I have Residue calculation that depends on a parameter r, for which I assume

$Assumptions = {r > 0};

I'm interested in the residue of the function


at $z=-r$.

The calculation is:

Residue[1/((z^(4/3) + Power[-1, (3)^-1] r^(4/3))   (z^2 - 
     r^2)), {z, -r}] // Simplify


However, If I first assign a value to r and then take the Residue

Residue[1/((z^(4/3) + Power[-1, (3)^-1] r^(4/3))   (z^2 - r^2)) /. 
  r -> 1, {z, -1}]

1/4 (-1)^(2/3)

The output is finite.

What am I missing? Thanks.

Edit 5.4.18

From the comment section of a different post of mine I learned that Residue in Mathematica versions prior to 11.1 may be buggy. Switching to 11.2, now I get:

$Assumptions = {r > 0};
Residue[1/((z^(4/3) + Power[-1, (3)^-1] r^(4/3)) (z^2 - r^2)), {z, -r}]
Residue[1/((z^(4/3) + Power[-1, (3)^-1] r^(4/3)) (z^2 - r^2)) /. 
  r -> 1, {z, -1}]

Residue[1/(((-1)^(1/3) r^(4/3) + z^(4/3)) (-r^2 + z^2)), {z, -r}]

Residue[1/(((-1)^(1/3) + z^(4/3)) (-1 + z^2)), {z, -1}]

Meaning, Residue fails and does not change the input in both cases.

Also, I decided to plot the complex function for $r=1$ using the function in celtschk's answer in a diffrent post, and got:

plot of the complex function

We can see that in $z=-r=-1$ there's a discontinuity, which may be consistent with a problematic Laurent series and the inexistence of the residue.

Still, if I remove the assumption that $r>0$, Residue returns the previously problematic output:

$Assumptions = {};
Residue[1/((z^(4/3) + Power[-1, (3)^-1] r^(4/3)) (z^2 - 
       r^2)), {z, -r}] // Simplify

-(1/(-2 (-r)^(7/3) + 2 (-1)^(1/3) r^(7/3)))

  • $\begingroup$ There is no Laurent series in a neighborhood of z=-r so it's hard to say what Residue might come up with in that situation. $\endgroup$ Commented Apr 3, 2018 at 14:28
  • $\begingroup$ the cube root of $-1$ is just $-1$. Take a look at the documentation of CubeRoot of the Wolfram language. It gives to me zero with this code Refine[Residue[1/((z^(4/3) - r^(4/3)) (z^2 - r^2)), {z, -r}], r > 0]. Well, using Power you probably will get the principal value of the cube roots of unity. $\endgroup$
    – Masacroso
    Commented Apr 3, 2018 at 14:36
  • $\begingroup$ with this code Refine[Residue[ 1/((z^(4/3) + (-1)^(1/3) r^(4/3)) (z^2 - r^2)), {z, -r}], r > 0] it gives this result -(1/(2 (-(-r)^(1/3) + (-1)^(1/3) r^(1/3)) r^2)) $\endgroup$
    – Masacroso
    Commented Apr 3, 2018 at 14:41
  • $\begingroup$ @DanielLichtblau, how do you know there is no Laurent series? If there isn't, then Residue should fail no? $\endgroup$ Commented Apr 4, 2018 at 11:01
  • 1
    $\begingroup$ Actually looking again I'm not sure. The fractional power gives rise to a series that, in Mathematica, uses Piecewise to handle Im[z] >= 0 vs. Im[z] < 0. But I am not sure the two are actually different in a way that would preclude the existence of a residue. $\endgroup$ Commented Apr 4, 2018 at 15:47


Your Answer

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

Browse other questions tagged or ask your own question.