Paperlane

no. 16 · Tuesday 17 November 2026

Inverse Gram Matrix Methods for Prioritization in Analytic Hierarchy Process: Explainability of Weighted Least Squares Optimization Method

Kevin Kam Fung Yuen

math.OC · v2 · code · 29 pp

Abstract

This paper proposes Inverse Gram Matrix (IGM) methods to prioritize Pairwise Reciprocal Matrices (PRMs) in the Analytic Hierarchy Process (AHP). The IGM methods include Pseudo-IGM, Normalized-IGM, and Lagrangian-IGM. Interestingly, the proposed IGM methods achieve the minimum Weighted Least Squares (WLS) error. Since clarity, explainability, usability, and verification of the closed-form solutions of WLS appear to be incomplete in the literature, comprehensive mathematical proofs, detailed computational demonstrations, and intensive simulation verification to extend the prior studies are offered in this study. After a simulation of one million random PRM instances was performed to verify the equivalence of the IGM methods, another simulation of ten thousand random PRM instances was performed to verify that any one of the proposed IGM methods is the exact closed-form solution to the WLS optimization problem. The proposed IGM methods under the WLS framework provide the promising alternatives to Saaty's Eigenvector method in AHP. The open-source R package igm4ahp, along with its user manual, tutorials, and simulation datasets, is publicly available on GitHub and permanently archived on Zenodo.

code
github.com/kkfyuen/igm4ahp
comments
29 pages, 2 tables. Package user manual, tutorials, and demonstrations: https://kkfyuen.github.io/igm4ahp/ . Code and simulation datasets on GitHub: https://github.com/kkfyuen/igm4ahp . Permanently archived on Zenodo: https://doi.org/10.5281/zenodo.24223557
fields
Optimization and Control
pdf ↗ page arXiv:2401.01190 ↗