Transforms and PDE – Unit II: Fourier Transforms Short Notes

Transforms and Partial Differential Equations

UNIT II – FOURIER TRANSFORMS
Short Notes | Definitions, Properties and Solved Problems

PART – A (Short Answer Questions)

1. Define Fourier Transform and Inverse Fourier Transform.

The Fourier Transform of f(x) is defined as:

F[f(x)] = F(s) = (1/√(2π)) ∫ f(x) eisx dx (from −∞ to ∞)

The Inverse Fourier Transform is defined as:

f(x) = F−1[F(s)] = (1/√(2π)) ∫ F(s) e−isx ds (from −∞ to ∞)

Note: F[f(x)] and f(x) = F ¹[F(s)] together form the Fourier Transform Pair.

2. Define Fourier Cosine Transform and its Inverse.

F [f(x)] = F (s) = √(2/π) ∫₀∞ f(x) cos(sx) dx

f(x) = √(2/π) ∫₀∞ F (s) cos(sx) ds (Inverse Fourier Cosine Transform)

Applicable when f(x) is an even function.

3. Define Fourier Sine Transform and its Inverse.

F [f(x)] = F (s) = √(2/π) ∫₀∞ f(x) sin(sx) dx

f(x) = √(2/π) ∫₀∞ F (s) sin(sx) ds (Inverse Fourier Sine Transform)

Applicable when f(x) is an odd function.

4. State the Convolution Theorem for Fourier Transforms. (AU 2000, 2003, 2008)

The convolution of f(x) and g(x) over (−∞, ∞) is:

f * g = ∫ ∞∞ f(u) g(x − u) du

Convolution Theorem: The Fourier Transform of the convolution of f(x) and g(x) equals the product
of their individual Fourier Transforms:

F[f(x) * g(x)] = F[f(x)] · G[g(x)] = F(s) · G(s)

5. State Parseval's Identity for Fourier Transforms. (AU 2002, 2007, 2008, 2010, 2011)
If F[f(x)] = F(s), then:

∫ ∞∞ |f(x)|² dx = ∫ ∞∞ |F(s)|² ds

This identity relates the total energy in the time domain to the total energy in the frequency domain.

Department of Mathematics Page 1 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

6. State Parseval's Identity for Fourier Sine and Cosine Transforms.

If F [f(x)] = F (s) and F [f(x)] = F (s), then:

(i) ∫₀∞ |f(x)|² dx = ∫₀∞ |F (s)|² ds = ∫₀∞ |F (s)|² ds

(ii) ∫₀∞ f(x)g(x) dx = ∫₀∞ F (s) G (s) ds = ∫₀∞ F (s) G (s) ds

7. State the Conjugate Symmetry Property of Fourier Transforms.

If F[f(x)] = F(s), then:

F[f(−x)] = ̅F(s) (complex conjugate of F(s))

Proof uses substitution −x = t with the Fourier integral definition.

8. Write the conditions for existence of Fourier Transform.

A function f(x) has a Fourier Transform if:
(i) f(x) is absolutely integrable over (−∞, ∞), i.e., ∫ ∞∞ |f(x)| dx is finite.
(ii) f(x) has only a finite number of maxima, minima and discontinuities in any finite interval.

9. Find the Fourier Transform of f(x) = 1 for |x| ≤ a and f(x) = 0 for |x| > a.
Sol:
F[f(x)] = (1/√(2π)) ∫ ₐₐ 1 · eisx dx
= (1/√(2π)) [eisx/is] ₐₐ
= (1/√(2π)) · (eias − e−ias) / (is)
= √(2/π) · sin(as) / s
Hence F[f(x)] = √(2/π) · (sin as)/s

10. State the unit impulse function (Dirac delta function) and its Fourier Transform.
The Dirac Delta function δ(x − a) is defined as:

δ(x − a) = lim(h→0) f(x) where f(x) = 1/h for a − h/2 ≤ x ≤ a + h/2, else 0

F[δ(x − a)] = e−ias

F[δ(x)] = 1 (for a = 0)

11. State the unit step function and its Fourier Transform.

uₐ(x) = u(a − x) = 1, x ≥ a | 0, x < a

F[uₐ(x)] = e−ias / (is)

F[u₀(x)] = 1/(is) = −i/s

12. Write the derivatives of Fourier Sine and Cosine Transforms.

Transform Formula

F {f'(x)} sF {f(x)} − f(0)

F {f''(x)} −s² F {f(x)} − f'(0)

Department of Mathematics Page 2 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

F {f'(x)} −sF {f(x)}

F {f''(x)} −s² F {f(x)} + sf(0)

13. If F [f(x)] = F (s), find d/ds{F (s)} and d/ds{F (s)}.

Sol: Using Leibniz rule for differentiation under the integral sign:
d/ds {F (s)} = −F {xf(x)}
d/ds {F (s)} = F {xf(x)}

