Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines. On the 6502 machine, I modified the address field of the instruction to make the instruction access wider range of memory. On the x86 machine, I modified some system codes, so OS/Applications cannot see my code by usual method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible be.
On a von Neumann architecture, you can create self-modifying programs precisely because the program being executed is stored in the same way
as the data which the program manipulates is; therefore, you can
overwrite the program itself with new instructions. That doesn't apply
to TMs where the TMs definition is entirely separate from the symbols on
the tape which the TM can manipulate.
André
On 9/11/2025 8:57 PM, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual
method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible be.
*My 2016 paper* https://www.researchgate.net/publication/307509556_Self_Modifying_Turing_Machine_SMTM_Solution_to_the_Halting_Problem_concrete_example
Basically a UTM that has extra features so that it can
modify an aspect of its own code that it is simulating.
It looks like lots of people have copycatted this since.
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on
x86,6502 machines.
On the 6502 machine, I modified the address field of the
instruction to make
the instruction access wider range of memory. On the x86
machine, I modified
some system codes, so OS/Applications cannot see my code by
usual method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation),
what is going on?
I have no idea what a 'self-modifying Turing Machine' could
possible be. On a von Neumann architecture, you can create self-
modifying programs precisely because the program being executed
is stored in the same way as the data which the program
manipulates is; therefore, you can overwrite the program itself
with new instructions. That doesn't apply to TMs where the TMs
definition is entirely separate from the symbols on the tape
which the TM can manipulate.
On 2025-09-11 21:34, olcott wrote:
On 9/11/2025 8:57 PM, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on
x86,6502 machines.
On the 6502 machine, I modified the address field of the
instruction to make
the instruction access wider range of memory. On the x86
machine, I modified
some system codes, so OS/Applications cannot see my code by
usual method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation),
what is going on?
I have no idea what a 'self-modifying Turing Machine' could
possible be.
*My 2016 paper*
https://www.researchgate.net/
publication/307509556_Self_Modifying_Turing_Machine_SMTM_Solution_to_the_Halting_Problem_concrete_example
Basically a UTM that has extra features so that it can
modify an aspect of its own code that it is simulating.
It looks like lots of people have copycatted this since.
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM.
It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code?
TMs
don't really have "code" per se, and if you want to view the
state transition table of a TM as "code-like", then a UTM (or any
other TM) would have no ability to modify its own "code".
A TM
can read and write from its tape, but its "code" isn't on that tape.
On 12/09/2025 02:57, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual
method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible
be. On a von Neumann architecture, you can create self- modifying
programs precisely because the program being executed is stored in the
same way as the data which the program manipulates is; therefore, you
can overwrite the program itself with new instructions. That doesn't
apply to TMs where the TMs definition is entirely separate from the
symbols on the tape which the TM can manipulate.
Thus quoth Turing in "Enumeration of computable sequences" in section 5
of his 1936 paper:
"Let us write down all expressions so formed from the table for the
machine and separate them by semi-colons. In this way we obtain a
complete description of the machine. In this description we shall
replace q_i by the letter "D" followed by the letter "A" repeated i
times, and S_j by "D" followed by "C" repeated j times. This new
description of the machine may be called the standard description (S.D).
It is made up entirely from the letters "A", " C", "D", "L", "R", "N",
and from ";".
If finally we replace "A" by "1", "C" by "2", "D" by "3", "L" by "4",
"R" by "5", "N" by "6", and ";" by "7" we shall have a description of
the machine in the form of an arabic numeral. The integer represented by this numeral may be called a description number (D.N) of the machine.
The D.N determine the S.D and the structure of the machine uniquely. The machine whose D.N is n may be described as M(n). To each computable
sequence there corresponds at least one description number, while to no description number does there correspond more than one computable
sequence. The computable sequences and numbers are therefore enumerable."
Should we choose to do so, we can therefore represent a TM using nothing
but a number.
We have this TM toolkit available to us:
1) read a number off a tape
2) change a number
3) write a number to a tape
4) load and run a "D.N" as a TM - this is exactly what a UTM /does/.
