Wave equation in an infinite medium

𝜕2 𝑦 2
2𝜕 𝑦
= 𝑐 𝜕𝑥2 for −∞ < 𝑥 < ∞, 𝑡 > 0
𝜕𝑡2

𝑦 𝑥, 0 = 𝑓(𝑥)
for −∞ < 𝑥 < ∞
𝜕𝑦
𝑥, 0 = 𝑔(𝑥)
𝜕𝑡
 𝜕2𝑦 2 𝜕2𝑦
ℱ[ 2 ] 𝜔 =c ℱ [ ] 𝜔
𝜕𝑡 𝜕𝑥2

𝜕2𝑦
ℱ[ 2] 𝜔 = iω 2𝑦ො 𝜔, 𝑡 = −𝜔2 𝑦ො 𝜔, 𝑡
𝜕𝑥

𝜕2𝑦 ∞ 𝜕2𝑦
ℱ[ 2 ] 𝜔 = ‫׬‬−∞ 2 𝑥, 𝑡 𝑒 −𝑖𝜔𝑥 𝑑𝑥
𝜕𝑡 𝜕𝑡
𝜕2 ∞ 𝜕2
= 2 ‫׬‬−∞ 𝑦(𝑥, 𝑡) 𝑒 −𝑖𝜔𝑥 𝑑𝑥 = 𝑦ො 𝜔, 𝑡
𝜕𝑡 𝜕𝑡2

𝜕2
𝑦ො 𝜔, 𝑡 = −c2𝜔2 𝑦ො 𝜔, 𝑡
𝜕𝑡2

𝜕2
𝑦ො 𝜔, 𝑡 + c2𝜔2 𝑦ො 𝜔, 𝑡 = 0
𝜕𝑡2

𝑦ො 𝜔, 𝑡 =𝑎𝜔 cos 𝜔𝑐𝑡 + 𝑏𝜔 sin(𝜔𝑐𝑡)

𝑦ො 𝜔, 0 = 𝑎𝜔 cos 0 + 𝑏𝜔 sin(0) = 𝑎𝜔

𝑦ො 𝜔, 0 = 𝑎𝜔 = ℱ 𝑦 𝑥, 0 መ
𝜔 = ℱ 𝑓(𝑥) 𝜔 = 𝑓(𝜔)

𝜕𝑦ො 𝜕𝑦
𝜔, 0 = 𝑐𝜔𝑏𝜔 = ℱ[ (𝑥, 0)] 𝜔
𝜕𝑡 𝜕𝑡
= ℱ[g(x)] 𝜔 = 𝑔ො 𝜔

1
𝑏𝜔 = 𝑔ො 𝜔
𝜔𝑐

1
𝑦ො 𝜔, 𝑡 = 𝑓መ 𝜔 cos 𝜔𝑐𝑡 + 𝑔ො 𝜔 sin(𝜔𝑐𝑡)
𝜔𝑐

1 ∞ 1
𝑦 𝑥, 𝑡 = መ
න [𝑓 𝜔 cos 𝜔𝑐𝑡 + 𝑔ො 𝜔 sin 𝜔𝑐𝑡 ]𝑒 𝑖𝜔𝑥 𝑑𝜔
2𝜋 −∞ 𝜔𝑐
 1 ∞ 𝑔ො 𝜔
𝑦 𝑥, 𝑡 = መ
න [𝑓 𝜔 cos 𝜔𝑐𝑡 + sin 𝜔𝑐𝑡 ]𝑒 𝑖𝜔𝑥 𝑑𝜔
2𝜋 −∞ 𝜔𝑐

1 ∞ መ 1 1 ∞ sin(𝜔𝑐𝑡) 𝑖𝜔𝑥
= ‫𝑓 ׬‬ 𝜔 cos 𝜔𝑐𝑡 𝑒 𝑖𝜔𝑥 𝑑𝜔 + ∙ ‫׬‬−∞[ 𝑔ො 𝜔 ∙ ]𝑒 𝑑𝜔
2𝜋 −∞ 2𝜋 𝑐 𝜔

