Home / Courses / GATE Physics / Practice / Fourier series

Fourier series

Mathematical methods · 33 questions and templates for GATE Physics, in order from warm-up to top rank. Check each answer, then read the full solution.

Loading questions… (they need JavaScript; nothing you type is sent anywhere)

Written questions with solutions

The same questions as a list, so you can print them or read without the practice box. Templates (fresh numbers) appear only in the practice box above.

  1. NATStandard

    In the complex form f=∑cneinx of a real function, cn=(an−ibn)/2 for n≥1. If a2=3 and b2=4, then |c2| is ______ (to one decimal place).
    Answer and solution

    Answer: 2.5 (±0.05)

    |c2|=12a22+b22=129+16=2.5. The amplitude of the second harmonic is a22+b22=5, shared equally between c2 and c−2.

    Why people get this wrong: Forgetting the factor 12; the real amplitude is split between e2ix and e−2ix.

  2. NATExam level

    For a symmetric square wave (values ±1), the fundamental (n=1) carries ______ % of the total mean-square power (to one decimal place).
    Answer and solution

    Answer: 81.1 % (±0.1)

    The mean square of the wave is 1. The fundamental b1sin⁡x with b1=4/π has mean square b12/2=8/π2=0.811. So 81.1% of the power is in the first harmonic; all the rest share under 19%.

    Why people get this wrong: Using b12 without the 12 (the average of sin2⁡) gives more than 100%.

  3. MSQExam level

    A real periodic function is written as f(x)=∑n=−∞∞cneinx. Which statements are true?
    1. c−n=cn∗
    2. c0 is the average value of f.
    3. cn=12π∫−ππf(x)e−inxdx
    4. |cn|=|bn| for every n≥1.
    Answer and solution

    Answer: A, B, C

    Multiply by e−imx and integrate: only n=m survives, giving the third formula. For real f, conjugating it gives c−n=cn∗. c0=12π∫f is the average. And |cn|=12an2+bn2, not |bn|.

    Why people get this wrong: Forgetting the minus sign in e−inx in the coefficient formula.

  4. MCQExam level

    A pulse train equals 1 for |x|<π/3 and 0 for π/3<|x|<π, repeated with period 2π. Which harmonics are missing from its Fourier series (apart from all sine terms)?
    1. every multiple of 3: n=3,6,9,…
    2. every even harmonic
    3. every odd harmonic
    4. none: every cosine harmonic is present
    Answer and solution

    Answer: A

    The function is even, so only cosines appear, with an=1π∫−π/3π/3cos⁡nxdx=2sin⁡(nπ/3)nπ. This vanishes when nπ/3 is a multiple of π, that is for n=3,6,9,…. In general a pulse of width 2d loses the harmonics with nd a multiple of π; only a 50% duty cycle loses all the even ones.

    Why people get this wrong: "Only odd harmonics" is true for a 50% duty cycle, not for this narrower pulse.

  5. MCQExam level

    A 2π-periodic function f(x)=∑cneinx is shifted to g(x)=f(x−x0). In the series g(x)=∑dneinx,
    1. |dn| differ from |cn|, but the phases are the same
    2. dn=cn for every n
    3. |dn|=|cn|, and the phase of the nth term changes by −nx0
    4. only the constant term d0 changes
    Answer and solution

    Answer: C

    g(x)=∑cnein(x−x0)=∑(cne−inx0)einx, so dn=cne−inx0. The modulus is unchanged and the phase shifts by an amount proportional to n. That is why the power in each harmonic (Parseval) does not depend on where you start the clock. The constant term, n=0, is the one coefficient that does not change.

    Why people get this wrong: A delay does not change how much of each harmonic there is, only its phase, and the phase shift grows with n.

  6. MSQExam level

    The half-wave rectified sine is f(x)=sin⁡x for 0<x<π and 0 for π<x<2π, repeated with period 2π. Which statements about its Fourier series are true?
    1. The constant term is 1/π
    2. It contains the term 12sin⁡x
    3. It contains cos⁡3x with a non-zero coefficient
    4. sin⁡x is the only sine term
    Answer and solution

    Answer: A, B, D

    Write f=12(sin⁡x+|sin⁡x|). Using |sin⁡x|=2π−4π∑k≥1cos⁡2kx4k2−1, f=1π+12sin⁡x−2π∑k≥1cos⁡2kx4k2−1. So the constant is 1/π (the average of a half sine over a full period), the only sine is 12sin⁡x, and the cosines are all of even order: no cos⁡3x.

    Why people get this wrong: Splitting into 12sin⁡x plus half of |sin⁡x| avoids the integrals and shows at once that only even cosines remain.

