Re: V5 --- Professor Sipser --- Execution trace of simulating termination analyzer HHH on DDD input

Op 24.aug.2024 om 15:21 schreef olcott:

On 8/24/2024 3:47 AM, Fred. Zwarts wrote:

Op 23.aug.2024 om 23:40 schreef olcott:

On 8/23/2024 2:24 AM, joes wrote:

Am Thu, 22 Aug 2024 12:42:59 -0500 schrieb olcott:

>

Only IF it will in fact keep repeating, which is not the case.

>
Only IF it *WOULD* in fact keep repeating, *which is the case*
Only IF it *WOULD* in fact keep repeating, *which is the case*
Only IF it *WOULD* in fact keep repeating, *which is the case*
Only IF it *WOULD* in fact keep repeating, *which is the case*

 

It is the case only if you still cheat with the Root variable, which makes that HHH processes a non-input, when it is requested to predict the behaviour of the input.

 <MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>
     If simulating halt decider H correctly simulates its input D
     until H correctly determines that its simulated D would never
     stop running unless aborted then
      H can abort its simulation of D and correctly report that D
     specifies a non-halting sequence of configurations.
</MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>
 The fact is that it *WOULD* in fact keep repeating,
thus *IT DOES* get the correct answer.

The fact is that it only happens because you make it so with cheating with the Root variable.
In the HHH without this cheat, the simulated and the simulating HHH are coded exactly in the same way to abort after two cycles. Therefore, the fact is that none of them would keep repeating.
It is just a dream that they would keep repeating. Dreams are no substitute for facts.

 

The input given to HHH in fact halts, as is seen in the direct execution and in the correct simulation by HHH1.

 The fact is that all deciders only report on the behavior
specified by their inputs and non-inputs are non-of-their-damn
business.

Exactly. So, they should not predict the behaviour of a hypothetical HHH that does not abort, but the input, which is a HHH that (when the cheat is removed) is coded to abort after two cycles.

 When HHH computes the mapping from its finite string input
of the x86 machine code of DDD to the the behavior that DDD
specifies HHH correctly predicts that DDD cannot possibly
stop running unless aborted.

And since the HHH is coded to abort, DDD *would* stop running. Dreaming that the 'unless' does not happen, i.e. dreaming of an HHH that does not abort, so that DDD does not stop does not help. Dreams are no substitute for facts.

 The reason that this seem so strange is not that I am incorrect.

You are incorrect, but you keep dreaming.

The reason is that everyone rejected simulation as a basis for a
halt decider out-of-hand without review. Because of this they
never saw the details of this behavior when a termination analyzer
correctly emulates an input that calls itself.

The reason that others see that DDD halts, is because they are not dreaming, but they look at facts: it was proven by the direct execution and by the correct simulation by HHH1. And they understand what programming is.

 They never notices that there could possibly be a case where
the behavior of the emulation of the machine specified by its
own Machine description (x86 language) could differ from
the direct execution of this same machine.

And they were right, because the *definition* of a correct simulation is that it agrees with the direct execution.
The only reason why the could differ is an error in the simulator.

 

But HHH cannot possibly simulate itself correctly.

 The ONLY measure of simulated correctly is that each x86
instruction of N instructions of DDD is emulated correctly and
in the correct order.

And without skipping the last (halting) part, and simulating the exact same instructions as in the direct simulation, starting with exactly the same initial values of all variables, so no cheating with the Root variable that differs depending on the level of simulation.

2 + 3 = 5 even if you don't believe in numbers.

And a halting program halts even if you do not believe in halting.