1 2
𝑠𝑖𝑛𝑎𝜔
1 2 𝐹 𝑈 𝑡+𝑎 −𝑈 𝑡−𝑎 =2
𝜔
−1
𝑠𝑖𝑛𝜔𝑐𝑡 𝐹 1
1 ∞
𝑒 𝑖𝜔𝑐𝑡
+𝑒 −𝑖𝜔𝑐𝑡 𝑈 𝑥 + 𝑐𝑡 − 𝑈 𝑥 − 𝑐𝑡 (Proof)
𝜔 2
መ
න 𝑓(𝜔) ∙ 𝑒 𝑖𝜔𝑥
𝑑𝜔
2𝜋 −∞ 2 1
2 = 𝑔 𝑥 ∗ 𝑈 𝑥 + 𝑐𝑡 − 𝑈 𝑥 − 𝑐𝑡
2𝑐
1 1 ∞ መ 1
𝑖𝜔(𝑥+𝑐𝑡) = 2𝑐 𝑔 𝑥 ∗ 𝑈 𝑥 + 𝑐𝑡 − 𝑔 𝑥 ∗ 𝑈 𝑥 − 𝑐𝑡
= ∙ ‫׬‬ 𝑓(𝜔) 𝑒 𝑑𝜔+
2 2𝜋 −∞ 1 ∞ ∞
1 1 ∞ መ =2𝑐 ‫׬‬−∞ 𝑔 𝜏 𝑈 𝑥 − 𝜏 + 𝑐𝑡 𝑑𝜏 − ‫׬‬−∞ 𝑔 𝜏 𝑈 𝑥 − 𝜏 − 𝑐𝑡 𝑑𝜏
∙ ‫׬‬−∞ 𝑓(𝜔) 𝑒 𝑖𝜔(𝑥−𝑐𝑡) 𝑑𝜔
2 2𝜋
1 𝑥+𝑐𝑡 𝑥−𝑐𝑡 1 𝑥+𝑐𝑡
1 = ‫׬‬−∞ 𝑔 𝜏 𝑑𝜏 − ‫׬‬−∞ 𝑔 𝜏 𝑑𝜏 =2𝑐 ‫𝑥׬‬−𝑐𝑡 𝑔 𝜏 𝑑𝜏
= ∙ 𝑓 𝑥 + 𝑐𝑡 + 𝑓 𝑥 − 𝑐𝑡 2𝑐
2
x −𝜏 + 𝑐𝑡 ≥ 0 ⇒ 𝜏 ≤ 𝑥 + 𝑐𝑡; x −𝜏 − 𝑐𝑡 ≥ 0 ⇒ 𝜏 ≤ 𝑥 − 𝑐𝑡
Wave equation in a semi-infinite medium
𝜕2𝑦 2 𝜕2𝑦
= 𝑐 − 𝐴 for x>0, t>0
𝜕𝑡2 𝜕𝑥2