So it seems to me that we have all the ingredients we need to load a
program from a tape, modify it, save the modification, reload the
program and run it again.
Self-modifying code.
Incidentally, this also answers a point that has been made ad nauseam elsewhere in this group about an input being a finite string. A TM /is/
a finite string - indeed, it's a finite string of simple digits. It is perfectly possible to supply to a decider a *complete* description of
the TM to be analysed, in a very well-defined form.
Therefore, if that finite string is not a wholly accurate and complete description of the program to be analysed, it fails properly to model a
TM as described by Turing, and therefore it cannot reasonably be used to draw inferences about The Halting Problem. A decider has no excuse to
ignore any aspect of its input, and must answer on the basis of the behaviour of the entire input program.
If we want to model this on an x86 emulator, we might reasonably do so
by passing in an executable filename (which is quite a good analogy to
"the tape"), but a pointer to the code stops being a defensible
mechanism the moment it is used as an excuse for the decider not being
able to see the whole input program. The decider gets to see
*everything*, and may omit *nothing*.
On 9/11/2025 8:57 PM, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual method. >>>
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible be.
*My 2016 paper* https://www.researchgate.net/publication/307509556_Self_Modifying_Turing_Machine_SMTM_Solution_to_the_Halting_Problem_concrete_example
Basically a UTM that has extra features so that it can
modify an aspect of its own code that it is simulating.
It looks like lots of people have copycatted this since.
You can represent a TM with a number, but that number isn't
present on the tape of the TM, which is the only thing a TM can
manipulate. You can feed a numeric representation of one TM (call
it I) to another TM (call it M), but neither I nor M is going to
have access to the number which represents *itself* on its tape.
You can create a TM which modifies some *other* TM, but not one
that modifies itself.
On 12/09/2025 05:03, André G. Isaak wrote:
<snip>
You can represent a TM with a number, but that number isn't present on
the tape of the TM, which is the only thing a TM can manipulate. You
can feed a numeric representation of one TM (call it I) to another TM
(call it M), but neither I nor M is going to have access to the number
which represents *itself* on its tape. You can create a TM which
modifies some *other* TM, but not one that modifies itself.
Please do correct me if I'm wrong, André, but doesn't that just imply a tape dance?
On 12/09/2025 04:51, André G. Isaak wrote:
How exactly would a UTM be able to modify "its own" code?
Read what Turing called its D.N, change it, write it, re-load, job done.
On 2025-09-11 22:01, Richard Heathfield wrote:
On 12/09/2025 04:51, André G. Isaak wrote:
How exactly would a UTM be able to modify "its own" code?
Read what Turing called its D.N, change it, write it, re-load,
job done.
Perhaps I'm just dense, but I'm not sure what you mean by "D.N."
On 2025-09-11 22:19, Richard Heathfield wrote:
On 12/09/2025 05:03, André G. Isaak wrote:
<snip>
You can represent a TM with a number, but that number isn't
present on the tape of the TM, which is the only thing a TM
can manipulate. You can feed a numeric representation of one
TM (call it I) to another TM (call it M), but neither I nor M
is going to have access to the number which represents
*itself* on its tape. You can create a TM which modifies some
*other* TM, but not one that modifies itself.
Please do correct me if I'm wrong, André, but doesn't that just
imply a tape dance?
I'm not sure that I understand the question...
On 12/09/2025 05:35, André G. Isaak wrote:
On 2025-09-11 22:01, Richard Heathfield wrote:
On 12/09/2025 04:51, André G. Isaak wrote:
How exactly would a UTM be able to modify "its own" code?
Read what Turing called its D.N, change it, write it, re-load, job done.
Perhaps I'm just dense, but I'm not sure what you mean by "D.N."
I refer you to my earlier reply in this thread:
Thus quoth Turing in "Enumeration of computable sequences" in section 5
of his 1936 paper:
"Let us write down all expressions so formed from the table for the
machine and separate them by semi-colons. In this way we obtain a
complete description of the machine. In this description we shall
replace q_i by the letter "D" followed by the letter "A" repeated i
times, and S_j by "D" followed by "C" repeated j times. This new
description of the machine may be called the standard description (S.D).
