Can logic be retroactively filled-in?

In a subjective simulation, the moon does not exist if you look away from it.

We know that in a simulation, the creators of the simulation can just make a moon.

Similarly, they can create something behind a door just as the door is being opened. In the subjective perspective, talking about "what really existed behind the door before it was open and observed" is quite meaningless.

But can creators of the simulation create logical arguments out of thin air?

Suppose humans in a simulation encounter the idea of the right angle triangle for the first time. Is it inevitable that they would come to the conclusion that a^2 + b^2 = c^2 ? Does this fact really exist outside of subjective experience?

We may fail to imagine how it could work at all. But many people fail to imagine how a moon could be garbage collected when not in use.

So are there parallel universes, if only in our minds, where logic works differently?

This is a question of the utmost importance, because it affects how we should look at results of computation. In "this timeline", all results of deterministic computation should converge. This is not in dispute. But if results of computation do not necessarily converge across "parallel universes" so to speak, then results of computation are no less malleable (to creators of simulation) than the thing behind the door.