4.6.26Ordinary Differential Equations

Transforms of standard functions — proofs

1,536 words7 min readdifficulty · medium

The one definition everything comes from

HOW to read it: ss is a parameter, tt is the integration variable. After integrating, tt disappears and we are left with a function of ss only.


1. Transform of f(t)=1f(t)=1

Derivation (from scratch). L{1}=0est1dt=[ests]0.\mathcal{L}\{1\} = \int_0^\infty e^{-st}\cdot 1\,dt = \left[\frac{e^{-st}}{-s}\right]_0^\infty.

  • Why this step? Antiderivative of este^{-st} w.r.t. tt is est/(s)e^{-st}/(-s) (chain rule in reverse).

At tt\to\infty: est0e^{-st}\to 0 provided s>0s>0 (else it blows up — this is where the condition s>0s>0 is born). At t=0t=0: e0=1e^{0}=1. So =01s=1s.= 0 - \frac{1}{-s} = \frac{1}{s}.


2. Transform of f(t)=tnf(t)=t^n (and eate^{at} along the way)

Derive L{eat}\mathcal{L}\{e^{at}\} first (it is the easiest and most reusable): L{eat}=0esteatdt=0e(sa)tdt=[e(sa)t(sa)]0=1sa,\mathcal{L}\{e^{at}\} = \int_0^\infty e^{-st}e^{at}\,dt = \int_0^\infty e^{-(s-a)t}\,dt = \left[\frac{e^{-(s-a)t}}{-(s-a)}\right]_0^\infty = \frac{1}{s-a},

  • Why s>as>a? We need sa>0s-a>0 so the exponential decays at infinity. This is just the L{1}\mathcal{L}\{1\} proof with ssas\to s-a.

Derive L{tn}\mathcal{L}\{t^n\} by a recursion (integration by parts). Let In=0esttndtI_n = \int_0^\infty e^{-st}t^n\,dt. Use udv=uvvdu\int u\,dv = uv - \int v\,du with u=tn, dv=estdtu=t^n,\ dv=e^{-st}dt: In=[tnests]0+ns0esttn1dt.I_n = \left[t^n\frac{e^{-st}}{-s}\right]_0^\infty + \frac{n}{s}\int_0^\infty e^{-st}t^{n-1}\,dt.

  • Why this choice? Differentiating tnt^n lowers its power, building a recursion.

The boundary term vanishes: at \infty, este^{-st} beats any polynomial (s>0s>0); at 00, tn=0t^n=0 for n1n\ge1. Hence In=nsIn1.I_n = \frac{n}{s}I_{n-1}. Unrolling from I0=L{1}=1/sI_0=\mathcal{L}\{1\}=1/s: In=nsn1s1s1s=n!sn+1.I_n = \frac{n}{s}\cdot\frac{n-1}{s}\cdots\frac{1}{s}\cdot\frac1s = \frac{n!}{s^{n+1}}.


3. Transforms of sinat\sin at and cosat\cos at

Derivation (the slick complex way). Use eiat=cosat+isinate^{iat} = \cos at + i\sin at. From step 2 with aiaa\to ia: L{eiat}=1sia=s+ia(sia)(s+ia)=s+ias2+a2.\mathcal{L}\{e^{iat}\} = \frac{1}{s-ia} = \frac{s+ia}{(s-ia)(s+ia)} = \frac{s+ia}{s^2+a^2}.

  • Why this step? Multiply top and bottom by the conjugate s+ias+ia to make the denominator real.

Now match real and imaginary parts. Since L\mathcal{L} is linear, L{eiat}=L{cosat}+iL{sinat}\mathcal{L}\{e^{iat}\}=\mathcal{L}\{\cos at\}+i\,\mathcal{L}\{\sin at\}: L{cosat}real=ss2+a2,L{sinat}imag=as2+a2.\underbrace{\mathcal{L}\{\cos at\}}_{\text{real}} = \frac{s}{s^2+a^2}, \qquad \underbrace{\mathcal{L}\{\sin at\}}_{\text{imag}} = \frac{a}{s^2+a^2}.


4. Transforms of sinhat\sinh at and coshat\cosh at

Derivation. Write hyperbolic functions in exponentials and use linearity + step 2: L{coshat}=L ⁣{eat+eat2}=12 ⁣(1sa+1s+a)=ss2a2.\mathcal{L}\{\cosh at\}=\mathcal{L}\!\left\{\tfrac{e^{at}+e^{-at}}{2}\right\}=\tfrac12\!\left(\frac{1}{s-a}+\frac{1}{s+a}\right)=\frac{s}{s^2-a^2}. L{sinhat}=L ⁣{eateat2}=12 ⁣(1sa1s+a)=as2a2.\mathcal{L}\{\sinh at\}=\mathcal{L}\!\left\{\tfrac{e^{at}-e^{-at}}{2}\right\}=\tfrac12\!\left(\frac{1}{s-a}-\frac{1}{s+a}\right)=\frac{a}{s^2-a^2}.

  • Why s>as>|a|? We need both eate^{at} and eate^{-at} pieces to converge, i.e. s>as>a and s>as>-a.
Figure — Transforms of standard functions — proofs

Forecast-then-Verify

Recall Predict before reading!

