Home / Courses / GATE Physics / Practice / Probability and statistics for physicists

Probability and statistics for physicists

Mathematical methods · 82 questions and templates for GATE Physics, in order from warm-up to top rank. Check each answer, then read the full solution.

New to this topic? Read the lesson first: Probability and statistics for physicists.

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. MCQWarm-up

    Two fair dice are thrown. The probability that the numbers add up to 7 is
    1. 112
    2. 736
    3. 16
    4. 111
    Answer and solution

    Answer: C

    There are 36 equally likely ordered outcomes. Sum 7 comes from (1,6), (2,5), (3,4), (4,3), (5,2), (6,1): 6 outcomes, so 636=16. It is the most likely sum.

    Why people get this wrong: 111 treats the 11 possible sums as equally likely, which they are not.

  2. MCQWarm-up

    A Poisson distribution has mean λ. Its variance is
    1. λ2
    2. λ
    3. λ(1−λ)
    4. λ
    Answer and solution

    Answer: D

    For P(k)=λke−λ/k!: ⟨k⟩=λ and ⟨k(k−1)⟩=λ2, so σ2=λ2+λ−λ2=λ. The standard deviation is λ: that is the N rule of counting experiments.

    Why people get this wrong: λ is the standard deviation, not the variance.

  3. MCQWarm-up

    In n independent trials, each a success with probability p, the variance of the number of successes is
    1. np(1−p)
    2. np
    3. np(1−p)
    4. n2p(1−p)
    Answer and solution

    Answer: A

    Each trial is a 0/1 variable with mean p and variance p−p2=p(1−p). Variances of independent variables add: np(1−p).

    Why people get this wrong: np is the mean.

  4. NATWarm-up

    A fair coin is tossed 4 times. The probability of exactly 2 heads is ______ (to three decimal places).
    Answer and solution

    Answer: 0.375 (±0.001)

    (42)(12)4=616=0.375.

    Why people get this wrong: Answering 12 because "half the tosses should be heads".

  5. MCQWarm-up

    For a Gaussian distribution, the fraction of values within one standard deviation of the mean is about
    1. 50%
    2. 95%
    3. 68%
    4. 99.7%
    Answer and solution

    Answer: C

    P(|x−μ|<σ)=erf(1/2)≈0.683. Within 2σ: 95.4%; within 3σ: 99.7%.

    Why people get this wrong: 95% is the two-sigma figure.

  6. MSQWarm-up

    Events A and B are independent, with 0<P(A),P(B)<1. Which statements are true?
    1. P(A∩B)=P(A)P(B)
    2. P(A|B)=P(A)
    3. A and B are mutually exclusive
    4. P(A∪B)=P(A)+P(B)−P(A)P(B)
    Answer and solution

    Answer: A, B, D

    Independence means P(A∩B)=P(A)P(B), equivalently P(A|B)=P(A). The union rule P(A)+P(B)−P(A∩B) then gives the last option. Mutually exclusive events have P(A∩B)=0≠P(A)P(B): if one happens the other cannot, which is the opposite of independence.

    Why people get this wrong: Exclusive and independent are different, almost opposite, ideas.

  7. MCQWarm-up

    A fair die is rolled twice. The probability of getting at least one six is
    1. 13
    2. 1136
    3. 16
    4. 2536
    Answer and solution

    Answer: B

    Use the complement: the chance of no six in both rolls is (56)2=2536, so at least one six has probability 1−2536=1136.

    Why people get this wrong: Adding 16+16=13 counts the double six twice.

  8. NATStandard

    A detector records 400 counts in a fixed time. The fractional uncertainty of this count is ______ %.
    Answer and solution

    Answer: 5 % (±0.01)

    Counts are Poisson, so σ=N=20 and σ/N=20/400=5%. To halve it you need four times as many counts.

    Why people get this wrong: Using 1/N instead of 1/N.

  9. MCQStandard

    N independent measurements each have standard deviation σ. The standard deviation of their mean is
    1. σ
    2. σ/N
    3. σ/N
    4. σN
    Answer and solution

    Answer: C

    Var(x‾)=Var(1N∑xi)=1N2Nσ2=σ2N, so the standard error is σ/N.

