Re: ChatGPT refutes the key rebuttal of my work

Liste des GroupesRevenir à theory 
Sujet : Re: ChatGPT refutes the key rebuttal of my work
De : richard (at) *nospam* damon-family.org (Richard Damon)
Groupes : comp.theory
Date : 16. Oct 2024, 03:11:57
Autres entêtes
Organisation : i2pn2 (i2pn.org)
Message-ID : <61ffc8131435005aaf8976ddbf109b8f16c77668@i2pn2.org>
References : 1 2 3 4 5 6 7 8 9 10 11 12 13
User-Agent : Mozilla Thunderbird
On 10/15/24 4:01 PM, olcott wrote:
On 10/15/2024 2:33 PM, joes wrote:
Am Tue, 15 Oct 2024 13:25:36 -0500 schrieb olcott:
On 10/15/2024 10:17 AM, joes wrote:
Am Tue, 15 Oct 2024 08:11:30 -0500 schrieb olcott:
On 10/15/2024 6:35 AM, Richard Damon wrote:
On 10/14/24 10:13 PM, olcott wrote:
On 10/14/2024 6:50 PM, Richard Damon wrote:
On 10/14/24 11:18 AM, olcott wrote:
On 10/14/2024 7:06 AM, joes wrote:
Am Mon, 14 Oct 2024 04:49:22 -0500 schrieb olcott:
On 10/14/2024 4:04 AM, Mikko wrote:
On 2024-10-13 12:53:12 +0000, olcott said:
>
https://chatgpt.com/share/6709e046-4794-8011-98b7-27066fb49f3e When
you click on the link and try to explain how HHH must be wrong when
it reports that DDD does not terminate because DDD does terminate it
will explain your mistake to you.
I did that, and it admitted that DDD halts, it just tries to justify
why a wrong answer must be right.
It explains in great detail that another different DDD (same machine
code different process context) seems to terminate only because the
recursive emulation that it specifies has been aborted at its second
recursive call.
Yes! It really has different code, by way of the static Root variable.
No wonder it behaves differently.
There are no static root variables. There never has been any "not a pure
function of its inputs" aspect to emulation.
 
Oh, did you take out the check if HHH is the root simulator?
>
 There is some code that was obsolete several years ago.
No, that code is still active. it is the source of the value for the variable Root that is passed around, and is checked in the code to alter the behavior.

 
Every termination analyzer that emulates itself emulating its input has
always been a pure function of this input up to the point where
emulation stops.
 
That point can never come in the complete simulation of a non-
terminating input, because it is infinite.
>
You and Richard never seemed to understand this previously.
But you don't understand the implications of that. A termination analyser can NOT be based on its own compete emulation of the input, and it can't use the fact that a different HHH looking at a DIFFERENT DDD (since for behavior purposes, DDD include the code of the HHH that it calls) to decide what this DDD does.

 
You err because you fail to understand how the same C/x86 function
invoked in a different process context can have different behavior.
Do explain how a pure function can change.
Non-terminating C functions do not ever return, thus cannot possibly be
pure functions.
 
By "pure" I mean having no side effects. You mean total vs. partial.
>
 You may be half right.
https://en.wikipedia.org/wiki/Pure_function
Only the analyzer must be pure.
The input is free to get stuck in an infinite loop.
Right, but it doesn't.

 
HHH is a pure function of its input the whole time that it is emulating.
DDD has no inputs and is allowed to be any finite string of x86 code.
Inputs to HHH are by no means required to ever return AT ALL.
 
I thought DDD was fixed to only call HHH(DDD)?
>
 Inputs are not required to be pure functions.
 
For Termination Analysis, maybe not, since it need to determine for ALL possible cases, which for non-pure functions include all external state value that might exist for it.
But for Halt Deciding, since an input is specified, it does, but that isn't actually needed here since you are trying to distance yourself form it.
But DDD MUST be a pure function, if it is to be a valid input for a termination analyser, as the Analyzer must be, and the code of DDD has nothing in itself that can break the purity. Now, if you exclude HHH for being part of DDD, then it isn't in the domain of valid inputs, since it isn't a complete program to be analysed.
Non-pure functions still need to be fully code-defined, they can just use data that isn't defined as part of their inputs.

Date Sujet#  Auteur
1 Jul 25 o 

Haut de la page

Les messages affichés proviennent d'usenet.

NewsPortal