It is made up entirely from the letters "A", " C", "D", "L", "R", "N",
and from ";".
If finally we replace "A" by "1", "C" by "2", "D" by "3", "L" by "4",
"R" by "5", "N" by "6", and ";" by "7" we shall have a description of
the machine in the form of an arabic numeral. The integer represented by this numeral may be called a description number (D.N) of the machine.
The D.N determine the S.D and the structure of the machine uniquely. The machine whose D.N is n may be described as M(n). To each computable
sequence there corresponds at least one description number, while to no description number does there correspond more than one computable
sequence. The computable sequences and numbers are therefore enumerable."
Should we choose to do so, we can therefore represent a TM using nothing
but a number.
We have this TM toolkit available to us:
1) read a number off a tape
2) change a number
3) write a number to a tape
4) load and run a "D.N" as a TM - this is exactly what a UTM /does/.
So it seems to me that we have all the ingredients we need to load a
program from a tape, modify it, save the modification, reload the
program and run it again.
Self-modifying code.
On 2025-09-11 23:00, Richard Heathfield wrote:
Should we choose to do so, we can therefore represent a TM
using nothing but a number.
We have this TM toolkit available to us:
1) read a number off a tape
2) change a number
3) write a number to a tape
4) load and run a "D.N" as a TM - this is exactly what a UTM /
does/.
OK. I think I get it now. But Turing is describing a method for
representing a *different* TM to a machine.
You can read a number
off the tape, but that number isn't going to be a number which
represents the TM currently running,
It's a number which
represents some *other* TM, i.e. the input to that TM. For any
given TM, its numerical representation isn't something that can
be guaranteed to be own its own tape (I mean, in principle you
could feed a TM a representation of itself, but their would be no
way to guarantee that the numeric representation fed to it would
be a representation of itself).
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines. On the 6502 machine, I modified the address field of the instruction to make the instruction access wider range of memory. On the x86 machine, I modified some system codes, so OS/Applications cannot see my code by usual method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
On 2025-09-12 01:30:54 +0000, wij said:There may be no consensus what exactly the SMTM is. But, it should be straightforward to come up with a definition of SMTM.
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method.
Self-modification means that the program is included in the configuration.
If the partial function from a configuration to the next configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
On 2025-09-11 21:34, olcott wrote:
On 9/11/2025 8:57 PM, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:*My 2016 paper*
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual
method.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible be. >>
https://www.researchgate.net/
publication/307509556_Self_Modifying_Turing_Machine_SMTM_Solution_to_the_Halting_Problem_concrete_example
Basically a UTM that has extra features so that it can
modify an aspect of its own code that it is simulating.
It looks like lots of people have copycatted this since.
Umm. A TM which takes as its input the description of another TM could
in principle modify the description of that TM, but then it wouldn't be
a UTM. It would be a TM which takes as input the description of another
TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state transition table of a TM as "code-like", then a UTM (or any other TM) would have no ability to modify its own "code". A TM can read and write from its tape,
but its "code" isn't on that tape.
André
every TM has a machine descriptionAnd therefore that description can be given to another TM to determine a property of the described TM, such as does the TM described halt when
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM could
in principle modify the description of that TM, but then it wouldn't be
a UTM. It would be a TM which takes as input the description of another
TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state transition
table of a TM as "code-like", then a UTM (or any other TM) would have
no ability to modify its own "code". A TM can read and write from its
tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
On 2025-09-12, olcott <polcott333@gmail.com> wrote:
On 9/11/2025 8:57 PM, André G. Isaak wrote:
On 2025-09-11 19:30, wij wrote:
Obviously TM can simulate SMTM. I have done such thing on x86,6502
machines.
On the 6502 machine, I modified the address field of the instruction
to make
the instruction access wider range of memory. On the x86 machine, I
modified
some system codes, so OS/Applications cannot see my code by usual method. >>>>
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is
going on?
I have no idea what a 'self-modifying Turing Machine' could possible be.
