Re: V5 --- Professor Sipser

On 8/21/2024 1:30 PM, Fred. Zwarts wrote:

Op 21.aug.2024 om 14:30 schreef olcott:

On 8/21/2024 3:01 AM, Mikko wrote:

On 2024-08-21 03:01:38 +0000, olcott said:

>
Professor Sipser must have understood that an HHH(DDD)
that does abort is supposed predict what would happen if it never
aborted.

>
Professor Sipser understood that what is not a part of the text is
not a part of the agreement. What H is required to predict is fully
determined by the words "halt decider H". The previous word
"simulating" refers to an implementation detail and does not affect
the requirements.
>

<MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>
</MIT Professor Sipser agreed to ONLY these verbatim words 10/13/2022>
It is crucial to the requirements in that it specifies that H is
required to predict (a) The behavior specified by the finite string D

Which is only complete if it includes all functions called by D.
Including the H that has the same behaviour as the simulating H.

(b) As measured by the correct partial simulation of D by H

Which does not really give a clue, because either a full simulation is
needed, or an algorithm that detects non-halting.

(c) When H would never abort its simulation of F

No, it must predict the behaviour of the input, including the H that
makes a partial simulation, not the behaviour of a hypothetical non-
input that does not abort. This means to predict the behaviour of the D
with the H that is called by D with the same behaviour as the
simulating H. No cheating with a Root variable to give the simulated H
a behaviour different from the simulating H.

(d) This includes H simulating itself simulating D

Itself, means the H with the same behaviour as the simulating H, i.e.
doing a partial simulation.
Anything else is cheating and making a prediction for a non-input.

You keep missing the idea that HHH does a partial simulation of DDD to
predict what would happen if this HHH never aborted its simulation of
DDD.