14. Find the Fourier Sine Transform of 1/x. (AU 2008, 2009)
Sol:
F [1/x] = √(2/π) ∫₀∞ (1/x) sin(sx) dx
Let sx = θ, then dx = dθ/s
= √(2/π) ∫₀∞ (sin θ/θ) dθ = √(2/π) · π/2
Hence F [1/x] = √(π/2)

15. Find the Fourier Cosine Transform of e−ax, a > 0.

Sol:
F [e−ax] = √(2/π) ∫₀∞ e−ax cos(sx) dx
= √(2/π) · a / (a² + s²)

16. Find the Fourier Sine Transform of e−ax, a > 0. (AU 2007, 2009, 2012)
Sol:
F [e−ax] = √(2/π) ∫₀∞ e−ax sin(sx) dx
= √(2/π) · s / (a² + s²)

PROPERTIES OF FOURIER TRANSFORMS – SUMMARY

Property Formula

1. Linearity F[c₁f₁(x) + c₂f₂(x)] = c₁F[f₁(x)] + c₂F[f₂(x)]

2. Change of Scale F[f(ax)] = (1/|a|) F(s/a)

3. Shifting (time) F[f(x − a)] = e<super>ias</super> F(s)

4. Shifting (freq) F[e<super>iax</super>f(x)] = F(s + a)

5. Modulation F[f(x)cos(ax)] = ½{F(s+a) + F(s−a)}

6. Transform of f'(x) F[f'(x)] = −is F(s)

7. Transform of f^n(x) F[f^n(x)] = (-is)^n F(s)

8. x^n multiplication F[x^n f(x)] = (-i)^n d^n/ds^n F(s)

9. Convolution F[f*g] = F(s) · G(s)

10. Parseval's ∫|f(x)|²dx = ∫|F(s)|²ds

Department of Mathematics Page 3 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

PROPERTIES OF FOURIER SINE AND COSINE TRANSFORMS

Property Formula

Change of Scale F {f(ax)} = (1/a)F (s/a); F {f(ax)} = (1/a)F (s/a)

Modulation (i) F {f(x)cos(ax)} = ½[F (s+a) + F (s−a)]

Modulation (ii) F {f(x)sin(ax)} = ½[F (s+a) + F (a−s)]

Modulation (iii) F {f(x)cos(ax)} = ½[F (s+a) + F (s−a)]

Modulation (iv) F {f(x)sin(ax)} = ½[F (s−a) − F (s+a)]

Derivative of F (s) d/ds{F (s)} = −F {xf(x)}

Derivative of F (s) d/ds{F (s)} = F {xf(x)}

PROOFS OF IMPORTANT PROPERTIES

1. Linearity Property
Statement: F[c₁f₁(x) + c₂f₂(x)] = c₁F[f₁(x)] + c₂F[f₂(x)]
Proof:
F[c₁f₁(x) + c₂f₂(x)] = (1/√(2π)) ∫ (c₁f₁(x) + c₂f₂(x)) eisx dx
= c₁ (1/√(2π)) ∫ f₁(x)eisx dx + c₂ (1/√(2π)) ∫ f₂(x)eisx dx
= c₁ F[f₁(x)] + c₂ F[f₂(x)] Proved.

2. Change of Scale Property (AU 2006, 2009, 2010, 2011)

Statement: If F[f(x)] = F(s), then F[f(ax)] = (1/|a|) F(s/a)
Proof: Let ax = t, so dx = dt/a. For a > 0:
F[f(ax)] = (1/√(2π)) ∫ f(ax) eisx dx
= (1/√(2π)) ∫ f(t) eis(t/a) dt/a = (1/a) F(s/a)
For a < 0, limits reverse, giving −(1/a)F(s/a). Combining: F[f(ax)] = (1/|a|)F(s/a). Proved.

3. Shifting Property (AU 1999, 2000, 2001, 2006, 2007, 2008, 2010, 2011, 2012)
Statement: (i) F[f(x − a)] = eias F(s) (ii) F[eiaxf(x)] = F(s + a)
Proof of (i): Put t = x − a, dt = dx:
F[f(x−a)] = (1/√(2π)) ∫ f(x−a) eisx dx = (1/√(2π)) ∫ f(t) eis(t+a) dt
= eisa · (1/√(2π)) ∫ f(t) eist dt = eias F(s) Proved.
Proof of (ii):
F[eiaxf(x)] = (1/√(2π)) ∫ eiaxf(x) eisx dx
= (1/√(2π)) ∫ f(x) ei(s+a)x dx = F(s+a) Proved.

