2
$\begingroup$

Suppose $f:\hat{\mathbb{C}}\to\hat{\mathbb{C}}$ were meromorphic, and suppose $f$ has a pole at $\infty$. I'm trying to understand the Laurent series of $f$ about $\infty$. By definition, $f$ has a pole at $\infty$ if $f(1/z)$ has a pole at $0$. If we let $w=1/z$, on an annulus around $0$, we have $$f(w)=\sum_{j=-n}^\infty a_jw^j=\sum_{j=-n}^\infty a_j(1/z)^j.$$ Thus, $$f(z)=\sum_{j=-n}^\infty a_jz^j $$ I'm wondering if this is the correct construction of the Laurent expansion of $f$ about infinity. My suspicion is raised because this looks like the expansion of $f$ about $0$, not $\infty$.

$\endgroup$

1 Answer 1

5
$\begingroup$

$g(w) = f(1/w)$ has a pole at $w=0$, so $$ g(w) = \sum_{j=-n}^\infty a_jw^j $$ for $0 < |w| < R$ with some $R > 0$. Then $$ f(z) = g(1/z) = \sum_{j=-n}^\infty a_j z^{-j} = \sum_{j=-\infty}^n a_{-j} z^j $$ for $1/R < |z| < \infty$, and that is the Laurent expansion of $f$ for the pole at infinity: A power series in $1/z$ plus the “principal part” which is a polynomial without constant term.

The residue at infinity is defined as $$ \operatorname {Res}(f,\infty )=-\operatorname {Res}\left({1 \over z^{2}}f\left({1 \over z}\right),0\right) \, . $$ For $f(z) = \sum_{j=-\infty}^n a_{-j} z^j$ this is $\operatorname {Res}(f,\infty ) = -a_{-1}$.


You got the wrong result because $$ f(w)=\sum_{j=-n}^\infty a_j w^j $$ holds for “small” $w$, so you cannot simply substitute $w=1/z$ by $z$ in that identity.

$\endgroup$
4
  • $\begingroup$ The Laurent expansion of $f$ has an infinite number of negative indexed terms? I thought that a function has a pole at a point if and only if its Laurent expansion had a finite number of negative indexed terms? $\endgroup$
    – Ty Perkins
    Commented Apr 26 at 7:02
  • 1
    $\begingroup$ @TyPerkins: That is only true for poles at some $z \in \Bbb C$, not for poles at $\infty$. $\endgroup$
    – Martin R
    Commented Apr 26 at 7:04
  • $\begingroup$ In the resulting Laurent expansion, do we consider the "principal" part of the expansion to then be the finite terms for which the exponent of $z$ is positive (and so the residue is still $a_{-1}$? Or is the principal part of the expansion still considered the negative indexed terms, and our residue becomes $a_1$? $\endgroup$
    – Ty Perkins
    Commented Apr 26 at 7:15
  • 1
    $\begingroup$ @TyPerkins: I am not sure if the “principal part at infinity” is a standardized term. But the “residue at infinity” is clearly defined, I have added that to my answer. $\endgroup$
    – Martin R
    Commented Apr 26 at 7:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .