Talk:The Schizophreniac From Muhosransk

Computational class

If integers were initialized so that first would be 0 and others would be 1 it would trivially be Turing-complete. Version in which all are 0 at the start is definetly TC too but the proof would be a bit more harder. --ChuckEsoteric08 (talk) 17:16, 28 October 2026 (UTC)

It's easy to see that there's also a construction for when register 1 is set to 1 and all others are set to 0. For example, if you have a program P that uses registers 3, 4, 5, 6, and 7, and assumes register 3 is 0 while 4, 5, 6, and 7 are 1, it can be emulated using ?1 { P } ?2 { -1 +2 +4 +5 +6 +7 }. –PkmnQ (talk) 17:58, 28 October 2026 (UTC)

I honestly can't get how you guys are not bothered by random evaluation order. --Blashyrkh (talk) 18:09, 28 October 2026 (UTC)

When one cell is initialized as 0 and others as 1 (PkmnQ version with different tape layout also initializes to it) we can simulate n-counter machine with m states by using m+n+1 registers, first register is used to initialize counters used by CM as well as cell for the first state to 0 (or set counters used to any other value, so that initialization wouldn't be part of a program) and itself to 1. There is a cell for each state which is 0 if it is a current state and 1 if not current state. --ChuckEsoteric08 (talk) 18:17, 28 October 2026 (UTC)

I see it now, thanks. So, random evaluation order is not a problem as long as all instructions within {} are executed exactly once and change both state registers and memory registers appropriately. But have you any idea how to deal with zero-initialized memory? --Blashyrkh (talk) 08:59, 29 October 2026 (UTC)

Currently I have no idea how would it be done --ChuckEsoteric08 (talk) 09:49, 29 October 2026 (UTC)

I/O?

Say, .x and ,x commands? --Blashyrkh (talk) 10:39, 29 October 2026 (UTC)

sure --dragoneater67 [talk|contribs] 10:40, 29 October 2026 (UTC)