𝜕𝑦
y(𝑥, 0) = (𝑥, 0) = 0 and y(0,t)=0
𝜕𝑡

𝜕𝑦
lim 𝑥, 𝑡 = 0
𝑥⟶∞ 𝜕𝑥

𝜕𝑦 𝜕 2𝑌(𝑥, 𝑠) 𝐴
2
𝑠 𝑌(𝑥, 𝑠) − 𝑠𝑦(𝑥, 0) − (𝑥, 0) = 𝑐 2 −
𝜕𝑡 𝜕𝑥2 𝑠

𝑠2 𝐴
𝑌’’ 𝑥, 𝑠 − 2 𝑌 𝑥, 𝑠 = 2
𝑐 𝑐𝑠

𝑠 𝐴
𝜆2 − ( )2 = 0; 𝑌𝑝 𝑥, 𝑠 = −
𝑐 𝑠3
𝑌𝑐 𝑥, 𝑠 =k1𝑒 𝑠𝑥/𝑐 + k2𝑒 −𝑠𝑥/𝑐

𝐴
𝑌 𝑥, 𝑠 =k1𝑒 𝑠𝑥/𝑐 + k2𝑒 −𝑠𝑥/𝑐 −
𝑠3

𝐴
𝑌(0, 𝑠) = 𝑘1 + 𝑘2 − 3 = 0
𝑠

𝐴
So 𝑘2 = − 𝑘1. Thus far,
𝑠3

𝐴 𝐴
𝑌 𝑥, 𝑠 =k1𝑒 𝑠𝑥/𝑐 + ( 3 − 𝑘1) 𝑒 −𝑠𝑥/𝑐 −
𝑠 𝑠3

𝜕
ℒ[ lim 𝑦(𝑥, 𝑡)] = lim 𝑌’(𝑥, 𝑠) = 0
𝑥⟶∞ 𝜕𝑥 𝑥⟶∞
 𝑠 𝑠 𝐴
𝑌’(𝑥, 𝑠) = k1𝑒 𝑠𝑥/𝑐 − ( 3 − k1)𝑒 −𝑠𝑥/𝑐 ; 𝑌 ′ 𝑥, 𝑠 = 0 when x → ∞, therefore 𝑘1 = 0
𝑐 𝑐 𝑠

𝐴 𝐴
𝑌(𝑥, 𝑠) = 𝑒 −𝑠𝑥/𝑐 −
𝑠3 𝑠3

The solution is y(x,t)= ℒ-1 𝑌 𝑥, 𝑠 . 𝑁𝑜𝑤

𝐴 −𝑠𝑥/𝑐 𝐴 𝑥 2 𝑥
ℒ-1 𝑒 = 𝑡− 𝑈 𝑡− ; Second-translation theorem
𝑠3 2 𝑐 𝑐

𝐴 𝑥 𝑥 𝐴 2
𝑦(𝑥, 𝑡) = (𝑡 − )2 𝑈 𝑡 − − 𝑡
2 𝑐 𝑐 2

𝑥 𝑥 0 𝑖𝑓 𝑥>𝑐𝑡
(𝑡 − )2 𝑈 𝑡 − =ቊ
𝑐 𝑐 (𝑡 − 𝑥/𝑐)2 𝑖𝑓𝑥≤𝑐𝑡
A problem on [0,L] Next consider boundary value problem:

𝜕2𝑦 2 𝜕2𝑦
= 𝑐 𝑓𝑜𝑟 0 < 𝑥 < 𝐿, 𝑡 > 0
𝜕𝑡2 𝜕𝑥2

𝜕𝑦 𝜕𝑦
y(𝑥, 0) = (𝑥, 0) = 0 𝑎𝑛𝑑 𝑦(0, 𝑡) = 0, 𝐸 (𝐿, 𝑡) = 𝑓(𝑡)
𝜕𝑡 𝜕𝑥

This models vibrations in an elastic bar of constant density 𝜌, uniform cross section of length L. f(t)
is a force per unit length acting parallel to the bar at the right end x=L. Here c2=E/𝜌, where E is
Young’s modulus for the material of the bar. The bar is initially at rest and lying flat along the
interval [0,L].
Apply the Laplace transform ℒ, with respect 𝑡, using the boundary conditions and the operational
rule for taking the transform of derivatives. This is similar to the problem just solved and we obtain.
𝑠2
𝑌 ′′ 𝑥, 𝑠 − 𝑌 𝑥, 𝑠 = 0
𝑐2
in which Y’(x,s)=𝜕𝑌/𝜕𝑥 and s is carried along as a parameter. This is a second order linear
homogeneous constant coefficient differential equation. The characteristic equation is

