Introduction to Laplace Transform
Laplace transform, intuitively, is feeding in a function of $f(t)$ and spits out another function $F(s)$. Though it seems magic, the $s$ in fact comes from the parameter of the transform. Let’s step back a bit and have a sequence $a(n)$, and
This seems similar to with $0<x<1$ a geometric series. If that’s so, $a(n)=1$ gives $A(x)=\displaystyle \frac{1}{1-x}$. $a(n)=\displaystyle \frac{1}{n!}$ leads to $e^x$ (regardless of domain of $x$). Rewrite $x^n=e^{n\ln x}$, name $n$ as $t$, $-s=\ln x$ just to look better ($s>0$), and switch to continuous case:
This is the Laplace transform, taking an input $f(t)$ and output $F(s)$ since we integrate over all $t$. In fact, from a probabilistic perspective, it’s like taking an expectation over one thing and left the parameter. And it should be written as $\mathcal L(·;s)$. Another notation is $f(t)\rightsquigarrow F(s)$.
Linearity of Laplace transform:
- $\mathcal L(f+g)=\mathcal L(f)+\mathcal L(g)$
- $\mathcal L(cf)=c\mathcal L(f)$
These come easily from the linearity of integration.
Let’s do some exercise:
-
what’s $1\rightsquigarrow?$
(It’s meaningless when $s<0$)
-
What’s $e^{at}\rightsquigarrow?$
-
What’s $e^{at}f(t)\rightsquigarrow?$
This is called the Exponential shift formula.
-
What’s $\cos at\rightsquigarrow?$ Note that the exponential shift rule also works for complex number $e^{(a+bi)t}$. And let’s remind ourselves the Euler inverse formula: $\displaystyle \cos at=\frac{e^{iat}+e^{-iat}}{2}$,
Note that as long as if we change the sigh of $i$ in a complex expression, and the expression does not change, then it's a real number.
-
Similarly, we can write $\sin at=\displaystyle \frac{e^{iat}-e^{-iat}}{2i}$, then
-
What’s $t^n\rightsquigarrow?$ (integration by parts and $v’=e^{-st}$)
let’s take a look on the $uv$ part:
Now the the right term is just zero. The left term will go to zero because when we do $n$ times Lopital, the above will be 1 but the bottom remains. Therefore,
We see that $\mathcal L(t^n)=\displaystyle\frac{n}{s}\mathcal L(t^{n-1})$, and therefore we can expect