4. Modulation Theorem (AU 2001, 2003, 2010)

Statement: If F[f(x)] = F(s), then F[f(x)cos(ax)] = ½{F(s+a) + F(s−a)}
Proof: Using cos(ax) = (eiax + e−iax)/2:

Department of Mathematics Page 4 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

F[f(x)cos(ax)] = (1/√(2π)) ∫ f(x) cos(ax) eisx dx

= (1/2) ∫ f(x)(eiax + e−iax) eisx dx
= (1/2)[∫ f(x)ei(s+a)xdx + ∫ f(x)ei(s−a)xdx]
= ½{F(s+a) + F(s−a)} Proved.

5. Transform of Derivatives (AU 2005, 2008)

Statement: If f(x) is continuous and f'(x) is absolutely integrable with f(x) → 0 as x → ±∞, then F[f'(x)] = −is
F(s)
Proof: Integrating by parts:
F[f'(x)] = (1/√(2π)) ∫ f'(x) eisx dx
= (1/√(2π)) [{e−isxf(x)} ∞∞ + is ∫ f(x) eisxdx]
= (1/√(2π)) [0 + is ∫ f(x)eisxdx] = −is F(s)
Similarly: F[f''(x)] = (-is)^2 F(s), and in general F[f^n(x)] = (-is)^n F(s). Hence Proved.

PART – B (Long Answer Problems with Solutions)

1. Find the Fourier transform of f(x) = 1 for |x| < a and f(x) = 0 for |x| > a. Hence deduce that (i) ∫₀∞ (sin
t/t) dt = π/2 (ii) ∫₀∞ (sin t/t)² dt = π/2. (AU 2003, 2004, 2005)
Sol:
F[f(x)] = (1/√(2π)) ∫ ₐₐ 1 · eisx dx
Since f(x) = 1 on (−a, a) and eisx = cos(sx) + i·sin(sx); sin(sx) is odd so the imaginary part vanishes:
F[f(x)] = (1/√(2π)) ∫ ₐₐ cos(sx) dx = √(2/π) · sin(as)/s = F(s)
Deduction (i): By the Fourier inversion formula:
f(x) = (1/√(2π)) ∫ ∞∞ F(s) e−isx ds
= (2/π) ∫₀∞ (sin(as)/s) cos(sx) ds
Putting x = 0, f(0) = 1, as = t:
1 = (2/π) ∫₀∞ (sin t/t) dt ∫₀∞ (sin t/t) dt = π/2
Deduction (ii): By Parseval's identity:
∫ ∞∞ |f(x)|² dx = ∫ ∞∞ |F(s)|² ds
∫ ₐₐ 1² dx = ∫ ∞∞ (2/π)(sin(as)/s)² ds
2a = (4/π) ∫₀∞ (sin(as)/s)² ds
Substituting as = t, ds = dt/a:
∫₀∞ (sin t/t)² dt = π/2

2. Find the Fourier transform of f(x) = 1 − x² for |x| < 1 and f(x) = 0 for |x| > 1. Hence evaluate (i) ∫₀∞
(s·cos s − sin s)/s³ · cos(s/2) ds = 3π/16 (ii) ∫₀∞ [(sin s − s·cos s)/s³]² ds = π/15. (AU 2000, 2001, 2004, 2005,
2006, 2007, 2010, 2011)
Sol:
F[f(x)] = (1/√(2π)) ∫ ¹¹ (1−x²)(cos sx + i·sin sx) dx
Since (1−x²)sin(sx) is odd, the imaginary part vanishes:

Department of Mathematics Page 5 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

F[f(x)] = (2/√(2π)) ∫₀¹ (1−x²) cos(sx) dx

Applying Bernoulli's theorem (integration by parts repeatedly):
= √(2/π) {(1−x²)(sin sx/s) − (−2x)(−cos sx/s²) + (−2)(−sin sx/s³)}₀¹
= √(2/π) {0 − 2cos s/s² + 2sin s/s³}
F[f(x)] = √(2/π) · (2·sin s − 2s·cos s)/s³ = F(s)
Deduction (i): By the inversion formula and using x = 1/2:
f(1/2) = 1 − (1/4) = 3/4
∫₀∞ (sins − s·cos s)/s³ · cos(s/2) ds = (π/4)·(3/4) = 3π/16
Deduction (ii): By Parseval's identity:
∫ ¹¹ (1−x²)² dx = ∫ ∞∞ |F(s)|² ds
2∫₀¹ (1−2x²+x⁴) dx = (8/π)∫₀∞ (sin s − s·cos s)²/s⁶ ds
[x − 2x³/3 + x⁵/5]₀¹ = (8/π)∫₀∞ ...
2(1 − 2/3 + 1/5) = 16/15 = (8/π)∫₀∞ (sin s−s·cos s)²/s⁶ ds
∫₀∞ [(sin s − s·cos s)/s³]² ds = π/15

3. Find the Fourier transform of f(x) = 1 − |x| for |x| < 1 and f(x) = 0 for |x| > 1. Hence evaluate (i) ∫₀∞
(sin t/t)² dt = π/2 (ii) ∫₀∞ (sin t/t)⁴ dt = π/3. (AU 2001, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012)
Sol:
Since f(x) is even (|x| symmetric), the imaginary part of the integral vanishes:
F[f(x)] = (2/√(2π)) ∫₀¹ (1−x) cos(sx) dx
Integrating by parts (Bernoulli's theorem):
= √(2/π) {(1−x)(sin sx/s) − (−1)(−cos sx/s²)}₀¹
= √(2/π) [{0 − cos s/s²} − {0 − 1/s²}]
F[f(x)] = √(2/π) · (1−cos s)/s² = F(s)
Note: (1 − cos s)/s² = 2sin²(s/2)/s²
Deduction (i): By inversion formula, put x = 0, f(0) = 1:
1 = (2/π) ∫₀∞ (1−cos s)/s² ds
∫₀∞ 2sin²(s/2)/s² ds = π/2
Let t = s/2:
∫₀∞ (sin t/t)² dt = π/2
Deduction (ii): By Parseval's identity:
∫ ¹¹ (1−|x|)² dx = ∫ ∞∞ (2/π)(1−cos s)²/s⁴ ds
2/3 = (2/π) ∫ ∞∞ (2sin²(s/2))²/s⁴ ds
Substituting t = s/2:
∫₀∞ (sin t/t)⁴ dt = π/3

4. Find the Fourier transform of f(x) = a² − x² for |x| < a and f(x) = 0 for |x| > a. Hence show that (i) ∫₀∞
(sin t − t·cos t)/t³ dt = π/4 (ii) ∫₀∞ (sin t − t·cos t)²/t⁶ dt = π/15. (AU 1996, 2001, 2004, 2008, 2009, 2010,
2011, 2012)
Sol:

Department of Mathematics Page 6 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

Since (a²−x²)sin(sx) is odd:

F[f(x)] = (2/√(2π)) ∫₀ₐ (a²−x²) cos(sx) dx
By Bernoulli's theorem:
= √(2/π) {(a²−x²)(sin sx/s) − (−2x)(−cos sx/s²) + (−2)(−sin sx/s³)}₀ₐ
= √(2/π) {0 − 2a·cos(as)/s² + 2sin(as)/s³}
F[f(x)] = 2√(2/π) · (sin(as) − as·cos(as))/s³ = F(s)
Deduction (i): By inversion formula, put x = 0, f(0) = a²:
∫₀∞ (sin(as) − as·cos(as))/s³ ds = πa²/4
Substituting as = t:
∫₀∞ (sin t − t·cos t)/t³ dt = π/4
Deduction (ii): By Parseval's identity:
∫ ₐₐ (a²−x²)² dx = ∫ ∞∞ |F(s)|² ds
2∫₀ₐ (a⁴−2a²x²+x⁴)dx = (8/π)∫₀∞ (sin as − as cos as)²/s⁶ ds
[a⁴x − 2a²x³/3 + x⁵/5]₀ₐ = a⁵(1−2/3+1/5) = 8a⁵/15
∫₀∞ [(sint−t·cost)/t³]² dt = π/15

5. Find the Fourier transform of e−a²x². Hence prove that e−x²/2 is self-reciprocal with respect to Fourier
transforms. Also find the Fourier cosine transform of e−x². (AU 2007, 2009)
Sol:
F[e−a²x²] = (1/√(2π)) ∫ ∞∞ e−a²x²+isx dx
Completing the square in the exponent:
−a²x² + isx = −[a²x² − isx + (is/2a)²] − s²/4a²
F[e−a²x²] = (1/√(2π)) e−s²/4a² ∫ ∞∞ e−(ax−is/2a)² dx
Let ax − is/2a = t, so adx = dt; using Gaussian integral ∫ e−t²dt = √π:
= (1/√(2π)) e−s²/4a² · √π/a
F[e−a²x²] = (1/a√2) e−s²/4a²
Self-reciprocal: Put a = 1/√2:
F[e−x²/2] = e−s²/2
This means e−x²/2 is its own Fourier transform — it is self-reciprocal.
Fourier Cosine Transform of e−x²:
Put a = 1 in the above (real part of F[e−x²eisx]):
F [e−x²] = (1/√2) e−s²/4

6. Find the Fourier sine transform of f(x) = e−ax, a > 0. Hence deduce that ∫₀∞ s·sin(sx)/(a²+s²) ds =
(π/2)e−ax. (AU 2007, 2009, 2012)
Sol:
F [e−ax] = √(2/π) ∫₀∞ e−ax sin(sx) dx
= √(2/π) [e−ax(−a·sin sx − s·cos sx)/(a²+s²)]₀∞
= √(2/π) [0 − (0 − s)/(a²+s²)]
F [e−ax] = √(2/π) · s/(a²+s²)