*My 2016 paper*
https://www.researchgate.net/publication/307509556_Self_Modifying_Turing_Machine_SMTM_Solution_to_the_Halting_Problem_concrete_example
We probably should define what self-modification means: how about: a
state transition occurs in a machine whereby the transition function calculates a new state in which some of the differences from the
previous state occur in information which are identifiable as "code".
It is difficult to give a definition of what is code and what
is data because it is continuum, with ambiguity ("code IS data").
Code-like data indicates operations that are to be performed
and is traversed in order to perform them; but it's not always
so clear-cut.
A Turing Machine is potentially self-modifying. It has a piece of
"hardware", which is the tape head and its processing rules. The tape is "software". It is a matter of interpretation whether the contents of
the tape are "code" or "data", and which symbols are in which category
at what stage of processing, and to what extent the operation of the
machine is manipulating "code" or just "plain data".
A Turing Machine cannot modify its own definition, which specifies
its initial configuration. It being self modifying doesn't mean it
it is capable of becoming a different Turing Machine.
Basically a UTM that has extra features so that it can
modify an aspect of its own code that it is simulating.
It looks like lots of people have copycatted this since.
Name two, and prove they even know about the existence of your paper.
On 9/12/2025 9:30 AM, olcott wrote:
every TM has a machine descriptionAnd therefore that description can be given to another TM to
determine a property of the described TM, such as does the TM
described halt when executed directly.
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM could
in principle modify the description of that TM, but then it wouldn't be
a UTM. It would be a TM which takes as input the description of another
TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state transition
table of a TM as "code-like", then a UTM (or any other TM) would have
no ability to modify its own "code". A TM can read and write from its
tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
TMs can't modify their state transition table.
(UTMs can of course
modify the table of the machine they're simulating.)
Its not about proof. That paper of mine merely
introduced the basic notion. Most everything
about SMTMs came after my paper.
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM could
in principle modify the description of that TM, but then it wouldn't be
a UTM. It would be a TM which takes as input the description of another
TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state transition
table of a TM as "code-like", then a UTM (or any other TM) would have
no ability to modify its own "code". A TM can read and write from its
tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
On 9/12/2025 8:38 AM, joes wrote:No, a machine cannot change itself whether simulated or not.
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:When a machine is being simulated by a UTM then this simulated machine
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM. It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state
transition table of a TM as "code-like", then a UTM (or any other TM)
would have no ability to modify its own "code". A TM can read and
write from its tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
can change its own machine description that is being simulated by this
UTM.
Am Fri, 12 Sep 2025 11:39:23 -0500 schrieb olcott:
On 9/12/2025 8:38 AM, joes wrote:
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:When a machine is being simulated by a UTM then this simulated machine
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM. It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state
transition table of a TM as "code-like", then a UTM (or any other TM) >>>>> would have no ability to modify its own "code". A TM can read and
write from its tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
can change its own machine description that is being simulated by this
UTM.
No, a machine cannot change itself whether simulated or not.
The UTM can of course change what machine it is simulating
On 9/12/2025 12:18 PM, joes wrote:
Am Fri, 12 Sep 2025 11:39:23 -0500 schrieb olcott:
On 9/12/2025 8:38 AM, joes wrote:
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:When a machine is being simulated by a UTM then this simulated machine
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM. It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't >>>>>> really have "code" per se, and if you want to view the state
transition table of a TM as "code-like", then a UTM (or any other
TM)
would have no ability to modify its own "code". A TM can read and
write from its tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM >>>>> that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
can change its own machine description that is being simulated by this
UTM.
No, a machine cannot change itself whether simulated or not.The simulated machine only needs to have access to its own machine description on a tape and then be able to write to this tape.
The UTM can of course change what machine it is simulating
Its UTM can provide this access.
Am Fri, 12 Sep 2025 11:39:23 -0500 schrieb olcott:
On 9/12/2025 8:38 AM, joes wrote:No, a machine cannot change itself whether simulated or not.
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:When a machine is being simulated by a UTM then this simulated machine
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM. It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't
really have "code" per se, and if you want to view the state
transition table of a TM as "code-like", then a UTM (or any other TM) >>>>> would have no ability to modify its own "code". A TM can read and
write from its tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM
that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
can change its own machine description that is being simulated by this
UTM.
