Home / Courses / GATE Physics / Practice / Limits, Taylor series and approximations

Limits, Taylor series and approximations

Mathematical methods · 22 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. MCQStandard

    In statistical physics, the sum Z=∑n=0∞e−nβε (with βε>0) equals
    1. e−βε1−e−βε
    2. 11+e−βε
    3. 11−e−βε
    4. 1eβε+1
    Answer and solution

    Answer: C

    It is a geometric series 1+r+r2+… with ratio r=e−βε<1, so the sum is 1/(1−r)=1/(1−e−βε).

    Why people get this wrong: The first option is the sum starting at n=1 (it misses the n=0 term, which is 1). It equals 1/(eβε−1), the Bose-Einstein occupancy, which is a different quantity.

  2. MCQExam level

    A model potential between two atoms is U(x)=U0[(a/x)12−2(a/x)6], with its minimum at x=a. Expanding U to second order about x=a, the effective spring constant U′′(a) is
    1. 36U0/a2
    2. 156U0/a2
    3. 12U0/a2
    4. 72U0/a2
    Answer and solution

    Answer: D

    U′=U0(−12a12x−13+12a6x−7), which is zero at x=a as expected. U′′=U0(156a12x−14−84a6x−8). At x=a: U′′=U0(156−84)/a2=72U0/a2. Near the minimum U≈−U0+12(72U0/a2)(x−a)2.

    Why people get this wrong: Taking only the repulsive term (156) or forgetting the factor 2 in front of the attractive term.

  3. MSQHard

    Let f(x)=e−1/x2 for x≠0 and f(0)=0. Which statements are true?
    1. Every derivative of f at x=0 is zero.
    2. The Maclaurin series of f converges for every x.
    3. The Maclaurin series of f equals f(x) for x≠0.
    4. f is not differentiable at x=0.
    Answer and solution

    Answer: A, B

    Each derivative of f for x≠0 is (a polynomial in 1/x) ×e−1/x2, and the exponential beats any power of 1/x, so every derivative tends to 0 at the origin; using the definition of the derivative, f(n)(0)=0 for all n. So f is differentiable (infinitely often) at 0. The Maclaurin series is 0+0x+0x2+⋯=0, which converges everywhere, but to 0, while f(x)>0 for x≠0. A function can be smooth without being equal to its Taylor series.

    Why people get this wrong: Believing that a convergent Taylor series always converges to the function. That needs the remainder to go to zero, which fails here.

  4. MCQHard

    The value of ∫0∞xe−xdx is
    1. π
    2. π/4
    3. 1/2
    4. π/2
    Answer and solution

    Answer: D

    It is Γ(3/2). Use Γ(z+1)=zΓ(z) and Γ(1/2)=π: Γ(3/2)=12Γ(1/2)=π/2≈0.886.

    Why people get this wrong: Using Γ(1/2)=π directly: the integrand x1/2e−x is xz−1e−x with z=3/2, not 1/2.

  5. NATHard

    The value of ∫0∞x2e−x2dx is ______ (to three decimal places).
    Answer and solution

    Answer: 0.443 (±0.002)

    Start from I(a)=∫0∞e−ax2dx=12π/a. Differentiate with respect to a: −∫0∞x2e−ax2dx=−14πa−3/2. At a=1 the integral is π/4=0.443.

    Why people get this wrong: Forgetting that the integral runs from 0, not −∞, doubles the answer to π/2.

  6. NATHard

    The sum ∑n=1∞nxn at x=1/3 is ______ (to two decimal places).
    Answer and solution

    Answer: 0.75 (±0.005)

    Differentiate the geometric series ∑n≥0xn=1/(1−x): ∑nxn−1=1/(1−x)2, so ∑nxn=x/(1−x)2. At x=1/3: (1/3)/(4/9)=3/4=0.75. This trick gives the mean energy of a quantum oscillator.

    Why people get this wrong: Writing 1/(1−x)2=2.25 forgets the extra factor x that turns nxn−1 into nxn.