Department of Mathematics Page 7 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

Deduction: By the inverse Fourier sine transform:

f(x) = √(2/π) ∫₀∞ F (s) sin(sx) ds
e−ax = (2/π) ∫₀∞ s/(a²+s²) · sin(sx) ds
∫₀∞ s·sin(sx)/(a²+s²) ds = (π/2) e−ax
Replacing s=x, x=m, a=1: ∫₀∞ x·sin(mx)/(1+x²) dx = (π/2)e−m

7. Find the Fourier sine transform of e−|x|. Hence show that ∫₀∞ xs·sin(ax)/(1+x²) dx = (π/2)e−a. (AU
2009, 2011)
Sol:
F [e−|x|] = √(2/π) ∫₀∞ e−x sin(sx) dx
= √(2/π) [e−x(−sin sx − s·cos sx)/(1+s²)]₀∞
= √(2/π) [0 + s/(1+s²)]
F [e−|x|] = √(2/π) · s/(1+s²)
By inversion formula:
e−|x| = (2/π) ∫₀∞ s/(1+s²) · sin(sx) ds
Replacing x=a and s=x:
∫₀∞ x·sin(ax)/(1+x²) dx = (π/2) e−a

8. Find the Fourier cosine transform of e−ax and Fourier sine transform of xe−ax. Also find F [xe−ax]
and F [xe−ax]. (AU 2011)
Sol: Using the result F [e−ax] = √(2/π) · a/(a²+s²):
We use the differentiation formula:
d/ds {F (s)} = −F {xf(x)}
F [xe−ax] = d/ds {F [e−ax]}
= d/ds {√(2/π) · s/(a²+s²)}
= √(2/π) · (a²+s²−2s²)/(a²+s²)²
F [xe−ax] = √(2/π) · (s²−a²)/(a²+s²)²
For F [xe−ax]:
F [xf(x)] = −d/ds {F [f(x)]} = −d/ds {√(2/π) · a/(a²+s²)}
F [xe−ax] = √(2/π) · 2as/(a²+s²)²

9. Find the Fourier cosine transform of e−ax + 3e−bx. (AU 2006, 2008, 2010, 2011)
Sol: Using linearity and the known result F [e−ax] = √(2/π)·a/(a²+s²):
F [e−ax + 3e−bx] = F [e−ax] + 3F [e−bx]
= √(2/π) · a/(a²+s²) + 3√(2/π) · b/(b²+s²)
= √(2/π) [a/(a²+s²) + 3b/(b²+s²)]

10. Find the Fourier sine transform of e−ax/x, a > 0. Hence deduce that ∫₀∞ (e−ax−e−bx)/x · sin(sx) dx =
tan ¹(b/s) − tan ¹(a/s). (AU 2005, 2007, 2009)
Sol: Let I = F [e−ax/x] = √(2/π) ∫₀∞ (e−ax/x) sin(sx) dx

Department of Mathematics Page 8 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

Differentiating I with respect to s by Leibniz rule:

dI/ds = √(2/π) ∫₀∞ e−ax cos(sx) dx = √(2/π) · a/(a²+s²)
Integrating both sides with respect to s:
I = √(2/π) · tan ¹(s/a) + C
At s = 0, I = 0, so C = 0. Therefore:
F [e−ax/x] = √(2/π) · tan ¹(s/a)
By linearity:
F [(e−ax−e−bx)/x] = √(2/π)[tan ¹(s/b)−tan ¹(s/a)]
= √(2/π)[tan ¹(b/s)−tan ¹(a/s)] (using tan ¹θ + tan ¹(1/θ) = π/2)
∫₀∞ (e−ax−e−bx)/x · sin(sx) dx = tan ¹(b/s) − tan ¹(a/s)

11. Evaluate ∫₀∞ dx/[(a²+x²)(b²+x²)] using the convolution theorem. (AU 2001, 2005, 2006, 2009, 2010,
2011)
Sol: Using Parseval's identity for cosine transforms:
∫₀∞ f(x)g(x) dx = ∫₀∞ F (s) G (s) ds
Let f(x) = e−ax and g(x) = e−bx. Then:
F [e−ax] = √(2/π) · a/(a²+s²) and F [e−bx] = √(2/π) · b/(b²+s²)
Applying the theorem:
∫₀∞ e−ax·e−bx dx = ∫₀∞ (2/π)·ab/[(a²+s²)(b²+s²)] ds
∫₀∞ e−(a+b)x dx = (2ab/π) ∫₀∞ ds/[(a²+s²)(b²+s²)]
[e−(a+b)x/(−(a+b))]₀∞ = 1/(a+b)
∫₀∞ dx/[(a²+x²)(b²+x²)] = π/[2ab(a+b)]