The UTM can of course change what machine it is simulating
The simulated machine only needs to have
access to its own machine description on
a tape and then be able to write to this tape.
Its UTM can provide this access.
On 2025-09-12, olcott <polcott333@gmail.com> wrote:
The simulated machine only needs to have
access to its own machine description on
a tape and then be able to write to this tape.
Its UTM can provide this access.
That's unproductive though; you are not creating a model of computation
that exceeds the power of the Turing one.
Every theorem that we know about Turing machines applies.
On 2025-09-12, olcott <polcott333@gmail.com> wrote:
The simulated machine only needs to have
access to its own machine description on
a tape and then be able to write to this tape.
Its UTM can provide this access.
That's unproductive though; you are not creating a model of computation
that exceeds the power of the Turing one.
Every theorem that we know about Turing machines applies.
On 9/12/2025 2:08 PM, Kaz Kylheku wrote:
On 2025-09-12, olcott <polcott333@gmail.com> wrote:As I show in my paper:
The simulated machine only needs to have access to its own machine
description on a tape and then be able to write to this tape.
Its UTM can provide this access.
That's unproductive though; you are not creating a model of computation
that exceeds the power of the Turing one.
Every theorem that we know about Turing machines applies.
This SMTM can simply remove the infinite loop that has been applied to itself.
On 9/12/2025 2:08 PM, Kaz Kylheku wrote:
On 2025-09-12, olcott <polcott333@gmail.com> wrote:
The simulated machine only needs to have
access to its own machine description on
a tape and then be able to write to this tape.
Its UTM can provide this access.
That's unproductive though; you are not creating a model of computation
that exceeds the power of the Turing one.
Every theorem that we know about Turing machines applies.
As I show in my paper:
This SMTM can simply remove the infinite loop that
has been applied to itself.
On 9/12/2025 12:18 PM, joes wrote:A machine has no access to its description. (The simulator does.)
Am Fri, 12 Sep 2025 11:39:23 -0500 schrieb olcott:
On 9/12/2025 8:38 AM, joes wrote:
Am Fri, 12 Sep 2025 08:30:01 -0500 schrieb olcott:When a machine is being simulated by a UTM then this simulated machine
On 9/11/2025 10:51 PM, André G. Isaak wrote:
Umm. A TM which takes as its input the description of another TM
could in principle modify the description of that TM, but then it
wouldn't be a UTM. It would be a TM which takes as input the
description of another TM and then somehow modifies it.
How exactly would a UTM be able to modify "its own" code? TMs don't >>>>>> really have "code" per se, and if you want to view the state
transition table of a TM as "code-like", then a UTM (or any other
TM)
would have no ability to modify its own "code". A TM can read and
write from its tape, but its "code" isn't on that tape.
A standard UTM could simulate the machine description of another UTM >>>>> that can modify its own machine description.
TMs can't modify their state transition table. (UTMs can of course
modify the table of the machine they're simulating.)
can change its own machine description that is being simulated by this
UTM.
No, a machine cannot change itself whether simulated or not.The simulated machine only needs to have access to its own machine description on a tape and then be able to write to this tape.
The UTM can of course change what machine it is simulating
Its UTM can provide this access.
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.Self-modification means that the program is included in the configuration. >> If the partial function from a configuration to the next configuration
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method. >>
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of 'orthodoxical'
reasons to persuade me.
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on
x86,6502 machines.
On the 6502 machine, I modified the address field of the
instruction to make
the instruction access wider range of memory. On the x86
machine, I modified
some system codes, so OS/Applications cannot see my code by
usual method.
Self-modification means that the program is included in the
configuration.
If the partial function from a configuration to the next
configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation),
what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you
define
it to mean (and nothing until you define).
But, it should bestraightforward to come up with a definition
of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following
lots of 'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM
uses
some magic that a Turing machine cannot simulate.
On 2025-09-12 13:18:37 +0000, wij said:As usual, informationless.
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method.
Self-modification means that the program is included in the configuration.
If the partial function from a configuration to the next configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define
it to mean (and nothing until you define).
SMTM uses the same magic as TM.But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of 'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses
some magic that a Turing machine cannot simulate.
