Abstract The integration of Artificial Intelligence (AI) into decision-making is transforming industry norms by enabling smarter, faster, and more reliable decisions. This paper presents a novel approach for supplier selection in manufacturing by combining the Analytic Hierarchy Process (AHP) with the Generative Pre-trained Transformer (GPT). Using GPT as virtual agents to mimic expert evaluations automates the supplier selection process, significantly improving efficiency. The study demonstrates the effectiveness of large language models as virtual experts in complex decision-making, highlighting AI’s crucial role in strategic manufacturing operations. Comparing GPT’s assessments with human expert judgments confirms the model’s reliability and effectiveness in optimizing supplier selection. This approach enhances decision-making processes and emphasizes the broad utility of AI-centric methodologies in various industries.

Abstract Hierarchical Risk Parity methods address some of the limitations of the classical mean-variance approach to portfolio selection by deriving a hierarchical structure. These methods are based on hierarchical clustering techniques and the recursive bisection of an ordered list of assets. When the number of assets is large, computational time becomes a limitation. This paper finds invariants of the allocation produced by simple asset permutations. We also study the size of the decision space to improve the understanding of the allocation algorithm. Building on these results, we propose a fast hierarchical risk parity portfolio selection method that reduces computational time while ensuring a similar performance.

The Population Stability Index (PSI) is a widely used measure in credit risk modeling and monitoring within the banking industry. Its purpose is to monitor for changes in the population underlying a model, such as a scorecard, to ensure that the current population closely resembles the one used during model development. If substantial differences between populations are detected, model reconstruction may be necessary. Despite its widespread use, the origins and properties of the PSI are not well documented. Previous literature has suggested using arbitrary constants as a rule-of-thumb to assess resemblance (or “stability”), regardless of sample size. However, this approach too often calls for model reconstruction in small sample sizes while not detecting the need often enough in large sample sizes. This paper introduces an alternative discrepancy measure, the Population Resemblance statistic (PRS), based on the Pearson chi-square statistic. Properties of the PRS follow from the non-central chi-square distribution. Specifically, the PRS allows for critical values that are configured according to sample size and the number of risk categories. Implementation relies on the specification of a set of parameters, enabling practitioners to calibrate the procedure with their risk tolerance and sensitivity to population shifts. The PRS is demonstrated to be universally competent in a simulation study and with real-world examples.

Abstract Convoy movement problem (CMP) involves the routing and scheduling of military convoys during war time, peacekeeping, or related operations. Although this study focuses on wartime CMP, the underlying methodologies have potential relevance for peacetime logistics and civilian emergency mobility planning. Wartime convoy routing is subject to stringent strategic constraints such as no road crossing, no halts enroute, and maintaining minimum headway distances. Existing literature faces challenges related to scalability, computational complexity, and limited use of search heuristics, highlighting the need for faster and reliable solution approaches. This study proposes 2 methodologies: (1) a heuristic based on the breadth first search [BFS] principal principle and (2) a Simulated Annealing [SA] meta-heuristic algorithm. Two categories of instances [CAT1 and CAT2] are generated using realistic parameters derived from minimum and maximum arc densities. The methodologies are tested on 70 hypothetical cases across both categories, and results are analyzed. Both approaches solve the problem within reasonable computation times. On average, the BFS heuristics yields solutions 6.20% from optimal for CAT1 and 19.43% for CAT2. SA What forms comparatively better, with deviations of 3.86% for CAT1 and 5.78% for CAT2. Overall, the findings indicate that BFS and SA each provide strong performance for specific classes of CMP instances when compared with optimal solutions obtained through conventional global optimization techniques.

In this paper, we apply MaxSAT to a system built around a physical binary sorter that functions as a load balancer between warehouses. We study a sorter operating in real time on 4.5 million elements per year. The optimization task is to minimize the switching needed to route a sequence of elements, under an arbitrary binary function that constrains the achievable load balance. The case study is drawn from a company operating hundreds of such sorters, which are essential to its internal logistics. We examine the problem from both theoretical and practical perspectives: we present an online model, analyze its properties, and evaluate several implementations across different scenarios. Experiments show switching reductions of 63 and 48% for two problem variants in a partial information setting.