  7. NATExam level

    A pulse train equals 1 for |x|<π/3 and 0 for π/3<|x|<π, repeated with period 2π. In f=a02+∑(ancos⁡nx+bnsin⁡nx), the coefficient a1 is ______ (to three decimal places).
    Answer and solution

    Answer: 0.551 (±0.002)

    a1=1π∫−π/3π/3cos⁡xdx=2sin⁡(π/3)π=3π=0.551.

    Why people get this wrong: Integrate only where the pulse is on; the factor in front is 1/π, not 1/2π.

  8. NATHard

    Use f(x)=a02+∑n≥1(ancos⁡nx+bnsin⁡nx) on (−π,π), with an=1π∫−ππfcos⁡nxdx and bn=1π∫−ππfsin⁡nxdx. For the full-wave rectified sine f(x)=|sin⁡x|, the coefficient a2 is ______ (to three decimal places).
    Answer and solution

    Answer: -0.424 (±0.002)

    an=2π∫0πsin⁡xcos⁡nxdx=1π∫0π[sin⁡(1+n)x+sin⁡(1−n)x]dx. For even n this gives 1π[21+n+21−n]=−4π(n2−1). For n=2: −43π=−0.424. (Odd n give zero: |sin⁡x| has period π.)

    Why people get this wrong: The ripple of a full-wave rectifier is at twice the mains frequency: there is no n=1 term.

  9. MSQHard

    Which statements about how fast Fourier coefficients fall off are true?
    1. The coefficients of |sin⁡x| fall off like 1/n.
    2. For a square wave they fall off like 1/n.
    3. For a triangle wave they fall off like 1/n2.
    4. For a periodic function with derivatives of every order they fall faster than any power of 1/n.
    Answer and solution

    Answer: B, C, D

    Jumps in f: 1/n (square wave). Jumps in f′ only: 1/n2 (triangle wave, and also |sin⁡x|, which is continuous with corners at multiples of π: an=−4/(π(n2−1))). A smooth periodic function: faster than any power, because integrating by parts can be repeated without limit.

    Why people get this wrong: Thinking |sin⁡x| behaves like a square wave because of the absolute value; it has corners, not jumps.

  10. NATHard

    Parseval's theorem applied to the series of x2 on (−π,π) (with a0/2=π2/3 and an=4(−1)n/n2) gives ∑n≥11/n4. Its value is ______ (to four decimal places).
    Answer and solution

    Answer: 1.0823 (±0.0002)

    Parseval: 1π∫−ππx4dx=a022+∑an2. Left: 2π45. Right: (2π2/3)22+16∑1n4=2π49+16∑1n4. So 16∑1n4=2π4(15−19)=8π445 and ∑1n4=π490=1.0823.

    Why people get this wrong: Writing a02/2 with a0=π2/3 (that is a0/2); a0=2π2/3.

  11. MCQHard

    As more terms are kept in the Fourier series of a function with a jump, the Gibbs overshoot
    1. keeps about the same height but moves closer to the jump and gets narrower
    2. shrinks steadily to zero height
    3. grows without limit
    4. moves away from the jump to the middle of the interval
    Answer and solution

    Answer: A

    The peak height tends to a fixed fraction (about 9%) of the jump, while its distance from the jump and its width shrink in proportion to 1/N. So the series still converges at every fixed point, but not uniformly near the jump.

    Why people get this wrong: Thinking convergence at every point means the maximum error goes to zero; near a jump it does not.

  12. NATHard

    An undamped oscillator x¨+4x=F(t) is driven by the square wave F=4π(sin⁡t+sin⁡3t3+…). In the steady state, the amplitude of the response at frequency 3 divided by the amplitude at frequency 1 is ______ (to two decimal places).
    Answer and solution

    Answer: 0.2 (±0.005)

    Each harmonic Fnsin⁡nt gives a response Fn4−n2sin⁡nt. For n=1: 4/π3. For n=3: 4/(3π)|4−9|=415π. Ratio: 4/(15π)4/(3π)=15=0.20.

    Why people get this wrong: Comparing the forcing amplitudes only (1/3), forgetting that each harmonic sees a different response factor 1/|ω02−n2|.

  13. MCQHard

    The half-range sine series of f(x)=x(π−x) on (0,π) has coefficients
    1. bn=8/(πn2) for odd n, 0 for even n
    2. bn=8/(πn3) for odd n, 0 for even n
    3. bn=4/(πn2) for odd n
    4. bn=2/n3 for every n
    Answer and solution

    Answer: B

    bn=2π∫0πx(π−x)sin⁡nxdx. Two integrations by parts give ∫0πx(π−x)sin⁡nxdx=2(1−(−1)n)n3, so bn=8πn3 for odd n and 0 for even n. The odd extension is smooth enough (continuous with a continuous slope) to give 1/n3.

