In Mathematica 9, 10.3 and 11.0 I can easily expand PolyLog[2, 2, x] around $x=1$ using

Series[PolyLog[2, 2, x], {x, 1, 0}, Assumptions -> x < 1]

to obtain

SeriesData[x, 1, {Pi^4/360}, 0, 1, 1]

In Mathematica 11.3 this works as well. Now if I put a minus sign in front of the PolyLog,

Series[-PolyLog[2, 2, x], {x, 1, 0}, Assumptions -> x < 1]

Mathematica 9, 10.3 and 11.0 of course still can do the expansion, but not Mathematica 11.3:

enter image description here

IMHO this looks like another bug in the expansion of special functions in the version 11.x. Or am I missing something here?

EDIT: In the meantime I discovered another glitch related to PolyLog[2, 2, x]. Consider

exp1 = (x PolyLog[4, x])/(1 - x)
exp2 = (x PolyLog[2, 2, x])/(1 - x)

Now doing

Series[exp1, {x, 1, 0}, Assumptions -> x < 1]
Series[exp2, {x, 1, 0}, Assumptions -> x < 1]

works fine in Mathematica versions 9 till 11.0

SeriesData[x, 1, {-Pi^4/90, -Pi^4/90 - Zeta[3]}, -1, 1, 1]
SeriesData[x, 1, {-Pi^4/360, -Pi^4/360 - Zeta[3]}, -1, 1, 1]

In Mathematica 11.3 the expansion of exp2 doesn't work

enter image description here

Likewise, the expansion of exp1+exp2 will result into errors. And this is where it gets interesting. For some reason even older Mathematica versions seem to have massive problems with the expansion of exp1+exp2, while they are perfectly capable to expand exp1 and exp2 separately. Just try to evaluate

Series[exp1+exp2, {x, 1, 0}, Assumptions -> x < 1]

with any Mathematica version you have in your disposal:

Mathematica 8 Mathematica 9 Mathematica 10.3 Mathematica 11.0

I find this somewhat surprising, since so far every time Mathematica 11.x returned a wrong expansion of such special functions, I could recover the correct result with Mathematica 8 or 9. However, this bug seems to affect all Mathematica versions starting at least with the version 8.

  • 5
    $\begingroup$ I agree it looks like a bug, and seems somehow familiar. I am away from the office this week, will check the status when I get back to see if it has been fixed. $\endgroup$ – Daniel Lichtblau Jun 11 '18 at 13:21
  • $\begingroup$ @DanielLichtblau Thanks for the quick reply. In the meantime I believe that I have discovered another issue with Series and PolyLog[2,2,x] which, however, affects all Mathematica versions, not just 11.3. It is in the edited version of this question. Perhaps you could have a look at this as well. $\endgroup$ – vsht Jun 12 '18 at 2:29
  • 1
    $\begingroup$ @vsht The sum in your edited example works well after applying Simplify or Together in Version 8.0, but not without. Series[exp1 + exp2 // Simplify, {x, 1, 0}, Assumptions -> x < 1] yields SeriesData[x, 1, { Rational[-1, 72] Pi^4, Rational[-1, 72] Pi^4 - 2 Zeta[3]}, -1, 1, 1] . $\endgroup$ – Akku14 Jun 12 '18 at 6:14
  • $\begingroup$ @Akku14 This actually reminds me of mathematica.stackexchange.com/questions/162257/… My reasoning is the same as there: The real expressions are usually too complicated to throw a Simplify (or even Together) at them. $\endgroup$ – vsht Jun 12 '18 at 6:27
  • 2
    $\begingroup$ Still an open problem, hoping to give it some attention when time permits. $\endgroup$ – Daniel Lichtblau Jul 7 '18 at 23:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.