Skip to content

T-norms

A triangular norm (T-norm) is a binary operation \(T:[0,1]^2\rightarrow[0,1]\) used to model fuzzy conjunction. A T-norm is commutative, associative, monotone in both arguments, and satisfies \(T(a,1)=a\).

frsutils uses T-norms both to aggregate feature-level similarities and in fuzzy-rough upper approximations. All implementations accept scalar or array-valued inputs through NumPy-compatible backends.

Implemented T-norms

Name Formula Parameters Registered aliases
Minimum \(\min(a,b)\) None minimum, min, goedel, standardintersection
Product \(ab\) None product, prod, algebraic
Łukasiewicz \(\max(0,a+b-1)\) None lukasiewicz, luk, bounded, boundeddifference
Drastic product \(a\) if \(b=1\); \(b\) if \(a=1\); otherwise \(0\) None drastic, drasticproduct
Einstein product \(ab/(2-a-b+ab)\) None einstein, einsteinproduct
Hamacher product \(0\) when \(a=b=0\); otherwise \(ab/(a+b-ab)\) None hamacher, hamacherproduct
Nilpotent minimum \(\min(a,b)\) if \(a+b>1\); otherwise \(0\) None nilpotent, nilpotentminimum
Yager \(1-\min\{1,[(1-a)^p+(1-b)^p]^{1/p}\}\) \(p>0\) yager, yg

The formulas above are the exact conventions implemented by the library. For array reduction, associative T-norms are applied along the aggregation axis; the Yager implementation uses the equivalent multi-argument expression.

References

  1. Klement, E. P., Mesiar, R., & Pap, E. (2000). Triangular Norms. Springer. https://doi.org/10.1007/978-94-015-9540-7
  2. Radzikowska, A. M., & Kerre, E. E. (2002). A comparative study of fuzzy rough sets. Fuzzy Sets and Systems, 126(2), 137–155. https://doi.org/10.1016/S0165-0114(01)00032-X