    Why people get this wrong: Expecting 1/n2; this extension has no corners, only jumps in the second derivative.

  14. MCQHard

    Given ∑n=1∞1n4=π490, the sum over odd n only, 1+134+154+…, is
    1. π490
    2. π448
    3. π496
    4. π28
    Answer and solution

    Answer: C

    The even terms are ∑1(2m)4=116∑1m4. So the odd terms are (1−116)π490=15π41440=π496. (It also follows from Parseval applied to the triangle wave.)

    Why people get this wrong: Taking the even part as half the total gives π4/180-type errors; even terms are a sixteenth, not a half.

  15. MCQHard

    f(x)=ex on (−π,π) is extended with period 2π and written as ∑cneinx. The coefficient cn is
    1. sinh⁡ππ(1+in)
    2. (−1)nsinh⁡π2π(1−in)
    3. (−1)ncosh⁡ππ(1−in)
    4. (−1)nsinh⁡ππ(1−in)
    Answer and solution

    Answer: D

    cn=12π∫−ππe(1−in)xdx=12π⋅e(1−in)π−e−(1−in)π1−in. Since e∓inπ=(−1)n, the numerator is (−1)n(eπ−e−π)=2(−1)nsinh⁡π. So cn=(−1)nsinh⁡ππ(1−in). Check n=0: c0=sinh⁡ππ, the mean of ex.

    Why people get this wrong: Losing the factor 2 of sinh⁡, or the alternating sign (−1)n from e∓inπ.

  16. MCQHard

    For f(x)=|x| on (−π,π), extended with period 2π, which statement is true?
    1. Its Fourier coefficients fall off like 1/n
    2. Its series converges to π/2 at x=π
    3. Its series contains sine terms
    4. Differentiating its series term by term gives the Fourier series of the square wave sgn⁡x
    Answer and solution

    Answer: D

    |x|=π2−4π∑n oddcos⁡nxn2. The extension is continuous (a triangle wave), so the coefficients fall like 1/n2 and the series converges to f(π)=π at x=π. It is even: no sines. Differentiating gives 4π∑oddsin⁡nxn, which is the square wave, the derivative of |x|.

    Why people get this wrong: π/2 is the mean value, not the value at the corner x=π, where the function is continuous.

  17. MCQHard

    A 2π-periodic function is odd, f(−x)=−f(x), and also symmetric about x=π/2, f(π−x)=f(x). Its Fourier series can contain only
    1. sin⁡nx with n even
    2. sin⁡nx with n odd
    3. cos⁡nx with n odd
    4. cos⁡nx with n even
    Answer and solution

    Answer: B

    Oddness removes the constant and all cosines. For the sines, sin⁡n(π−x)=sin⁡nπcos⁡nx−cos⁡nπsin⁡nx=(−1)n+1sin⁡nx. The condition f(π−x)=f(x) keeps only terms with (−1)n+1=1, that is n odd. The square wave and sin3⁡x are examples.

    Why people get this wrong: sin⁡2x is odd but is anti-symmetric about π/2, so even n drop out.

  18. MSQHard

    A 2π-periodic function f has Fourier coefficients an, bn. Let g(x)=f(x)+f(x+π). Which statements are true?
    1. g has period π
    2. The series of g has only even harmonics (n=0,2,4,…)
    3. The constant term of g is twice that of f
    4. The odd harmonics of g are twice those of f
    Answer and solution

    Answer: A, B, C

    g(x+π)=f(x+π)+f(x+2π)=g(x), so the period is π. Term by term, cos⁡n(x+π)=(−1)ncos⁡nx and likewise for sines, so g has coefficients (1+(−1)n)an and (1+(−1)n)bn: doubled for even n (including n=0), zero for odd n.

    Why people get this wrong: Shifting by half a period flips the sign of every odd harmonic, so they cancel in the sum.

  19. NATHard

    The function f(x)=x on (−π,π) is approximated by a single term bsin⁡x, with b chosen to make the mean-square error 12π∫−ππ(f−bsin⁡x)2dx as small as possible. The smallest mean-square error is ______ (to three decimal places).
    Answer and solution

    Answer: 1.29 (±0.003)

    The best b is the Fourier coefficient: b1=1π∫−ππxsin⁡xdx=2ππ=2. The error is the mean of f2 minus the mean power in the kept term: π23−b122=3.290−2=1.290. (Directly: the mean of x2 is π2/3, the mean of −2bxsin⁡x is −2b⋅1=−4, and the mean of b2sin2⁡x is b2/2=2; total π2/3−2.) Keeping more harmonics shrinks the error towards zero, by Parseval.