On Sat, 2025-09-13 at 12:46 +0300, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.Self-modification means that the program is included in the configuration. >>>> If the partial function from a configuration to the next configuration >>>> is Turing computable the machine can be simulated.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method. >>>>
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define
it to mean (and nothing until you define).
As usual, informationless.
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of
'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses
some magic that a Turing machine cannot simulate.
SMTM uses the same magic as TM.
Just by assuming the TM can modify its transition table is enough the definition
of a SMTM.
Because neither does TM clearly specify where the transition table
exists and how the transition table is interpreted and acts upon...
(e.g. does
TM also uses read/write head to read its own transition table?).
On 13/09/2025 14:52, wij wrote:
On Sat, 2025-09-13 at 12:46 +0300, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method.
Self-modification means that the program is included in the configuration.
If the partial function from a configuration to the next configuration >>>>> is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define
it to mean (and nothing until you define).
As usual, informationless.
But, it should bestraightforward to come up with a definition of SMTM. >>>>
Artificial Idiot says TM cannot simulate SMTM, and following lots of
'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses
some magic that a Turing machine cannot simulate.
SMTM uses the same magic as TM.
Just by assuming the TM can modify its transition table is enough the definition
of a SMTM.
If the SMTM follows clear rules that specify how it transitions between states, including how its
own state transition table is modified, then a TM will be able to simulate that.
Because neither does TM clearly specify where the transition table
exists and how the transition table is interpreted and acts upon...
The TM is just a mathematical structure. It's transition table is a partial function with a
specified range and codomain. It is not part of the TM's tape.
(e.g. does
TM also uses read/write head to read its own transition table?).
No, the transition table is not on the tape, and is not visible to the program the TM implements.
Mike.
On 13/09/2025 17:14, Mike Terry wrote:
On 13/09/2025 14:52, wij wrote:
On Sat, 2025-09-13 at 12:46 +0300, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method.
Self-modification means that the program is included in the configuration.
If the partial function from a configuration to the next configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me? (AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define it to mean (and nothing until you define).
As usual, informationless.
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of 'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses some magic that a Turing machine cannot simulate.
SMTM uses the same magic as TM.
Just by assuming the TM can modify its transition table is enough the definition
of a SMTM.
If the SMTM follows clear rules that specify how it transitions between states, including how its
own state transition table is modified, then a TM will be able to simulate that.
Because neither does TM clearly specify where the transition table
exists and how the transition table is interpreted and acts upon...
The TM is just a mathematical structure. It's transition table is a partial function with a
specified range and codomain. It is not part of the TM's tape.
should have been "..specified domain and codomain"
(e.g. does
TM also uses read/write head to read its own transition table?).
No, the transition table is not on the tape, and is not visible to the program the TM implements.
Mike.
On Sat, 2025-09-13 at 17:29 +0100, Mike Terry wrote:
On 13/09/2025 17:14, Mike Terry wrote:
On 13/09/2025 14:52, wij wrote:
On Sat, 2025-09-13 at 12:46 +0300, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method.
Self-modification means that the program is included in the configuration.
If the partial function from a configuration to the next configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me? (AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define it to mean (and nothing until you define).
As usual, informationless.
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of
'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses some magic that a Turing machine cannot simulate.
SMTM uses the same magic as TM.
Just by assuming the TM can modify its transition table is enough the definition
of a SMTM.
If the SMTM follows clear rules that specify how it transitions between states, including how
its
own state transition table is modified, then a TM will be able to simulate that.
Because neither does TM clearly specify where the transition table exists and how the transition table is interpreted and acts upon...
The TM is just a mathematical structure. It's transition table is a partial function with a
specified range and codomain. It is not part of the TM's tape.
should have been "..specified domain and codomain"
(e.g. does
TM also uses read/write head to read its own transition table?).
No, the transition table is not on the tape, and is not visible to the program the TM
implements.
Mike.
The part you might missed: Theory serves our handling of reality.I just have more cases of undecidability in math.
From x86 assembly, we already roughly know what the theory should be.Â
The definition of SMTM (reading part) could be the same as the definition of TM.
