I'm using Mathematica to calculate the inverse Laplace transform (ILT) of various functions--functions that I do not know ahead of time. I was stepping through the process and decided to write my own code to simplify the expressions and calculate the inverse transform, but I stumbled upon the general equation that it appears Mathematica uses to calculate the transforms. If we have a function


where P(s) and Q(s) are both polynomials and P(s) is a lower order than Q(s), I have been able to calculate the ILT by evaluating


where c is a root of Q(s) and n the total number of roots. According to wikipedia, the formula


is an acceptable way to calculate the residue of these irreducible quadratics as long as (according to wikipedia) $P(s)$ and $Q(s)$ are holomorphic and $Q(c)=0$ and $Q^{'}(c)\neq0$. Is there anything else I should be aware of? My irreducible quadratics should always be holomorphic, correct? I have very little knowledge of complex calculus.

Second, I did find a site (Theorem 12.21) that uses this formula residue calculation for irreducible polynomials, however they use go back and plug the result of the residue equation into a formula that is in the s domain still, do partial fraction decomposition, then take the ILT. Why is that necessary if the above equation for $y(t)$ works? If someone could help me understand I would greatly appreciate it. I can provide an example of the types of functions I'm looking at if that would be useful (they are basic).

  • Let $H(s) = \int_0^\infty e^{at}e^{-st}dt = \frac{1}{s-a}$ so that $H^{(k-1)}(s)=\frac{(k-1)! (-1)^{k-1}}{(s-a)^k}=\int_0^\infty e^{at}(-t)^{k-1}e^{-st}dt$ is the Laplace transform of $(-t)^{k-1} e^{at}1_{t > 0}$.

  • If $\deg(P) < \deg(Q)$ then $\frac{P(s)}{Q(s)}=\frac{P(s)}{\prod_{j=1}^l (s-a_j)^{k_j}} = \sum_{j=1}^l\sum_{k=1}^{k_j} \frac{c_{j,k}}{(s-a_j)^k}$ (partial fraction decomposition)

  • Therefore $F(s)=\frac{P(s)}{Q(s)}$ is the Laplace transform of $f(t)=\sum_{j=1}^l\sum_{k=1}^{k_j} \frac{c_{j,k}}{(k-1)!} t^{k-1} e^{a_j t}1_{t > 0}$ (the domain of convergence is $\Re(s) > \sup_j \Re(a_j)$)

  • The residue theorem is really another way to compute the coefficients $c_{j,k}$ ie. to do the same thing, applying the Cauchy integral theorem to the inverse Laplace transform integral


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.