𝑠 2
𝜆2 −( ) =0
𝑐

with roots ±s/c. The general solution for Y(x,s) is

𝑌 𝑥, 𝑠 =k1𝑒 𝑠𝑥/𝑐 + k2𝑒 −𝑠𝑥/𝑐

We need two initial conditions to solve for k1 and k2. First, transform 𝑦(0, 𝑡) = 0 to obtain
𝑌(0, 𝑠) = 0. Then

𝑌(0, 𝑠) = 𝑘1 + 𝑘2 = 0
so k1=-k2. This means that Y(x,s) has the form

𝑌(𝑥, 𝑠) = 𝑘 sinh(𝑠𝑥/𝑐)
in which k is an arbitrary constant. To determine k, transform the other initial condition
𝜕𝑦
𝐸 𝐿, 𝑡 = 𝑓(𝑡) to obtain 𝐸𝑌 ′ 𝐿, 𝑠 = 𝐹(𝑠).
𝜕𝑥
Applying this to 𝑌(𝑥, 𝑠) = 𝑘 sinh(𝑠𝑥/𝑐) to obtain

𝑠 𝑠𝐿
𝐸𝑘 cosh =𝐹 𝑠 ,
𝑐 𝑐
or
𝑐 1
𝑘 = F(s)
𝐸 𝑠𝐿
𝑠 cosh( )
𝑐
We now have
𝑐 sinh(𝑠𝑥/𝑐)
𝑌(𝑥, 𝑠) = F(s) 𝑠𝐿
𝐸 𝑠 cosh( 𝑐 )
The solution to the original problem is
𝑦(𝑥, 𝑡) = ℒ−1[𝑌(𝑥, 𝑠)](𝑡)
Because of the generality of the problem, f(t) is unspecified and we cannot proceed beyond this
point. However, there are special cases of interest in which we can complete the solution. We will
consider two such cases.
Suppose 𝑓(𝑡) = 𝐾, constant.
Now
𝐾
𝐹(𝑠) = ℒ[𝐾] =
𝑠
so
𝑐𝐾 sinh(𝑠𝑥/𝑐)
𝑌(𝑥, 𝑠) =
𝐸 𝑠2cosh(𝑠𝐿)
𝑐

We can take the inverse transform of this expression by making use of the geometric series:
1
= σ∞ 𝑛 𝑛
𝑛=0 −1 𝜉 , when 𝜉 < 1
1+𝜉
Now write
sinh(𝑠𝑥/𝑐) 𝑒 𝑠𝑥/𝑐 − 𝑒 −𝑠𝑥/𝑐
= 𝑠𝐿/𝑐
𝑠𝐿 𝑒 + 𝑒 −𝑠𝐿/𝑐
cosh( )
𝑐

𝑒 𝑠𝑥/𝑐 𝑒 −𝑠𝐿/𝑐 −𝑒 −𝑠𝑥/𝑐 𝑒 −𝑠𝐿/𝑐

=
1+𝑒 −𝑠𝐿/𝑐 𝑒 −𝑠𝐿/𝑐

𝑒 − 𝐿−𝑥 𝑠/𝑐 −𝑒 − 𝐿+𝑥 𝑠/𝑐

=
1+𝑒 −2𝑠𝐿/𝑐

𝐿−𝑥 𝑠 𝐿+𝑥 𝑠
− − 1
= (𝑒 𝑐 −𝑒 𝑐 )
1+𝑒 −2𝑠𝐿/𝑐

𝐿−𝑥 𝑠 𝐿+𝑥 𝑠
− −
= (𝑒 𝑐 −𝑒 𝑐 ) σ∞ 𝑛
𝑛=0 −1 (𝑒
−2𝑠𝐿/𝑐 n
)

𝐿−𝑥 𝑠 𝐿+𝑥 𝑠
− −
= (𝑒 𝑐 −𝑒 𝑐 ) σ∞ 𝑛 −2𝑛𝑠𝐿/𝑐
𝑛=0 −1 𝑒
Here we used 𝜉 = 𝑒 −2𝑠𝐿/𝑐 in the geometric series, using the fact that this exponential is always less
than 1 for x>0. Finally, we can write