  7. NATHard

    The value of limx→0⁡(sin⁡xx)1/x2 is ______ (to three decimal places).
    Answer and solution

    Answer: 0.846 (±0.002)

    This is a 1∞ form. Take logs: L=lim⁡1x2ln⁡sin⁡xx. Since sin⁡xx=1−x26+O(x4) and ln⁡(1+u)≈u, the log is −x26+O(x4). So L=−16 and the limit is e−1/6≈0.846.

    Why people get this wrong: Saying 1∞=1. The base tends to 1 only as fast as x2, exactly matched by the exponent 1/x2.

  8. MSQHard

    Which of these series converge?
    1. ∑n≥21nln⁡n
    2. ∑n≥21n(ln⁡n)2
    3. ∑n≥1sin⁡1n
    4. ∑n≥1(1−cos⁡1n)
    Answer and solution

    Answer: B, D

    Integral test for the first two: ∫dxxln⁡x=ln⁡ln⁡x→∞ (diverges), while ∫dxx(ln⁡x)2=−1ln⁡x stays finite (converges). For the third, sin⁡(1/n)≈1/n, so by limit comparison with the harmonic series it diverges. For the fourth, 1−cos⁡(1/n)≈12n2, so it converges by comparison with ∑1/n2.

    Why people get this wrong: All four have terms that go to zero; that alone decides nothing. A tiny extra ln⁡n power can switch the verdict.

  9. MCQHard

    The value of ∫0∞e−2x−e−5xxdx is
    1. ln⁡52
    2. ln⁡25
    3. 3
    4. 310
    Answer and solution

    Answer: A

    Define I(a)=∫0∞e−2x−e−axxdx. Differentiating under the integral sign, I′(a)=∫0∞e−axdx=1a. Since I(2)=0, I(a)=ln⁡a2, and I(5)=ln⁡52≈0.916. It is positive, as it must be: e−2x>e−5x for x>0.

    Why people get this wrong: Splitting into ∫e−2x/x−∫e−5x/x gives two divergent pieces; only the difference is finite. The sign check rules out ln⁡25.

  10. NATHard

    Let F(t)=∫tt2sin⁡(tx)xdx for t>0. The value of F′(1) is ______ (to three decimal places).
    Answer and solution

    Answer: 0.841 (±0.002)

    Substitute u=tx (so dx/x=du/u): F(t)=∫t2t3sin⁡uudu, with no t left inside. Then by the fundamental theorem, F′(t)=sin⁡t3t3⋅3t2−sin⁡t2t2⋅2t=3sin⁡t3−2sin⁡t2t. At t=1: 3sin⁡1−2sin⁡1=sin⁡1≈0.841.

    Why people get this wrong: Using only the limit terms of the Leibniz rule and forgetting that t also sits inside the integrand. The substitution removes that worry.

  11. NATHard

    The value of limn→∞⁡(n!)1/nn is ______ (to three decimal places).
    Answer and solution

    Answer: 0.368 (±0.002)

    Take logs: ln⁡(n!)1/nn=1n∑k=1nln⁡k−ln⁡n=1n∑k=1nln⁡kn. This is a Riemann sum for ∫01ln⁡xdx=[xln⁡x−x]01=−1. So the limit is e−1≈0.368. (Stirling's formula n!≈(n/e)n2πn gives the same.)

    Why people get this wrong: Guessing 1 because "n! grows like nn". It grows like (n/e)n, and that e survives.

  12. MSQHard

    About radii of convergence of Maclaurin series, which statements are true?
    1. The series of 11+x2 has radius 1, although the function is smooth for every real x
    2. The series of e−x2 converges for every real x
    3. The series of ln⁡(1+x) converges at x=1
    4. The series of tan⁡x converges for every real x
    Answer and solution

    Answer: A, B, C

    The radius is the distance to the nearest singularity in the complex plane. 11+x2 blows up at x=±i, distance 1, so R=1 even though nothing goes wrong on the real line. e−x2 is entire, so R=∞. ln⁡(1+x) has R=1, and at x=1 the series is 1−12+13−…, which converges (to ln⁡2) by the alternating series test. tan⁡x has poles at ±π/2, so its radius is π/2, not infinite.

    Why people get this wrong: Looking only for real trouble spots: the radius for 1/(1+x2) is set by complex poles.

