<Caveat>
This re-post of the thread represents a major update on the technical
reasons why the answer to the questions asked be a "Yes". It's
certainly not meant to be complete and additional updates,
clarification would be needed. But I think we now do have very strong
technical reasons to accept the thesis that in general Mathematical
Reasoning about the naturals is relativistic.
</Caveat>
**********************************************************************
Hi all,
I posted in the past a few threads related to the issue of whether
or not the nature of mathematical reasoning, viz-a-viz FOL, is
genuinely relativistic, in the sense that there would always be
a formula in the language of arithmetic that is truth-undecidable:
it's genuinely impossible to decide its truth value, in the
underlying concept of the natural numbers.
Toward the aim of seeing this issue be in a little more formal
investigation I'd very much appreciate your assistance if you could
kindly forward my questions below to some Institutional Mathematical
Departments for possible (re)solution.
The formula cGC in L(PA) will be defined in details below but briefly
it would stand for "There are infinitely many counter examples of GC
(GoldBach Conjecture)".
Then my 2 questions are:
Q1: Is it reasonable to accept, as a foundational thesis, that
it's impossible to know the truth value of cGC in out *current*
concept of the Natural Numbers?
Q2: If the answer to Q1 is a "Yes", then would it be reasonable to
consider our mathematical reasoning in FOL be relativistic, in
the sense mentioned above?
If the answer to either questions is a "No", please help explaining
the reasons you'd have in supporting your position.
Thank You Kindly and Best Regards,
-Nam Nguyen
namduc...@shaw.ca
-----------------------------------------------------------------------
This post will be divided into 6 sections:
- Section 1: Definitions & Conventions.
- Section 2: Notes. (Some important, relevant notes are mentioned
in the section).
- Section 3: Rules - K & nK. This section concerns how we'd
syntactically "encode" the "Knowing" and "not Knowing"
(or "impossible to know").
- Section 4: "Lemma" Theses. This section contains certain
meta assertions we'd accept on the basis of certain
combination of be self-explanation intuition, explanation
of based from what is known, or meta proofs.
The assumption here is we'd accept these as starting point
theses, unless we could protest with a counter proof or a
clear counter intuition.
- Section 5: Prelude. A brief description of why we should accept
some theses that eventually lead to the suggested that
it's impossible to know the truth value of cGC and ~cGC.
- Section 6: Motivation. We'll explain the motivation for accepting
some theses that would lead to the ultimate proposed
thesis which is:
nK(cGC) and nK(~cGC).
- Section 7: Ramification. We'll explore some preliminary fallout of
nK(cGC) and nK(~cGC).
*****
Section 1 - Definitions & Conventions
=====================================
Def-00: The natural numbers collectively is a language model [of L(PA)]
of which the universe U is non-finite.
Please note the cardinality of a finite set is just the number
of elements in the set, with the empty set is of 0 cardinality.
Def-01: A formula is "positively assertive", or just "positive", iff
the formula contains no negation sign '~', up to logical
equivalence, with the exception where '~' is required for the
expression "P -> Q". For example, the formula prime(x) as defined
below is a positive formula, or just positive., while the
formula ~(x=x) -> A is not positive.
Def-02: prime(x) <-> Ax1x2[(S0<x /\ (x=x1*x2)) -> ((x1=S0 /\ x=x2) \/
(x2=S0 /\ x=x1))]
Def-03a: even1(x) <-> Ey[x=y+y]
Def-03b: even2(x) <-> Ey[x=2*y]
Def-03c: even(x) <-> (even1(x) \/ even2(x))
Def-04a: odd(x) <-> Ey[x=(y+y+S0)]
Def-04b: odd2(x) <-> ~even2(x)
Def-04c: odd(x) <-> (~even1(x) \/ ~even2(x))
Def-05a: GC(x) <-> (even(x) /\ (SS0<x)) -> Ep1p2[prime(p1) /\ prime(p2)
-> (x=p1+p2)]
Def-05b: aGC(x) <-> (even(x) /\ (SS0<x)) -> Ap1p2[prime(p1) /\
prime(p2) -> (p1+p2<x \/ x<p1+p2)]
Def-06a: GC <-> Ax[GC(x)]
Def-06b: aGC <-> Ax[aGC(x)]
Def-07a: Assuming there's a defined P(x), the statement "There are
infinitely many examples of P" would be symbolized as
'(I)P(*)' and is defined as:
(I)P(*) <-> Ex[P(x)] /\ AxEy[P(x) -> (P(y) /\ Ez(y = x + z))]
This is called I-form (Inductive) of infinity expression.
Def-07b: Assuming there's a defined P(x), the statement "There are
infinitely many examples of P" would be symbolized as
'(aI)P(*)' and is defined as:
(aI)P(*) <-> Ex[P(x)] /\ AxEy[P(x) -> (P(y) /\ (x < y))]
This is called aI-form (anti-Inductive) of infinity expression.
