This revision of pointfind should work:
pointfind = @(d) points_id(find(fix(distances*10^(length(num2str(rem(d,1)))-2))/(10^(length(num2str(rem(d,1)))-2)) == d));
(It obviously all goes on one line, although it wraps here.)
You are dealing with differences in precision between the values stored in distances and the precision in the argument ‘d’ you present to pointfind. The revised version of pointfind truncates the elements of distances to the precision of ‘d’ before it does the comparison. It works with decimal fractions now.