Suppose I have a function of the form $$f(x)=\frac{1}{(x-a)(x-b)^2(x-c)^3}$$

Clearly, I have a simple pole at $a$, and poles of order 2,3 at $b,c$, respectively. By definition, the residue at $x=\alpha$ is the coeffecient of the term $(x-\alpha)^{-1}$ in the Laurent expansion of $f(x)$ centered at $x=\alpha$. From my experience, this residue at $x=\alpha$ is equivalent to the coefficient of the term $(x-\alpha)^{-1}$ in the partial fraction decomposition of $f(x)$. However, the partial fraction of $f(x)$ will not be a Laurent series since it has no center (I.e., there are denominators of the form $x-\beta$ with $\beta \not = \alpha$). Is there a rule that says that the coefficient of $(x-\alpha)^{-1}$ in the partial fraction decompositon is the residue at $x=\alpha$, even if the partial fraction decompoition is not a Laurent expansion?


I'm sure you'll find the following theorem very useful.

Theorem: If $f(z)$ has a pole of order $m$ at $z_0$, then $$R [ f,z_0 ] = \lim_{z \to z_0} \frac{1}{( {m - 1} )!} \frac{d^{m - 1}}{dz^{m - 1}} ( ( z - z_0 )^m f( z )).$$


The Laurent series for $f(z)$ around $z_0$ is given by $$f( z ) = \frac{a_{ - m}}{( z - z_0 )^m} + \cdots + \frac{a_{ - 2}}{( z - z_0 )^2} + \frac{a_{ - 1}}{( z - z_0 )^1} + a_0 + a_1 ( z - z_0 ) + \cdots $$

Multiplying the series by $( z - z_0 )^m$, we get $$\begin{array}{c}{\left( {z - {z_0}} \right)^m}f\left( z \right) = {a_{ - m}} + {a_{ - m + 1}}\left( {z - {z_0}} \right) + \ldots + {a_{ - 2}}{\left( {z - {z_0}} \right)^{m - 2}}\\ + {a_{ - 1}}{\left( {z - {z_0}} \right)^{m - 1}} + {a_0}{\left( {z - {z_0}} \right)^m} + a_1 ( z - z_0)^{m + 1} + \ldots \end{array}$$

Differentiating $m-1$ times, we obtain $$\frac{d^{m - 1}}{dz^{m - 1}} ( ( z - z_0 )^m f ( z ) ) = ( m - 1 )! a_{ - 1} + m! a_0 ( z - z_0 ) + \frac{( m + 1 )!}{2}{a_1}{( z - z_0 )^2} + \cdots $$

Hence, it follows that $$\lim_{z \to z_0} \frac{d^{m - 1}}{dz^{m - 1}} ( ( z - z_0 )^m f( z ) ) = ( m - 1 )! a_{ - 1}$$ and the theorem is proved.

From the previous theorem, you can conclude that if $f(z)$ has a simple pole at $z_0$, then $$R\left[ f,z_0 \right] = \lim_{z \to z_0} ( z - z_0 )f( z ).$$

  • 1
    $\begingroup$ Please. Look at my edits to your answer. Your style of writing TeX code is bizarre. It contains zillions of purposeless complications that can only make editing harder. I haven't cleaned up all of it; some of it is still there. \mathop{\lim}\limits_{z \to {z_{0}} could have been just \lim_{z \to z_0}, and lots of it is even much worse than that. $\endgroup$ – Michael Hardy Feb 3 '14 at 22:50
  • $\begingroup$ \frac{{{a_{ - m}}}}{{{{\left( {z - {z_0}} \right)}^m}}} could be coded as \frac{a_{ - m}}{( z - z_0 )^m}. $\endgroup$ – Michael Hardy Feb 3 '14 at 22:53
  • $\begingroup$ I'm really sorry. As I don't master LaTeX yet, sometimes I resort to MathType (dessci.com/EN/products/mathtype). To be honest, perhaps I'm just lazy :/ $\endgroup$ – etothepitimesi Feb 3 '14 at 22:56
  • $\begingroup$ Maybe I'll complain to the authors of that software. It should be possible for them to do far better than that. To write \mathop{\lim}\limits_{} is to neglect really basic stuff. $\endgroup$ – Michael Hardy Feb 3 '14 at 23:06
  • $\begingroup$ It can be my fault... I cut/past/delete/undo/redo lots of stuff, so the code just gets messy and messier. $\endgroup$ – etothepitimesi Feb 3 '14 at 23:10

Yes, there is such a rule. If you have a partial fraction decomposition

$$f(z) = h_\alpha(z) + h_\beta(z) + h_\gamma(z) + g(z),$$

where $h_w$ is the principal part of $f$ in $w$, and $g$ is the remaining holomorphic part of $f$, then you can develop $h_\beta,\, h_\gamma$, and $g$ in Taylor series about $\alpha$ to obtain the Laurent series of $f$ in the annulus $0 < \lvert z-\alpha < d$, where $d$ is the distance to the nearest other singularity ot the boundary of the domain of $f$, whatever is smaller.

The point is that everything but the fractions $\frac{c_k}{(z-\alpha)^k}$ in the partial fraction decomposition of $f$ is holomorphic in a neighbourhood of $\alpha$, so can be expanded into a power series around $\alpha$ and therefore doesn't influence the coefficient of $(z-\alpha)^{-1}$.


Theorem: If $f(x) = g(x)h(x)$ and $g$ is holomorphic near $a$ and $h$ has a simple pole at $a$, then the residue of $f$ at $a$ is $g(a)$ times the residue of $h$ at $a$.

  • $\begingroup$ @DanielFischer : When I wrote "Space-aliens were behind the JFK assassination", it was a typo. I really meant "All recursively enumerable sets are Diophantine." And in this answer, I've added the word "simple". $\endgroup$ – Michael Hardy Feb 3 '14 at 22:27
  • $\begingroup$ That is an impressive typo. Your cat must be proud. $\endgroup$ – Daniel Fischer Feb 3 '14 at 22:30

Your Answer

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

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