From andrew at anselan.com Sat Sep 7 23:46:02 2024 From: andrew at anselan.com (andrew buchanan) Date: Sun, 8 Sep 2024 03:46:02 +0000 (UTC) Subject: [Retros] Stelvio validates very old PG References: <1052097379.2738456.1725767162248.ref@mail.yahoo.com> Message-ID: <1052097379.2738456.1725767162248@mail.yahoo.com> https://pdb.dieschwalbe.de/P0002278 is now C+. Reto Aschwanden writes in Discord: "We have a new length record for tested SPGs! 52.0 moves, breaking the previous record by a full 4 moves. This is the first of the many-rooks-length-record problems to be fully verified. Needless to say, I am very impressed by the composers that they were able to come up with this "I had to adjust Stelvio to be able to cope with this SPG, as one technical limit was breached due to the length of the problem, which then caused an error. I split up the solving process into several chunks, all in all it took an estimated 10000 hours of CPU time. Since strategy seeking can be well parallelized here (>50fold on my machine), and strategy playing 8fold, solving this in one stretch would take about 3 weeks on my server. "In this family, there are a few other problems in 55.5/56.5/57.5 moves, see. https://pdb.dieschwalbe.de/search.jsp?expression=PROBID%20IN%20'P0000136;%20P0001644;%20P0002279'. There were even more recent attempts, one of them possibly not cooked (yet) in 58.5 moves (this version is not in the PDB). "Solving one of these seems way harder though, but I'll look into it... The ultimate goal would be solving them all, but that seems sooooo faaaar away..." Congratulations to Andrey Frolkin who is still with us, for this amazing achievement. And congratulations to Reto for constructing software that can validate it. Best,Andrew -------------- next part -------------- An HTML attachment was scrubbed... URL: