From: Osher Doctorow [EMAIL PROTECTED], Wed. Nov. 28, 2000 3:10PM

First-order LBP analysis of Fermat's Last Theorem (FLT for short in what
follows - see my contribution from the last few days) asserts that
first-order LBP maximum entropy equations/functions are the only admissible
or acceptable kinds, which means that functions must be linear (in either
independent or dependent variables), quadratic, cross-product (like kxy),
simple exponential (like a exp(bx) with b real or complex), or have
vanishing or "neglect-able" 3rd and higher partial derivatives (the latter
mostly applies to mathematical physics as in general relativity).  It turns
out that only x + y = z and x^^2 y^^2 = z^^2 (where ^^ denotes
exponentiation) are admissible under 1st-order LBP and not the higher degree
equations, which "proves" FLT under lst-order conditions.   Notice that in
general elements x, z, y can be arbitrary objects (tensors, vector,
pseudovectors, bivectors, etc.), but in FLT they're usually taken as
(positive) integers, and in LBP they are taken as continuous (e.g., all
nonnegative reals) on the nonnegative real line unless some other connected
domain is specified.

Second-order LBP analysis of FLT is instructive for its applicability to
other Mersenne/prime equations.  Here we use second-order LBP maximum
entropy, which permits multiplication, addition, or subtraction of elements
of 1st-order LBP, but not division, not inversion except subtraction, not
limits unless they enter into the previous types, and not composition of
functions. It turns out that x^^n + y^^n = z^^n needs to have two of the
quantities x, y, or z "independent", and without loss of generality I will
choose x and y to be "independent" and z = f(x, y).   However, if z is a
function of x and y, then z must be a two-variable polynomial (with terms
like kxy, kx^^3y^^5, etc.) since multiplication of x and y times themselves
only yield these (they do not yield fractional or rational powers of x
and/or y without inverse and/or limiting operations).   Therefore, z^^n must
be a two-variable polynomial unless it is a constant.  However, there is no
known equation in mathematics for which x^^n + y^^n = k^^n (constant) for n
> 2 integer (for n = 2 it is a circle, for n = 1 it is a line).   Therefore,
we have an identity x^^n + y^^n = x^^n + y^^n by setting equal powers of x
and y equal in pairs.  An identity (in this sense) is a trivial equation and
is excluded from LBP since trivial equations make no statements about
functions and hence none about their LBP entropy.  This gives a second-order
LBP proof of FLT.  Alternatively, notice that if we took partial derivatives
with respect to x, symbol Dx, of both sides of x^^n + y^^n = z^^n, we would
obtain n! = Dx...x z^^n  where Dx...x is the nth partial, and since an nth
partial is also a function of x and y, we would either end up with n! = n!
(trivial) or n! = two-variable polynomial in x and y, which by equating
terms with equal powers on the left and right sides yields n! = n!.  This
second method does not involve the fact that there is no known function of
form x^^n + y^^n = k^^n for k constant and x, y independent and n > 2
integer.  We can also exclude n! = x^^m + y^^m and similar types from it by
going back to Leibniz' rule and using induction on z = f(x,y).

Osher Doctorow

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

Reply via email to