Def-07c: P(*) <-> ((I)P(*) \/ (aI)P(*))
This is the general form of infinity.
Def-08: cGC <-> aGC(*)
Def-09: Given a meta statement M, by K(M) and nK(M) we mean,
we can and can not, respectively, assert/verify that
M is true, by consistently and cohesively using
foundational definitions and meta theorems, possibly
coupled with accepted theses and rules regarding to
K(M) and nK(M) mentioned in later sections.
Conv-01a: The symbol '=>' is used for inference in meta level.
Conv-01b: By 'card(U)' we'd mean the cardinality of the set U.
Conv-01c: By 'set(AxP(x))' we'd mean the set of all the naturals
x's each of which P(x), and where P is positively defined.
Conv-02: Given a positive formula A, by 'nK(A)' we mean it's not
impossible in meta to assert the truth of A. Pleas refer
to the Notes section below for more details.
Conv-03a: If A is a formula, then "A" is the meta statement "A is true".
Conv-03a: If A is a formula, the meta statement K("A") can be written
as K(A); and similarly nK("A") as nK(A).
Section 2 - Notes
=================
Note-01: That nK(M) is true doesn't mean M itself is a false statement.
It simply means we can not assert that its truth even if it's
true.
Note-02: For a formula A in L(PA), the ability to assert K(A) or nK(A)
in meta level would follow the guidelines below, the rules
of inference about K(A) and nK(A), the accepted theses, and
existing meta truths or theorems about the naturals, as
detailed in below sections.
The knowing/assertion 'K(A)' shall be obtained only by any
combination of the below methods:
- By finite verification of the truth of A, for Tarski's
model theoretical truth satisfaction.
- By using Induction principle reasoning on the known truths.
- By adhere rules mentioned in Section 3.
Note-03: FToA (Fundamental Theorem of Arithmetic): All numbers greater
than S0 is a product of primes, and is uniquely represented by
factorization of these primes.
Section 3 - Rules - K & nK
==========================
Rule-01: nK(M1) => nK(M1 and M2) [M2 is a meta statement].
Rule-02a: nK(ExP(x)) => nK(AxP(x))
Rule-02b: nK(ExP(x)) => nK(P(*))
Rule-03: ((P -> Q) and nK(P)) => nK(Q)
Section 4 - Prelude
===================
[TBD...]
Section 5 - "Lemma" Theses
==========================
All these theses are named "Anti-Induction" (AI), and are indexed
by Greek alphabets. Also, AI(omega) is actually a meta theorem
(MT) but is listed here instead; it will be proven in Section 5.
AI(alpha): On the basis of (P1 or P2 or ... Pn) alone, we can't
assert none of P1, P2, ... Pn as true. In notation:
(P1 or P2 or ... Pn) => (nK(P1) and nK(P1) and ... nK(Pn))
AI(beta): GC => nK(GC)
AI(gamma): Ex[aGC(x)] => nK(aGC(*))
AI(omega): nK(cGC)
Except for AI(alpha), which is self-explanatory, the other
theses will be explained in some degree in in the below sections.
Section 6 - Motivation
======================
- AI(beta)
=========
This thesis states that if GC is true then it's impossible to know
it so.
Why should we accept this thesis? Below is the explanation.
There's the familiar meta statement we've adopted as "theorem":
(*) If GC is false then it's decidable in PA.
Since that means we'd find in the naturals as a model of L(PA)
a counter example of GC. So then if GC is undecidable in PA it
must be true, which means in so far as PA is consistent it must
not be able to prove GC if GC is true (and there's still a chance
this might be the case.) But syntactically PA contains the very
Induction Principle that we'd use to construct the very language
model we'd refer to as the naturals. So if's it's impossible
to know a proof in PA in this case, then it should equally be
impossible to construct the truth of GC in N. So if we assume
we know GC and there's no counter example then it's impossible
to know GC is true.
- AI(gamma):
==========
This basically states that from there existing one counter
example of GC, it'd be impossible to know there are infinitely
many counter examples.
This is really an application of Rule-02b. But the motivation
of AI(gamma) is the following.
It's not always true in model construction where infinite
relations must be constructed we'd necessarily know all the
existences of the element in the relations. The phrase 'AC'
should remind us that we could assume Ex[P(x)] _without_
showing P(k) for a particular constant k. And in this case,
from knowing Ex[P(x)] it'd be impossible to know if there
are finitely many, or infinitely many counter examples of
GC.
- AI(omega): nK(cGC) and nK(~cGC)
==========
Since GC is either true or false we would consider
2 cases.
Case 1 - GC is true
===================
Now, GC -> ~cGC, but by AI(beta) nK(GC). So it must also be nK(~cGC).
Case 2 - GC is false
====================================
In this case it doesn't matter whether or not there
are finitely or infinitely many counter examples.
By AI(gamma), nK(aGC(*)), but by Def-08, cGC <-> aGC(*)
and so nK(cGC).
QED.
Section 6 - Ramification
========================
[Continued...]