"Niek Otten" wrote...
<maximum change can't be set to machine precision

Of course there is no such thing as "machine precision", but I used
your words because it's a nice way to describe precision of the
algorithms used.

Machine precision means, in the case of IEEE double precision floating
point, numbers that can be represented exactly using 1 sign bit, 11
bits for the base-2 order and 52 bits for the base-2 fractional part.

For something that 'of course', Google shows well over 80,000 hits for
"machine precision". Maybe they're all cruft, but a lot of them seem
to be hosted by universities. What do you know that the professors
involved don't?

If I set Maximum change to 0.000000000000001 I get the same answer
as with your formula, in both cases (using Excel) with same number
of significant digits: 1.42857142857143. . . .
....

This may work in this case, but sometimes the significant digits are
further away from the decimal point. If the highest/lowest order
significant digit were 12 decimal places away from the decimal point,
you're going to have trouble with the maximum change entry.