[FOM] primer on vagueness
Vladimir Sazonov
V.Sazonov at csc.liv.ac.uk
Sat May 28 13:06:22 EDT 2005
The main point of my previous posting was that although the
existence of 2^1024 = 2^{2^10} is provable by means of
classical logic from some simple assumptions, there is a
mathematically rigorous way to consider this number as not
existing (non-feasible) under a restricted version of
classical logic. So to speak, we need here a microscope
instead of the habitual "classical" telescope.
Anyway, if we are still interested in as short as possible
formal classical existence proof of 2^1024, we can make
the sub-derivation
M(0), M(0) => M(1), M(1),...,M(1023), M(1023) => M(1024), M(1024)
from that posting (and therefore the whole derivation of existence
of 2^1024) to be considerably shorter.
First, infer from the sentence (ii) forall x (M(x) => M(x+1))
derived in the previous posting the following sequence of
corollaries, step-by-step:
(*) forall x (M(x) => M(x + 2^i)), i = 0,1,...,10.
Indeed, the case i=0 is just (ii). Then, assuming that this is
proved for some i, we can infer this for i+1:
M(x) => M(x + 2^i) and M(x + 2^i) => M(x + 2^i + 2^i)
entail M(x) => M(x + 2^{i+1}).
Then apply the last formula in (*) to x = 0 and use M(0)
and modus ponens to get M(1024). This proof contains only
10 of routine steps instead of 1024.
Probably a compression of this kind was mentioned by
Charles Silver and Michael Sheard. But even more can be done:
By using this idea we can feasibly derive existence even of
exponential tower 2_1000 = 2^{2^{...^2...}} of the height 1000
which denotes the number incomparably larger than that denoted
by the short tower 2^1024 = 2^{2^10}. (Note, that some iterated
abbreviations for formulas are used here.) Such a derivation can
be extracted from V. Orevkov's proof of non-elementary complexity
of cut elimination (1979) and is also presented in my paper
"On feasible numbers" to demonstrate that non-restricted classical
logic is not appropriate for formalizing feasibility.
Vladimir Sazonov
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
More information about the FOM
mailing list