[Numpy-discussion] atol in allclose
Charles R Harris
charlesr.harris at gmail.com
Tue Nov 28 21:36:06 CST 2006
On 11/28/06, Robert Kern <robert.kern at gmail.com> wrote:
> Matthew Brett wrote:
> > Hi,
> > Sorry to keep cluttering the list, but I was a bit surprised by this
> > behavior of allclose:
> > In :allclose([1.0], [1.0], rtol=0)
> > Out:True
> > In :allclose([1.0], [1.0], rtol=0, atol=0)
> > Out:False
> > The docstring seems to imply that atol will not be used in this
> > comparison - or did I misunderstand?
> The docstring is a somewhat loose interpretation of the actual
> You've exposed a corner case. The long and the short of it is this:
> d = less(absolute(x-y), atol + rtol * absolute(y))
> rtol was never intended to be less than the epsilon of your chosen
> point type much less 0. That's what == is for.
> That said, less() should probably be less_equal(), and the docstring can
> clearly describe what is going on.
I think that's right. Do you want to make the change?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Numpy-discussion