1

In the course of using the Cauchy Residue Theorem for a particular problem, I have a pole of order $n$ that yields the residue

$$ R(z)= \lim_{s\rightarrow z} \frac{(s-z)^n}{s[A+Bf(s)]^n} $$

where $A$, $B$, and $z$ which is a zero of $[A+Bf(s)]$ are real constants, $s$ is the Laplace variable, and $f(s)$ is a known infinitely differentiable function of $s$. The indeterminate limit has to be resolved with $n$ applications of L'Hôpital's Rule which requires the $n$th derivative of the term in the denominator of the expression above.

In short therefore, I would like a compact (or at least fairly compact) expression for

$$ \frac{d^n}{dx^n}\Bigl[x\Bigl(A+Bf(x)\Bigr)^n\Bigr] = A^n\frac{d^n}{dx^n}\Bigl[x\Bigl(1+rf(x)\Bigr)^n\Bigr] $$

where $r=B/A$. All of my attempts have resulted in one horrendous book-keeping mess after another. Any help will as always be deeply appreciated.

With a binomial expansion, I've managed to get it down to

$$ \frac{d^n}{dx^n}\Bigl[x\Bigl(A+Bf(x)\Bigr)^n\Bigr] = A^n\sum_{k=0}^{n} {n\choose k} r^{n-k} \frac{d^n}{dx^n}\Bigl[xf(x)\Bigr]^{n-k} $$

but then each term by term derivative inside appears to open up another monstrosity!

UPDATE: I was WAY overthinking this and thanks to user Raoul pointing it out, the Limit of a power is the power of the limit rule CONSIDERABLY simplifies my life!

https://www.mathdoubts.com/power-rule-of-limits/

1 Answers1

0

If $z \neq 0$, your limit is $$ \lim_{s \to z} \frac{(s-z)^n}{(A+B f(s))^n} = \frac{1}{z} \left ( \lim_{s \to z} \frac{s-z}{A+B f(s)} \right )^n = \frac{1}{zc^n} $$ where $c$ is the residue of $\frac{1}{A+B f(s)}$ at $z$. If $z = 0$, the limit does not exist.

Raoul
  • 2,360
  • Ah! Does this follow from the fact that the limit of a power is the power of the limit? If, so Yes the expression is only valid for the non-zero poles. Thanks – Sharat V Chandrasekhar Apr 09 '21 at 15:05
  • Yes to both. At $z=0$, the denominator still tends to the residue, but it is also multiplied by $z$, which gives you something indeterminate. – Raoul Apr 09 '21 at 15:09
  • Fortunately, $z=0$ is not a zero of $A+Bf(s)$ and so the residue correspond to $z=0$ is simply $R(0) =1/ [A+Bf(0)]^n$ – Sharat V Chandrasekhar Apr 09 '21 at 15:12