12. Evaluate ∫₀∞ dx/(a²+x²)² and ∫₀∞ x²dx/(a²+x²)² using Parseval's identity. (AU 2001, 2010, 2011)
Sol (i): Using Parseval's identity with f(x) = g(x) = e−ax:
∫ ∞∞ |f(x)|² dx = ∫ ∞∞ |F (s)|² ds
∫ ∞∞ e−2ax dx = ∫ ∞∞ (2/π)·a²/(a²+s²)² ds
1/(2a) × 2 = (2a²/π) × 2 ∫₀∞ 1/(a²+s²)² ds
∫₀∞ 1/(a²+s²)² ds = π/(4a³) (replacing s by x)
Sol (ii): Using F [e−ax] = √(2/π)·s/(a²+s²):
∫ ∞∞ e−2ax dx = ∫ ∞∞ (2/π)·s²/(a²+s²)² ds
∫₀∞ x²/(a²+x²)² dx = π/(4a)

13. Solve the integral equation ∫₀∞ f(x)cos(λx) dx = 1 − λ for 0 ≤ λ ≤ 1 and 0 for λ > 1. Hence evaluate ∫₀∞
(sin t/t)² dt. (AU 2001)
Sol:
√(2/π) ∫₀∞ f(x)cos(λx) dx = √(2/π){1−λ for 0≤λ≤1, 0 for λ>1}
F [f(x)] = √(2/π){1−λ for 0≤λ≤1, 0 for λ>1}
By the inverse Fourier cosine transform:

Department of Mathematics Page 9 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

f(x) = √(2/π) ∫₀¹ (1−λ) cos(λx) dλ

= (2/π) [(1−λ)(sin λx/x) − (−1)(−cos λx/x²)]₀¹
f(x) = (2/π) · (1−cos x)/x² = (4/πx²) sin²(x/2)
Also: F [f(x)] = √(2/π)(1−λ) for 0≤λ≤1, so by Parseval:
∫₀∞ f²(x) dx = ∫₀¹ (2/π)(1−λ)² dλ = (2/3π)
∫₀∞ (4/πx²)² sin⁴(x/2) dx = 2/3π
Substituting t = x/2:
∫₀∞ (sin t/t)² dt = π/2

14. Find the Fourier sine and cosine transforms of xn−1 and prove that 1/√x is self-reciprocal under
both Fourier sine and cosine transforms. (AU 2005, 2007)
Sol:
Using the Gamma function: Γn = ∫₀∞ e−x xn−1 dx. Replacing x by ax:
Γn = aⁿ ∫₀∞ e−ax xn−1 dx ∫₀∞ e−ax xn−1 dx = Γn/aⁿ
Putting a = is:
∫₀∞ e−isx xn−1 dx = Γn/(is)ⁿ = (−i)ⁿ Γn/sⁿ
Since (−i)ⁿ = [cos(π/2)−i sin(π/2)]ⁿ = e−inπ/2 = cos(nπ/2) − i sin(nπ/2)
Equating real and imaginary parts:
F [xn−1] = √(2/π) · Γn/sⁿ · cos(nπ/2)
F [xn−1] = √(2/π) · Γn/sⁿ · sin(nπ/2)
Self-reciprocal (1/√x): Put n = 1/2; Γ(1/2) = √π:
F [1/√x] = √(2/π) · √π/√s · cos(π/4) = 1/√s
F [1/√x] = 1/√s
1/√x is self-reciprocal under both Fourier cosine and sine transforms.

15. Find the Fourier cosine transform of e−x². (AU 2006, 2008, 2010, 2011)
Sol: Let I = F [e−x²] = √(2/π) ∫₀∞ e−x² cos(sx) dx
This is the real part of √(2/π) ∫ ∞∞ e−x²+isx dx. Completing the square:
F [e−x²] = (1/√2) e−s²/4 · (1/√π) √π = (1/√2) e−s²/4

16. Find the Fourier cosine transform of f(x) = e−ax, a > 0. Hence deduce ∫₀∞ cos(2x)/(x²+16) dx and ∫₀∞
x·sin(2x)/(x²+16) dx. (AU 2008, 2011)
Sol:
F [e−4x] = √(2/π) · 4/(16+s²)
By inverse cosine transform:
e−4x = (2/π) ∫₀∞ 4/(16+s²) · cos(sx) ds
∫₀∞ cos(sx)/(16+s²) ds = (π/8) e−4x
Put x = 2:
∫₀∞ cos(2x)/(x²+16) dx = (π/8) e−8
Differentiating with respect to x (or s):

Department of Mathematics Page 10 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