Q: Without computing, guess L{cosat}\mathcal{L}\{\cos at\} given you already know L{sinat}=as2+a2\mathcal{L}\{\sin at\}=\frac{a}{s^2+a^2}. Predict: numerator should be ss (cos→even, value 11 at t=0t=0), same denominator. Verify: ss2+a2\frac{s}{s^2+a^2} ✓. Note the trick: replacing aiaa\to ia turns sin/cos\sin/\cos into sinh/cosh\sinh/\cosh, i.e. s2+a2s2a2s^2+a^2\to s^2-a^2.


Common mistakes (steel-manned)


Active-recall flashcards

What is the defining integral of the Laplace transform of f(t)f(t)?
0estf(t)dt\int_0^\infty e^{-st}f(t)\,dt, for ss in the region of convergence.
L{1}=?\mathcal{L}\{1\}=? and its condition
1/s1/s, valid for s>0s>0.
L{eat}=?\mathcal{L}\{e^{at}\}=?
1sa\dfrac{1}{s-a} for s>as>a.
Why does the L{1}\mathcal{L}\{1\} integral need s>0s>0?
So est0e^{-st}\to 0 as tt\to\infty; otherwise the area diverges.
L{tn}=?\mathcal{L}\{t^n\}=? and how is it derived?
n!/sn+1n!/s^{n+1}; from the recursion In=nsIn1I_n=\frac ns I_{n-1} via integration by parts, starting at I0=1/sI_0=1/s.
L{sinat}\mathcal{L}\{\sin at\} and L{cosat}\mathcal{L}\{\cos at\}?
as2+a2\dfrac{a}{s^2+a^2} and ss2+a2\dfrac{s}{s^2+a^2}.
How do you get sin/cos transforms from L{eiat}\mathcal{L}\{e^{iat}\}?
Compute 1sia=s+ias2+a2\frac{1}{s-ia}=\frac{s+ia}{s^2+a^2}, then take real part (cos) and imaginary part (sin).
L{coshat}\mathcal{L}\{\cosh at\} and L{sinhat}\mathcal{L}\{\sinh at\}?
ss2a2\dfrac{s}{s^2-a^2} and as2a2\dfrac{a}{s^2-a^2}, for s>as>|a|.
How does the trig denominator become the hyperbolic one?
Replace aiaa\to ia: s2+a2s2a2s^2+a^2\to s^2-a^2.

Recall Feynman: explain to a 12-year-old

Imagine you weigh a function. The weight you put on each moment tt is este^{-st} — heavy near the start, fading later. You add up "function × weight" over all time to get one number per choice of ss. For simple functions like a flat line, growing exponential, or wiggling sine, that adding-up gives clean fractions. We prove each one just by doing the adding-up (an integral) carefully, and noting how big ss must be so the total doesn't run off to infinity.

Connections

Concept Map

contains

needs s large enough

set f=1

shift s to s-a

shift s to s-a gives

integration by parts recursion

base case I0

use Euler with e^iat

complex a gives

d/dt becomes times s

Laplace transform integral

Decaying weight e^-st

Abscissa of convergence

L of 1 = 1/s

L of e^at = 1/ s-a

L of t^n = n!/s^n+1

L of sin at and cos at

Calculus becomes algebra

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Dekho, Laplace transform ek aisi machine hai jo time-domain ki function f(t)f(t) ko leke usko ss-domain mein convert kar deti hai, ek hi formula se: 0estf(t)dt\int_0^\infty e^{-st}f(t)\,dt. Yeh este^{-st} ek "weight" ki tarah kaam karta hai — shuru mein heavy, baad mein fade ho jaata hai. Agar ss kaafi bada ho to yeh weight function ko itna daba deta hai ki total area finite aa jaata hai. Isiliye har transform ke saath ek condition aati hai jaise s>0s>0 ya s>as>a — yeh ignore mat karna, yeh answer ka part hai.

Ab proofs simple hain. L{1}=1/s\mathcal{L}\{1\}=1/s to bas este^{-st} ka direct integral hai. L{eat}\mathcal{L}\{e^{at}\} same cheez hai bas ss ki jagah sas-a — isko "shift" bolte hain. L{tn}=n!/sn+1\mathcal{L}\{t^n\}=n!/s^{n+1} ke liye integration by parts se recursion banate hain In=nsIn1I_n=\frac{n}{s}I_{n-1}, aur unroll karne par factorial nikal aata hai. Yaad rakhna n!n! bhulna sabse common galti hai.

Sin aur cos ke liye sabse smart tarika: eiat=cosat+isinate^{iat}=\cos at+i\sin at. Toh L{eiat}=1sia=s+ias2+a2\mathcal{L}\{e^{iat}\}=\frac{1}{s-ia}=\frac{s+ia}{s^2+a^2} nikalo, real part cos hai aur imaginary part sin. Hyperbolic functions ke liye sinh,cosh\sinh,\cosh ko exponentials mein todo, phir same exponential formula laga do — denominator s2a2s^2-a^2 aata hai (minus sign!). Mnemonic: trig waale mein PLUS, hyperbolic waale mein MINUS. Yeh chhoti table physics, signals, aur ODE solving sab mein baar baar aati hai, isiliye proof samajhna zaroori hai — ratta nahi.

Go deeper — visual, from zero

Test yourself — Ordinary Differential Equations

Connections