𝑐𝐾 𝑛 1 𝑐𝐾 𝑛 1
𝑌(𝑥, 𝑠) = 𝑒− 𝐿−𝑥 𝑠/𝑐 σ∞
𝑛=0 −1 2𝑒
−2𝑠𝑛𝐿/𝑐 − 𝑒− 𝐿+𝑥 𝑠/𝑐 σ∞
𝑛=0 −1 2 𝑒 −2𝑛𝑠𝐿/𝑐
𝐸 𝑠 𝐸 𝑠
2𝑛+1 𝐿−𝑥 2𝑛+1 𝐿+𝑥
𝑐𝐾 𝑛 1 − 𝑠 𝑐𝐾 𝑛 1 − 𝑠
= σ∞
𝑛=0 −1 2𝑒
𝑐 − σ∞
𝑛=0 −1 2𝑒
𝑐
𝐸 𝑠 𝐸 𝑠

Now recall that, for any nonzero number 𝛼

1 −𝑎𝑠
ℒ-1 𝑒 = (t − 𝛼)𝑈(𝑡 − 𝛼)
𝑠2
in which 𝑈 is the unit step function. Assuming that we can take the transform of the geometric series
term by term, the solution is

𝑐𝐾 ∞ 2𝑛+1 𝐿−𝑥 2𝑛+1 𝐿−𝑥 𝑐𝐾 ∞ 2𝑛+1 𝐿+𝑥

𝑦 𝑥, 𝑡 = σ𝑛=0 −1 𝑛(𝑡 − )𝑈 𝑡− − σ𝑛=0 −1 𝑛(𝑡 − )𝑈 ቀ𝑡 −
𝐸 𝑐 𝑐 𝐸 𝑐
2𝑛+1 𝐿+𝑥
ቁ
𝑐

Look at this solution for a particular choice of the constants. Suppose c=2 and K=E=L=1. Now the solution
is

2𝑛+1−𝑥 2𝑛+1−𝑥 2𝑛+1+𝑥 2𝑛+1+𝑥

𝑦 𝑥, 𝑡 = 2 σ∞
𝑛=0 −1
𝑛 𝑡− 𝑈 𝑡− − 2 σ∞
𝑛=0 −1
𝑛 𝑡− 𝑈 𝑡−
2 2 2 2
Diffusion equation:
Fick’s first law

𝜕𝑢
𝐽 = −𝐷
𝜕𝑥

𝐽: amount of substance flowing through (a small) unit area during (a small) unit time
𝐷: diffusion coefficient
𝑢(𝑥, 𝑡): concentration of the substance
diffusion flow

∆𝑥
u(x,t) u(x+∆𝑥,t)
Amount of substance flowing through the unit time (volume)

𝜕𝑢 𝑥, 𝑡 𝜕𝑢 𝑥 + ∆𝑥, 𝑡
𝐴 ∙ [(−𝐷 ∙ ) − (−𝐷 ∙ )]
𝜕𝑥 𝜕𝑥

flow in flow out

𝜕𝑢
= ∙ 𝐴 ∙ ∆𝑥
𝜕𝑡

𝜕𝑢 𝑥+∆𝑥,𝑡 𝜕𝑢 𝑥,𝑡
𝜕𝑢 𝜕𝑥
− 𝜕𝑥
⟹ =𝐷∙
𝜕𝑡 ∆𝑥
∆𝑥 ⟶ 0

𝜕𝑢 𝜕2𝑢
=𝐷∙ 2
𝜕𝑡 𝜕𝑥
This is the Fick’s second law
 Heat equation, or diffusion equation