The writting part 'in theory' could almost be free-imagination which need not be
more specific than TM if the abstract definition of TM is accepted (of course,
the rule must be compatible with TM's rule).
No formal expression whose expressive power exceeds that of algorithm or procedural language. That is what I feel so far.
From POOH we can see the clue. Undecidable is not derived from any other formal
model, any other theory cannot defy the 'undecidable' result from HP. 'undecidable'
can explain many logic paradoxes. The reverse is questionable.
Another example from POOH:
DD is simulated by x86 emulator, thus the HHH inside, too. The HHH in main
is run direct by real CPU. What is the consequence I am not sure, but is already
invalid as a proof.
If this above invalid reason is accepted, since there is only one copy of x86Â
emulator, HHH is not reentrant-safe. If you try to make it reentrant-safe, not hardÂ
to see, an infinite loop is formed.
I see those 0/1 decision logic is for convenience of reasoning by other formal model
or for easier explanation, those codes are unreachable codes.
When there is reality, theory (for that reality) is not as much important. Math. concept is pure abstract? I would say no. At least there is foundation in
psychology or physiology.
On 13/09/2025 10:46, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.Self-modification means that the program is included in the configuration. >>>> If the partial function from a configuration to the next configuration >>>> is Turing computable the machine can be simulated.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method. >>>>
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define
it to mean (and nothing until you define).
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of
'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses
some magic that a Turing machine cannot simulate.
It isn't difficult to imagine a pseudoturing machine (PTM) that would
do the job.
On Sat, 2025-09-13 at 12:46 +0300, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on x86,6502 machines.Self-modification means that the program is included in the configuration. >>>> If the partial function from a configuration to the next configuration >>>> is Turing computable the machine can be simulated.
On the 6502 machine, I modified the address field of the instruction to make
the instruction access wider range of memory. On the x86 machine, I modified
some system codes, so OS/Applications cannot see my code by usual method. >>>>
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation), what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you define
it to mean (and nothing until you define).
As usual, informationless.
But, it should bestraightforward to come up with a definition of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following lots of> > >>> 'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid SMTM uses
some magic that a Turing machine cannot simulate.
SMTM uses the same magic as TM.
On 2025-09-13 10:16:47 +0000, Richard Heathfield said:
On 13/09/2025 10:46, Mikko wrote:
On 2025-09-12 13:18:37 +0000, wij said:
On Fri, 2025-09-12 at 09:51 +0300, Mikko wrote:
On 2025-09-12 01:30:54 +0000, wij said:
Obviously TM can simulate SMTM. I have done such thing on
x86,6502 machines.
On the 6502 machine, I modified the address field of the
instruction to make
the instruction access wider range of memory. On the x86
machine, I modified
some system codes, so OS/Applications cannot see my code by
usual method.
Self-modification means that the program is included in the
configuration.
If the partial function from a configuration to the next
configuration
is Turing computable the machine can be simulated.
Is my understanding of TM/SMTM different from what AI told me?
(AI's answer seems full of 'academic' type of explanation),
what is going on?
Can't say wihout knowing what the Artificial Idiot said.
There may be no consensus what exactly the SMTM is.
There needen't be. You asked the question so it means what you
define
it to mean (and nothing until you define).
But, it should bestraightforward to come up with a definition
of SMTM.
Artificial Idiot says TM cannot simulate SMTM, and following
lots of 'orthodoxical'
reasons to persuade me.
Without a definition of SMTM is is possible that some valid
SMTM uses
some magic that a Turing machine cannot simulate.
It isn't difficult to imagine a pseudoturing machine (PTM) that
would do the job.
Maybe one could imagine a PTM that could do something that a Turing
machine cannot do but nobody knows how to implement one.
| Sysop: | DaiTengu |
|---|---|
| Location: | Appleton, WI |
| Users: | 1,069 |
| Nodes: | 10 (0 / 10) |
| Uptime: | 83:11:18 |
| Calls: | 13,728 |
| Calls today: | 3 |
| Files: | 186,961 |
| D/L today: |
5,820 files (1,536M bytes) |
| Messages: | 2,416,634 |