How evalf chop works?

[Paul Royik]

expr = 2682469501914725726236807146020965585039500052171867725591732162347*exp(-60)/48648670709168952944186714683693506574127

N(expr) outputs 0.482830727370627

but

N(expr, chop=True) outputs 0

In some other expressions chop doesn't make 0.

So, how chop works?

[Paul Royik]

Nobody could tell how chop works?

[Aaron Meurer]

I'm guessing chop chops some intermediate expression that is smaller
than the cutoff in this case. It probably should be considered a bug.

Aaron Meurer

> --
> You received this message because you are subscribed to the Google Groups
> "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sympy+un...@googlegroups.com.
> To post to this group, send email to sy...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sympy.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sympy/3ef42fe1-b3cb-45d3-80ef-6af88a181bb2%40googlegroups.com.
>
> For more options, visit https://groups.google.com/d/optout.

[Paul Royik]

OK.

Thank you.

Could you move it to issues?

[Chris Smith]

When chop=True it uses the current precision to replace small arguments with an exact zero. If you want to specify the threshold for chopping, pass in a value instead of a boolean. The exp(-60) is about 8x10^-27 so if you specify chop smaller than that you will not get 0 for the result:

>>> expr.n(chop=1e-27)

0.482830727370613

>>> expr.n(chop=1e-26)

0

[Paul Royik]

Thank you.