    Why people get this wrong: The mean of sin2⁡x is 12, so a term of amplitude 2 carries mean square 2, not 4.

  20. NATTop rank

    Using x(π−x)=8π∑n oddsin⁡nxn3 on (0,π), the sum 1−133+153−173+… equals ______ (to three decimal places).
    Answer and solution

    Answer: 0.969 (±0.002)

    Put x=π/2: π24=8π(1−127+1125−…) because sin⁡(nπ/2) alternates +1,−1 for odd n. So the sum is π332=0.969.

    Why people get this wrong: Dropping the alternating sign from sin⁡(nπ/2).

  21. MSQTop rank

    For the Fourier series of a piecewise smooth function with a jump, which statements are true?
    1. The height of the overshoot does not go to zero as more terms are added.
    2. The series still converges to f(x) at every point where f is continuous.
    3. The overshoot is about 9% of the full jump.
    4. A similar overshoot appears for a continuous, piecewise smooth function such as |x|.
    Answer and solution

    Answer: A, B, C

    The Gibbs overshoot settles at about 0.0895× the jump, so the maximum error never falls below that: the convergence is not uniform. But at any fixed point of continuity the partial sums do converge to f (the overshoot slides towards the jump and passes every fixed point). For a continuous piecewise smooth function the series converges uniformly and there is no Gibbs overshoot.

    Why people get this wrong: Mixing up pointwise convergence (true at points of continuity) with uniform convergence (false near a jump).

  22. NATTop rank

    The full-wave rectified sine has a0/2=2/π and a2k=−4π(4k2−1) (all other coefficients zero). Parseval's theorem then gives ∑k≥11(4k2−1)2. Its value is ______ (to four decimal places).
    Answer and solution

    Answer: 0.1169 (±0.0002)

    Parseval: 1π∫−ππsin2⁡xdx=1=a022+∑a2k2=(4/π)22+16π2∑1(4k2−1)2. So 16π2S=1−8π2 and S=π2−816=0.1169.

    Why people get this wrong: Using a0/2 in place of a0 in the term a02/2.

  23. MCQTop rank

    The series x=2∑(−1)n+1sin⁡(nx)/n holds on (−π,π). Which operation on it is always safe?
    1. Integrating it term by term
    2. Differentiating it term by term
    3. Squaring it term by term
    4. Multiplying every coefficient by n
    Answer and solution

    Answer: A

    Integrating term by term works for the Fourier series of any piecewise continuous function (integration smooths, and the coefficients get an extra 1/n). Differentiating here gives 2∑(−1)n+1cos⁡nx, whose terms do not even go to zero, so it diverges: the periodic sawtooth has jumps, whose derivative is not an ordinary function. Multiplying the coefficients by n is the same mistake. Squaring term by term misses all the cross terms.

    Why people get this wrong: Differentiating a series with jumps; it only works when the periodic extension is continuous and piecewise smooth.

  24. NATTop rank

    An undamped oscillator x¨+10x=F(t) is driven by the square wave F(t)=4π(sin⁡t+sin⁡3t3+sin⁡5t5+…). In the steady response, the ratio of the amplitude at frequency 3 to the amplitude at frequency 1 is ______.
    Answer and solution

    Answer: 3 (±0.01)

    Each harmonic 4nπsin⁡nt gives a response 4nπsin⁡nt10−n2. For n=1: 4π⋅19. For n=3: 43π⋅11. Ratio =1/31/9=3. The third harmonic, though three times weaker in the force, sits close to the natural frequency 10≈3.16 and dominates the motion.

    Why people get this wrong: Assuming the response mirrors the force's 1/n spectrum; near resonance a weak harmonic wins.

  25. NATTop rank

    Apply Parseval's theorem to the pulse train equal to 1 for |x|<π/3 and 0 for π/3<|x|<π (period 2π). It gives ∑n=1∞sin2⁡(nπ/3)n2= ______ (to three decimal places).
    Answer and solution

    Answer: 1.097 (±0.002)

    With a0/2=1/3 and an=2sin⁡(nπ/3)nπ, Parseval reads 12π∫f2dx=(a02)2+12∑an2. The left side is the fraction of time the pulse is on, 13. So 13−19=2π2∑sin2⁡(nπ/3)n2, giving ∑=29⋅π22=π29=1.097.

    Why people get this wrong: ∫f2 is the on-time; forgetting to subtract the square of the constant term gives π2/6.

← All topics · Timed mock test →