  13. MCQHard

    The radius of convergence of ∑n≥1n!nnxn is
    1. 1
    2. 1e
    3. ∞
    4. e
    Answer and solution

    Answer: D

    Ratio of successive coefficients: (n+1)!/(n+1)n+1n!/nn=nn(n+1)n=(1+1n)−n→1e. The series converges for |x|/e<1, so R=e. (Stirling agrees: n!/nn≈2πne−n.)

    Why people get this wrong: 1/e is the limit of the coefficient ratio; the radius is its reciprocal.

  14. MSQHard

    For f(x)=x1/x with x>0, which statements are true?
    1. f has its largest value at x=e
    2. f(x)→1 as x→∞
    3. f(x)→0 as x→0+
    4. f increases for all x>0
    Answer and solution

    Answer: A, B, C

    ln⁡f=ln⁡xx, whose derivative 1−ln⁡xx2 is positive for x<e and negative for x>e: the maximum is at x=e, f(e)=e1/e≈1.44, and f falls after it. As x→∞, ln⁡x/x→0, so f→1. As x→0+, ln⁡x/x→−∞, so f→0.

    Why people get this wrong: Unlike xx (which tends to 1 at 0+), x1/x tends to 0 there.

  15. NATTop rank

    The value of limx→0⁡(1+x)1/x−ex is ______ (to three decimal places).
    Answer and solution

    Answer: -1.359 (±0.002)

    Write (1+x)1/x=eln⁡(1+x)/x. Since ln⁡(1+x)=x−x2/2+x3/3−…, the exponent is 1−x/2+x2/3−… So (1+x)1/x=e⋅e−x/2+…≈e(1−x/2). The numerator is −ex/2, and the limit is −e/2=−1.359.

    Why people get this wrong: Treating (1+x)1/x as exactly e near 0 gives 0; the question is about how fast it approaches e.

  16. MSQTop rank

    Which statements are true?
    1. ∑n≥1(−1)n+1n=ln⁡2
    2. The series ∑n≥1(−1)n+1n converges absolutely.
    3. Reordering the terms of ∑n≥1(−1)n+1n can change its sum.
    4. ∑n≥11n(n+1)=1
    Answer and solution

    Answer: A, C, D

    Put x=1 in ln⁡(1+x)=x−x2/2+… (allowed, as the series converges there): the sum is ln⁡2. Absolute convergence would need ∑1/n to converge, which it does not, so the convergence is only conditional. For conditionally convergent series, Riemann showed that reordering can give any sum at all. Finally 1n(n+1)=1n−1n+1, a telescoping sum equal to 1.

    Why people get this wrong: Assuming the order of terms never matters. It does not matter only for absolutely convergent series.

  17. MCQTop rank

    For small ϵ>0, to first order in ϵ, ∫0∞e−x2−ϵx4dx is
    1. π2(1−3ϵ8)
    2. π2(1−3ϵ2)
    3. π2(1−3ϵ4)
    4. π2(1−ϵ2)
    Answer and solution

    Answer: C

    Expand e−ϵx4≈1−ϵx4. Then the integral is ∫0∞e−x2dx−ϵ∫0∞x4e−x2dx. With u=x2: ∫0∞x4e−x2dx=12Γ(52)=12⋅34π=3π8. So the result is π2−3π8ϵ=π2(1−3ϵ4).

    Why people get this wrong: 1−3ϵ/8 forgets to divide the correction by the leading π/2 when factoring it out.

  18. NATTop rank

    Differentiating under the integral sign with respect to a, find ∫01x3−1ln⁡xdx= ______ (to three decimal places).
    Answer and solution

    Answer: 1.386 (±0.002)

    Let I(a)=∫01xa−1ln⁡xdx, so I(0)=0. Since ∂∂axa=xaln⁡x, I′(a)=∫01xadx=1a+1. Then I(a)=ln⁡(1+a) and I(3)=ln⁡4=1.386. The integrand is finite at both ends (→3 as x→1, →0 as x→0).

    Why people get this wrong: Fix the constant with I(0)=0; without it the answer is undetermined.

← All topics · Timed mock test →