𝜕𝑢 𝜕2𝑢
= 𝑘 2 for -∞ < 𝑥 < ∞, 𝑡 > 0
𝜕𝑡 𝜕𝑥

𝑢(𝑥, 0) = 𝑓(𝑥) 𝑓𝑜𝑟 -∞ < 𝑥 < ∞

Take the Fourier transform of the heat equation with respect to x to get

𝜕𝑢 𝜕2𝑢
ℱ = 𝑘ℱ[ 2 ]
𝜕𝑡 𝜕𝑥

Because x and t are independent, ℱ passes through 𝜕/𝜕𝑡, and

𝜕𝑢 𝜕
ℱ = 𝑢ො 𝜔, 𝑡
𝜕𝑡 𝜕𝑡

in which 𝜔 appears as a parameter. Next use the operational rule for the Fourier transform to write

𝜕2𝑢
ℱ[ 2 ]= −𝜔2 𝑢ො 𝜔, 𝑡
𝜕𝑥

The transformed heat equation is

𝜕
𝑢ො 𝜔, 𝑡 +𝑘𝜔2 𝑢ො 𝜔, 𝑡 = 0
𝜕𝑡
with solutions

2
𝑢ො 𝜔, 𝑡 = 𝑎𝜔 𝑒 𝑘𝑡
−𝜔

Since 𝑢(𝑥, 0) = 𝑓(𝑥)

𝑢ො 𝜔, 0 = 𝑓መ 𝜔 = 𝑎𝜔

We now have the Fourier transform of the solution:

2
𝑢ො 𝜔, 𝑡 = 𝑓መ 𝜔 𝑒 −𝜔 𝑘𝑡
Applying the inverse Fourier transform. Since this transform is complex valued and the solution for a
temperature distribution is real, the solution is the teal part of this inverse:

2
𝑢 𝑥, 𝑡 = ℱ −1 [𝑓መ 𝜔 𝑒 −𝜔 𝑘𝑡 ](𝑡)
2
Let 𝑔ො 𝜔, 𝑡 = 𝑒 −𝜔 𝑘𝑡

Then according to convolution theorem

𝑢 𝑥, 𝑡 = 𝑓 𝑥 ∗ 𝑔(𝑥, 𝑡)
 1 ∞ −𝑘𝑡𝜔2 𝑖𝜔𝑥 𝑔′ 𝑥 1
𝑔(𝑥) = ‫𝑒 ׬‬ 𝑒 d𝜔 ⟹ = − 2𝑘𝑡 𝑥
2𝜋 −∞ 𝑔 𝑥

1 ∞ −𝑘𝑡𝜔2 1 2
𝑔′(𝑥) = න 𝑒 𝑖𝜔 𝑒 𝑖𝜔𝑥 d𝜔 ⟹ 𝑔(𝑥) = 𝑐𝑒 4𝑘𝑡𝑥
−
2𝜋 −∞
−𝑖 ∞ 2
= න −𝜔𝑒 −𝑘𝑡𝜔 𝑒 𝑖𝜔𝑥 d𝜔
2𝜋 −∞
∞
−𝑖 𝑑 2
= න 𝑒 −𝑘𝑡𝜔 𝑒 𝑖𝜔𝑥 d𝜔
4𝜋𝑘𝑡 −∞ 𝑑𝜔

∞
−𝑖 2 ∞ 2
= 𝑒 −𝑘𝑡𝜔 𝑒 𝑖𝜔𝑥 ฬ − 𝑖𝑥 න 𝑒 −𝑘𝑡𝜔 𝑒 𝑖𝜔𝑥 d𝜔
4𝜋𝑘𝑡 −∞ −∞

=0
𝑥 ∞
2
𝑥
=− න 𝑒 −𝑘𝑡𝜔 𝑖𝜔𝑥
𝑒 d𝜔 = − 𝑔(𝑥)
4𝜋𝑘𝑡 −∞ 2𝑘𝑡
 1 ∞ −𝑘𝑡𝜔2
𝑐=𝑔 0 = න 𝑒 d𝜔
2𝜋 −∞