∫₀∞ x·sin(2x)/(x²+16) dx = (π/2) e−8

17. Find the Fourier sine transform of x/(a²+x²) and Fourier cosine transform of 1/(a²+x²). (AU 2002,
2010)
Sol: Let I = F [1/(a²+x²)] = √(2/π) ∫₀∞ cos(sx)/(a²+x²) dx
By differentiating under the integral and solving the resulting ODE:
d²I/ds² = a²I I = Ae−as + Beas
Using boundary conditions I is bounded as s→∞, so B=0; I₀ = √(π/2)/a, giving:
F [1/(a²+x²)] = √(π/2) · e−as/a
Then using F {xf(x)} = −d/ds{F (s)}:
F [x/(a²+x²)] = √(π/2) · e−as

18. Solve the integral equation ∫₀∞ f(x)cos(λx) dx = e−λ. Also show that ∫₀∞ cos(λx)/(1+λ²) dλ = (π/2)e−x.
(AU 1996)
Sol: Given ∫₀∞ f(x)cos(λx) dx = e−λ
√(2/π) ∫₀∞ f(x)cos(λx) dx = √(2/π) e−λ
F [f(x)] = √(2/π) e−λ
By inverse cosine transform:
f(x) = √(2/π) ∫₀∞ √(2/π) e−λ cos(λx) dλ = (2/π) ∫₀∞ e−λ cos(λx) dλ
= (2/π) [e−λ(−cos(λx) + x sin(λx))/(1+x²)]₀∞
f(x) = (2/π) · 1/(1+x²)
Taking Fourier cosine transform of f(x) = 2/[π(1+x²)]:
∫₀∞ cos(λx)/(1+x²) dx = (π/2)e−λ

19. Prove that the Fourier Transform of the Fourier Transform of f(x) is 2π·f(−x).
Sol: Let F[f(x)] = F(s). Then:
F[F(s)] = (1/√(2π)) ∫ ∞∞ F(s) eixs ds
By the Fourier inversion formula:
(1/√(2π)) ∫ ∞∞ F(s) eixs ds = f(x)
But (1/√(2π)) ∫ F(s) eixs ds = f(−x) (replacing x by −x in inversion formula)
F[F(s)](x) = 2π·f(−x)

20. Find the Fourier sine transform of f(x) = sin(x) for 0 < x < π and f(x) = 0 for x ≥ π. (AU 2002, 2010)
Sol:
F [f(x)] = √(2/π) ∫₀π sin(x) sin(sx) dx
= (1/2)√(2/π) ∫₀π [cos((s−1)x) − cos((s+1)x)] dx
= (1/2)√(2/π) [{sin(s−1)x/(s−1) − sin(s+1)x/(s+1)}₀π]
= (1/√(2π)) [{sin(s−1)π/(s−1)} − {sin(s+1)π/(s+1)}]
= √(2/π) · sin(sπ) / (1−s²) (for s ≠ 1)
For s = 1: F [f(x)] = √(2/π) ∫₀π sin²(x) dx = √(π/2)/2

Department of Mathematics Page 11 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

F [f(x)] = √(2/π) · sin(sπ) / (1−s²) (s ≠ 1)

21. Find the function whose Fourier sine transform is e−as/s, a > 0. Hence find the value of f(x). (AU
2010, 2011)
Sol: Given F [f(x)] = e−as/s. By inverse sine transform:
f(x) = √(2/π) ∫₀∞ (e−as/s) sin(sx) ds = I (say)
Differentiating I w.r.t. x:
f'(x) = √(2/π) ∫₀∞ e−as cos(sx) ds = √(2/π) · a/(a²+x²)
Integrating with respect to x:
f(x) = √(2/π) · tan ¹(x/a) + C
Since f(0) = 0 (from integral with F = e−as/s at x=0), C = 0:
f(x) = √(2/π) · tan ¹(x/a)

STANDARD RESULTS – QUICK REFERENCE

f(x) F[f(x)] or F /F (s)

e<super>−ax</super>, a>0 (FCT) √(2/π) a/(a²+s²)

e<super>−ax</super>, a>0 (FST) √(2/π) s/(a²+s²)

1/x (FST) √(π/2)

1/(a²+x²) (FCT) √(π/2) e<super>−as</super>/a

x/(a²+x²) (FST) √(π/2) e<super>−as</super>

e<super>−a²x²</super> (1/a√2) e<super>−s²/4a²</super>

e<super>−x²/2</super> e<super>−s²/2</super> (self-reciprocal)

1/√x (FCT & FST) 1/√s (self-reciprocal)

x<super>n−1</super> (FCT) √(2/π) Γn/sⁿ cos(nπ/2)

x<super>n−1</super> (FST) √(2/π) Γn/sⁿ sin(nπ/2)

e<super>−as</super>/s (inv FST) √(2/π) tan ¹(x/a)

δ(x−a) (FT) e<super>−ias</super>

Unit step uₐ(x) e<super>−ias</super>/(is)

