Improved solutions for eight instances of Dataset 2 of the Medical Student Scheduling benchmark

Solutions for eight instances of Dataset 2 of the Medical Student Scheduling (MSS) benchmark, each improving the best solution published for its instance. The benchmark is by Zanazzo, Ceschia, Dovier and Schaerf (Journal of Scheduling 28(2):233-246, 2025, DOI 10.1007/s10951-024-00806-z). The problem is a maximisation. Published values are from Table 8 of the paper and match the solution files in the authors' repository, both consulted on 28 September 2026: I40_24_1: 7799 to 7839 I80_12_1: 8614 to 8615 I80_24_1: 16453 to 16455 I160_12_1: 17748 to 17750 I160_24_1: 31470 to 31510 I240_24_1: 50278 to 50291 I320_12_1: 31897 to 31910 I320_24_1: 65943 to 65951 These are feasible solutions. No optimality and no new upper bound are claimed for any of them. Every solution is in the authors' own format and can be checked with the authors' own MiniZinc model (https://bitbucket.org/satt/mss_data, Software/Minizinc/mss.mzn) and a public solver: the model accepts each schedule and prints the stated objective. The deposit also contains an independent evaluator written from scratch from that model, calibrated against all 29 published Dataset 2 solutions, with its complete output, the validation logs and SHA-256 digests of every file. Instances, published solutions and the model belong to their authors and are not redistributed here. Author affiliation and other benchmark results: Team Banzai S.L.U. (https://team-banzai.com).