I attempted a proof this way. Don't know, whether this is sufficient...
Using b=t^(1/t), u=log(t), log(b) = u/t , t being the fixpoint
Assume all the following lines as limit-expressions for @->0 (saving me typo-load)
Is this -at least in principle - sufficient?
Gottfried
Using b=t^(1/t), u=log(t), log(b) = u/t , t being the fixpoint
Assume all the following lines as limit-expressions for @->0 (saving me typo-load)
Code:
ยด
b^^(-2+@) - log_b(@) = b^^inf == fixpoint
b^^(-2+@) - log(@)/log(b) = b^^inf
= t
log(b)* b^^(-2+@) - log(@) = u/t * t
= u
log(b)* b^^(-2+@) = u + log(@)
exponentiate
b^(b^^(-2 + @) = t * @
= b^^(-1 + @)
exponentiate again, using base b
b^(b^^(-1 + @)) = b^(t * @ )
= t^@
= b^^(0 + @) = t^@
then this is
lim @->0 b^^(0 + @) = 1
lim @->0 t^@ = 1
saying
lim { @->0 } b^^(-2 + @ ) - log_b( @ ) = b^^inf
is correct in the limit.Is this -at least in principle - sufficient?
Gottfried
Gottfried Helms, Kassel

