14 Feb
2014
14 Feb
'14
7:18 a.m.
On Fri, Feb 14, 2014 at 12:08 AM, Dan Asimov <dasimov@earthlink.net> wrote:
It seems that in some sense the density ought to have a spike at any x = 1/2^n in (-1,1), since there are countably many sign patterns with that sum. For all other x in [-1,1], there is a unique sign pattern {e_j} that gives x = Sum e_j/2^j. Don’t know if there is any rigorous way to make sense of this.
But the height of the spike should be countable/uncountable, or 0. Andy