnorm

• Conditionally cancellative triangular subnorms are binary operations on [0,1]² that relax full cancellation while maintaining weak associativity and monotonicity.
• They are constructed by combining strict t-norms with monotone generator functions through pseudo-inverse techniques to ensure precise algebraic behavior.
• Their definitive characterization resolves a long-standing problem in fuzzy set theory, enhancing tools for aggregation, uncertainty modeling, and data fusion.

A conditionally cancellative triangular subnorm is a binary operation defined on $[0,1]^2$ exhibiting weak forms of associativity, monotonicity, and cancellation, extending the classical theory of triangular norms and subnorms. Such functions arise naturally from combining strict t-norms with monotone generator functions via pseudo-inverse constructions, and they play a pivotal role in the classification of continuous Archimedean proper t-subnorms. The definitive characterization of these structures, as established in (Chen et al., 14 Jan 2026), resolves longstanding open problems in fuzzy set theory.

1. Algebraic Fundamentals

A triangular norm (t-norm) is a mapping $T: [0,1]^2 \rightarrow [0,1]$ fulfilling commutativity $(T1)$, associativity $(T2)$, monotonicity $(T3)$, and the existence of a neutral element $(T4)$: $$\begin{aligned} &(T1)\quad T(x,y)=T(y,x),\ &(T2)\quad T\left(T(x,y),z\right) = T\left(x, T(y,z)\right),\ &(T3)\quad T(x,y) \leq T(x,z)\quad \text{for } y\leq z,\ &(T4)\quad T(x,1) = x. \end{aligned}$$

A triangular subnorm (t-subnorm) relaxes $(T4)$ but enforces

$$(T5)\quad T(x,y)\leq \min\{x,y\}\quad\forall x,y\in [0,1].$$

T-norms are always t-subnorms; those failing $(T4)$ are termed proper.

Associated with cancellation properties are:

• The cancellation law: $x\star y = x\star z \implies x=0$ or $y=z$;
• The conditional cancellation law: $x\star y = x\star z > 0 \implies y=z$.

2. Construction via Monotone Generators and Pseudo-Inverses

Given a monotone $f: [0,1]\rightarrow [0,1]$, one defines its pseudo-inverse as

$f^{(-1)}(u) = \sup\{x\in [0,1]: f(x) < u\}$

(where $f$ is non-decreasing), satisfying

$$f^{(-1)}\circ f \leq \mathrm{id},\qquad f\circ f^{(-1)}(u) = u \text{ for } u \in \mathrm{Ran}(f).$$

With strict t-norm $T$ and monotone $f$, the operation

$F(x, y) = f^{(-1)}(T(f(x), f(y)))$

produces a structure whose algebraic properties depend intricately on the functional form of $f$ and the set-theoretic configuration of its range.

3. Structural Characterization

The decisive result (Theorem 4.4 in (Chen et al., 14 Jan 2026)) stipulates:

Let $T$ be a strict t-norm, and $f$ be monotone with range $M$ and “holes”—

$$M = \left([0,1] \setminus \bigcup_{k\in K} [b_k, d_k]\right)\cup \{c_k: k \in K\}$$

for countable $K$, gap intervals $[b_k, d_k]$, and distinguished points $c_k$.

Define

$$Q = \{\omega \in [0,1] : f^{-1}(\{\omega\})\ \text{has at least two points} \},\qquad f(0^+) = \lim_{x\downarrow 0} f(x),$$

and let $\mathfrak L(M)$ be the set encoding associativity-violating gap points.

Then $F$ is a conditionally cancellative t-subnorm if and only if:

• (i) $f$ is non-increasing, $f(x)=0$ for $x\in(0,1]$; or
• (ii) $f$ is non-decreasing and

$$\begin{aligned} &(\text{ii}_1)\quad\mathfrak L(M)\cap(M\setminus C) = \emptyset,\ &(\text{ii}_2)\quad T(M\setminus C, M) \subseteq M\cup [0,f(0^+)],\ &(\text{ii}_3)\quad T(Q, M) \subseteq [0,f(0^+)]. \end{aligned}$$

Condition (ii) implies $f$ is constant on $[0,\tau]$ for some $\tau$ and strictly increasing on $[\tau,1]$; no further gap points are introduced by $T$.

4. Analytical Properties and Proof Outline

Conditional cancellation is verified by analyzing the effect of $T$ on $Q$ and $M$:

• Failure occurs if $T(Q,M)\not\subset[0,f(0^+)]$ or $T(M\setminus C,M)\not\subset M\cup[0,f(0^+)]$. Associativity is breached if some triple $(x,y,z)$ yields a failure, occurring if any of $(\mathrm{ii}_1)$–$(\mathrm{ii}_3)$ are violated.

Propositions 4.1, 4.2, and 4.3 in (Chen et al., 14 Jan 2026) formalize these equivalences, ensuring that the three algebraic–set-theoretic conditions fully characterize the property.

5. Illustrative Examples

Example	Generator $f$	$T$	Property
(A) Cond. cancellative	$f(x)=\frac{1}{2}$ for $x\leq\frac{1}{2}$;<br>$f(x)=x$ for $x>\frac{1}{2}$	$xy$	$F$ is cond. cancellative t-subnorm;<br>associative, fails cancellation.
(B) Cancellative	$f(x)=x/2$ for $x<1$;<br>$f(1)=1$	$\frac{xy}{2-(x+y-xy)}$	$F$ is cancellative t-subnorm.
(C) Not cond. cancell.	$f(x)=\frac{1}{4}x+\frac{1}{4}$ for $x<1$;<br>$f(1)=1$	$\min\{x,y\}$	Fails conditional cancellation (and so cancellation).

The first two cases are verified by applying Theorem 4.4, demonstrating the effectiveness of the characterization.

6. Resolution of Mesiarová’s Problem and Implications

The work resolves the open problem, posed by Mesiarová in 2004, of characterizing additive generators $f$ for which $T(x,y)=f^{-1}(f(x)+f(y))$ defines continuous Archimedean proper t-subnorms. Standard generator duality reduces this to the general question for monotone $f$ and strict $T$ for which

$F(x,y) = f^{(-1)}(T(f(x),f(y)))$

is a continuous, conditionally cancellative, proper t-subnorm.

Theorem 5.1 in (Chen et al., 14 Jan 2026) confirms that a continuous $F$ emerges precisely when $f$ is continuous, non-decreasing, strictly increasing on $[\tau,1]$, with $\mathrm{Ran}(f)$ gap–structure as above, $f(1)<1$, and the three algebraic conditions plus a boundary-continuity condition: $$\mathrm{Ran}(f)\cap [T(f(x^-),f(y^-)),\, T(f(x^+),f(y^+))] \ \text{has at most one point for each } x,y\in (\tau,1].$$ All conditionally cancellative t-subnorms generated in this manner satisfy these combinatorial and analytic constraints, thereby completing the classification of continuous Archimedean proper t-subnorms.

7. Mathematical and Theoretical Significance

Conditionally cancellative triangular subnorms provide critical algebraic tools in fuzzy logic, aggregation theory, and the study of generalized means. Their definitive characterization by set-theoretic properties of monotone generators and strict t-norms brings closure to a central open problem in the field, enabling rigorous identification and utilization of these structures in both theoretical and applied contexts. The established equivalence between conditional cancellation and Archimedean property in the continuous case further refines the landscape of aggregation operators, supporting advanced applications in uncertainty modeling and data fusion.