20
$\begingroup$

Fixed in 11.3


NSum[Log[k]/k^k, {k,1,Infinity}, WorkingPrecision->50]
(* 0.219947267975228664843531307905860703797097130 *)

But

Sum[Log[k]/k^k, {k,1,Infinity}]
(* -(PolyLog^(1,0))[0,1/k] *)

What does "k" in the result? Bug ?

$\endgroup$
6
  • 1
    $\begingroup$ It's really Times[-1,Derivative[1,0][PolyLog][0,Power[k,-1]]]. I get the same result on 11.1.1.0 on Mac. I'd call it a bug. $\endgroup$
    – John Doty
    Dec 10, 2017 at 13:53
  • 1
    $\begingroup$ Problem continues thru v11.2 on Mac $\endgroup$
    – Bob Hanlon
    Dec 10, 2017 at 14:26
  • 1
    $\begingroup$ This is quite obviously a bug. NSum[Log[k]/k^k, {k, 1, Infinity}] evaluates to 0.219947, and it obviously shouldn't have a $k$ in the output. I'm adding the tag. (V10.0.1 on Mac OSX). $\endgroup$
    – march
    Dec 11, 2017 at 4:24
  • $\begingroup$ BTW, Wolfram Alpha gives the same troublesome result. $\endgroup$ Dec 11, 2017 at 6:53
  • $\begingroup$ Versions 5.2.0 and 8.0.4 give the same wrong result, moreover, the result from 5.2.0 contains the variable "K$18" instead of "k". $\endgroup$
    – innaiz
    Dec 11, 2017 at 7:26

1 Answer 1

2
$\begingroup$

Fyi,

In 11.3 the result is returned as unevaluated. So the issue was bypassed. Much better to return unevaluated, than return something not correct.

Mathematica graphics

Compare to version 11.2

Mathematica graphics

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.