IMPORTANT INTEGRAL RESULTS (DEDUCED FROM TRANSFORMS)

Integral Result

∫₀∞ sin(t)/t dt π/2

∫₀∞ (sin t/t)² dt π/2

∫₀∞ (sin t/t)⁴ dt π/3

∫₀∞ (sint−t·cost)/t³ dt π/4

Department of Mathematics Page 12 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

∫₀∞ (sint−t cost)²/t⁶ dt π/15

∫₀∞ (scoss−sins)²/s⁶ ds π/15

∫₀∞ 1/(a²+x²)² dx π/(4a³)

∫₀∞ x²/(a²+x²)² dx π/(4a)

∫₀∞ dx/[(a²+x²)(b²+x²)] π/[2ab(a+b)]

∫₀∞ x·sin(ax)/(1+x²) dx (π/2)e<super>−a</super>

∫₀∞ s·sin(sx)/(a²+s²) ds (π/2)e<super>−ax</super>

∫₀∞ cos(λx)/(1+x²) dx (π/2)e<super>−λ</super>

ADDITIONAL PART – A QUESTIONS

Q. Find the Fourier cosine transform of f(x) = cos x for 0 < x < 1 and 0 for x ≥ 1.
Sol:
F [f(x)] = √(2/π) ∫₀¹ cos(x) cos(sx) dx = (1/2)√(2/π) ∫₀¹ [cos(s+1)x + cos(1−s)x] dx
= (1/2)√(2/π) [{sin(s+1)x/(s+1) + sin(1−s)x/(1−s)}₀¹]
F [f(x)] = √(1/(2π)) [sin(s+1)/(s+1) + sin(1−s)/(1−s)]

Q. If F [f(x)] = F (s) and F [f(x)] = F (s), state the relationship between them when f'(x)→0 as
x→∞.

F {f'(x)} = sF {f(x)} − f(0)

F {f'(x)} = −sF {f(x)} [if f(x)→0 as x→∞]

Q. Find the Fourier sine transform of f(x) = sin(x) in 0 < x < π and verify using known properties.
Sol: Already solved in Problem 20. Result:
F [sin x · I(0,π)] = √(2/π) · sin(sπ)/(1−s²) for s ≠ 1

Q. What is meant by "self-reciprocal" under Fourier transforms? Give two examples.

A function f(x) is said to be self-reciprocal under a Fourier transform if its transform equals the
function itself (up to scaling).

Examples:

(i) F[e−x²/2] = e−s²/2 (self-reciprocal under FT)

(ii) F [1/√x] = 1/√s and F [1/√x] = 1/√s (self-reciprocal under FCT and FST)

PROOF OF PARSEVAL'S IDENTITY FOR FOURIER TRANSFORMS

Statement: If F[f(x)] = F(s), then ∫ ∞∞ |f(x)|² dx = ∫ ∞∞ |F(s)|² ds

Proof: By the Convolution Theorem:

Department of Mathematics Page 13 All Rights Reserved

Transforms and PDE – Unit II: Fourier Transforms Short Notes

F[f*g] = F(s)·G(s) f*g = F ¹[F(s)G(s)]

∫ ∞∞ f(u)g(x−u)du = ∫ ∞∞ F(s)G(s) e−isx ds
Put x = 0:
∫ ∞∞ f(u)g(−u)du = ∫ ∞∞ F(s)G(s) ds
Take g(u) = ̅f(−u) (complex conjugate of f(−u)). Then G(s) = ̅F(s) and g(−u) = ̅f(u):
∫ ∞∞ f(u)̅f(u) du = ∫ ∞∞ F(s)̅F(s) ds
∫ ∞∞ |f(x)|² dx = ∫ ∞∞ |F(s)|² ds Proved.

PROOF OF CONVOLUTION THEOREM FOR FOURIER TRANSFORMS

Statement: F[f*g] = F(s)·G(s)

Proof:
F[f(x)*g(x)] = F[∫ ∞∞ f(u)g(x−u)du]
= (1/√(2π)) ∫ ∞∞ eisx [∫ ∞∞ f(u)g(x−u)du] dx
Changing order of integration:
= ∫ ∞∞ f(u) [∫ ∞∞ eisxg(x−u)dx] du
Let x − u = t, dx = dt:
= ∫ ∞∞ f(u) eisu[(1/√(2π))∫ eistg(t)dt] du
= ∫ ∞∞ f(u) eisu G(s) du = G(s) ∫ ∞∞ f(u) eisu du
= F(s) · G(s) Proved.

Note: All integrals from −∞ to ∞ use the normalization constant 1/√(2π). Results marked (AU year) indicate
questions from Anna University examinations. sin(0)=0, cos(0)=1, e−∞=0, sin(nπ)=0, cos(nπ)=(−1)ⁿ.

Department of Mathematics Page 14 All Rights Reserved