Let 𝑣 = 𝑘𝑡𝜔

∞
1 2
= න 𝑒 −𝑣 d𝑣
2𝜋 𝑘𝑡 −∞
1 1
−4𝑘𝑡𝑥2
⟹ 𝑔(𝑥, 𝑡) = 𝑒
= 𝜋 2 𝜋𝑘𝑡

1
=
2 𝜋𝑘𝑡
 ∞ 2
Side note: ‫׬‬−∞ 𝑒 −𝑥 𝑑𝑥 = 𝜋

Proof:

∞ 2 2 ∞ 2 ∞ 2
Let ∞ −𝑥2 𝛼 2 = ‫׬‬−∞ 𝑒 −𝑥 𝑑𝑥 = ‫׬‬−∞ 𝑒 −𝑥 𝑑𝑥 ‫׬‬−∞ 𝑒 −𝑦 𝑑𝑦 =
‫׬‬−∞ 𝑒 𝑑𝑥 = 𝛼 ∞ ∞ −(𝑥2+𝑦2) 2𝜋 ∞ 2
‫׬‬−∞ ‫׬‬−∞ 𝑒 𝑑𝑥𝑑𝑦 = ‫׬‬0 ‫׬‬0 𝑒 −𝑟 𝑟 𝑑𝑟𝑑𝜃

𝑥 = 𝑟 cos 𝜃; 𝑦 = 𝑟 sin 𝜃; 𝑥2 + 𝑦2 = 𝑟2

2𝜋 ∞ 2 ∞ 1 2 1
= ‫׬‬0 1 𝑑𝜃 ‫׬‬0 𝑟𝑒 −𝑟 𝑑𝑟 = 2𝜋 ∙ − 2 𝑒 −𝑟 ȁ 0 = 2𝜋 ∙ 2 = 𝜋

𝛼= 𝜋
‫)𝑡 ‪𝑢 𝑥, 𝑡 = 𝑓 𝑥 ∗ 𝑔(𝑥,‬‬

‫‪1‬‬ ‫∞‬ ‫𝑡𝑘‪− 𝑥−𝜉 2/4‬‬

‫‪=2‬‬ ‫׬‬ ‫)𝜉(𝑓‬ ‫𝑒‬ ‫𝜉𝑑‬
‫∞‪𝜋𝑘𝑡 −‬‬
BIBO (Bounded input bounded output)

∞ ∞
𝑦 𝑡 =න ℎ 𝑡 − 𝜏 𝑥 𝜏 𝑑𝜏 = න ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
−∞ −∞

∞ ∞
𝑦 𝑡 = න ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏 ≤ න ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
−∞ −∞

∞
𝐼𝑓 𝑥 𝑡 ≤ 𝐵 𝑓𝑜𝑟 − ∞ < 𝑡 < ∞ ≤ 𝐵 ‫׬‬−∞ ℎ 𝜏 𝑑𝜏

∞
Therefore, if ‫׬‬−∞ ℎ 𝜏 𝑑𝜏 ≤ ∞, then 𝑦 𝑡 <∞ (bounded output)
 ∞
𝐻(𝑠) = න ℎ(𝑡)𝑒 −𝑠𝑡 𝑑𝑡
0

∞
𝐻(𝑠) ≤ න ℎ(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡
0

∞
= න ℎ(𝑡) 𝑒 −(𝜎+𝑗𝜔)𝑡 𝑑𝑡
0

∞
= න ℎ(𝑡) 𝑒 −𝜎𝑡 ∙ 𝑒 𝑗𝜔𝑡 𝑑𝑡
0

∞ 𝜎=0⟶ 𝐻(𝑠) 𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒𝑙𝑦 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒

= න ℎ(𝑡) 𝑒 −𝜎𝑡 𝑑𝑡
0 ⟹ 𝑅𝑒 𝑠 ≥ 0 𝑖𝑠 𝑡ℎ𝑒 region of convergence