    Why people get this wrong: σ/N would mean the errors cancel completely; independent errors cancel only partly, which gives N.

  10. MCQStandard

    A variable is uniformly distributed on [0,6]. Its variance is
    1. 3
    2. 6
    3. 12
    4. 3
    Answer and solution

    Answer: A

    For uniform on [a,b]: σ2=(b−a)212=3612=3. (A digital meter that rounds to the last digit has a reading error of step/12 for this reason.)

    Why people get this wrong: Using (b−a)2/4 or the half-width as the standard deviation.

  11. MCQStandard

    The full width at half maximum of a Gaussian of standard deviation σ is
    1. 2σ
    2. σ2ln⁡2
    3. 2σ
    4. 22ln⁡2σ≈2.355σ
    Answer and solution

    Answer: D

    e−x2/2σ2=12 at x=±σ2ln⁡2, so FWHM =22ln⁡2σ≈2.355σ. Spectral line widths are often quoted this way.

    Why people get this wrong: σ2ln⁡2 is the half width.

  12. NATStandard

    A condition affects 1% of a population. A test detects it in 99% of those who have it, and gives a false positive for 5% of those who do not. Someone tests positive. The probability that they have the condition is ______ (to three decimal places).
    Answer and solution

    Answer: 0.167 (±0.002)

    Bayes: P(C|+)=P(+|C)P(C)P(+)=0.99×0.010.99×0.01+0.05×0.99=0.00990.0594≈0.167. Most positives are false because the healthy group is so much larger.

    Why people get this wrong: Answering 0.99, which is P(+|C), not P(C|+).

  13. MSQStandard

    Which statements about the Poisson distribution with mean λ are true?
    1. It is the limit of the binomial for n→∞, p→0 with np=λ fixed
    2. P(0)=e−λ
    3. P(k)=λke−λ/k!
    4. Its variance is λ2
    Answer and solution

    Answer: A, B, C

    (nk)pk(1−p)n−k→λkk!e−λ in that limit (rare events, many chances). P(0)=e−λ follows. The variance is λ, not λ2.

    Why people get this wrong: Mean equals variance for Poisson.

  14. MCQStandard

    A random walker takes 100 steps of 1 m, each equally likely to the left or the right. Its root-mean-square distance from the start is
    1. 100 m
    2. 50 m
    3. 0 m
    4. 10 m
    Answer and solution

    Answer: D

    Steps are independent with mean 0 and variance 1 m², so ⟨x2⟩=N=100 m² and xrms=100=10 m. The mean displacement is 0, but the rms is not. This N growth is diffusion.

    Why people get this wrong: The mean displacement (0) is not the rms displacement.

  15. MCQStandard

    For f=xy with independent uncertainties σx and σy, the relative uncertainty of f is
    1. σxx+σyy
    2. (σxx)2+(σyy)2
    3. σxσyxy
    4. σx2+σy2
    Answer and solution

    Answer: B

    σf2=(∂xf)2σx2+(∂yf)2σy2=y2σx2+x2σy2. Dividing by f2=x2y2: relative errors add in quadrature. The straight sum is a worst-case bound, used only for correlated or systematic errors.

    Why people get this wrong: Adding linearly overestimates independent errors.

  16. NATStandard

    A nucleus has decay constant λ=0.2 s⁻¹. The probability that it does not decay in the next 5 s is ______ (to three decimal places).
    Answer and solution

    Answer: 0.368 (±0.002)

    The survival probability is e−λt=e−1≈0.368. A nucleus has no memory: the result is the same whatever its age.

    Why people get this wrong: Using 1−λt=0, a linear approximation valid only for λt≪1.

  17. MCQStandard

    Waiting times between counts in a Poisson process of rate λ follow p(t)=λe−λt. The mean waiting time is
    1. λ
    2. 1/λ
    3. ln⁡2/λ
    4. 1/λ2
    Answer and solution

    Answer: B

    ⟨t⟩=∫0∞tλe−λtdt=1/λ (the mean life τ). The median is ln⁡2/λ (the half-life), and the variance is 1/λ2.

    Why people get this wrong: ln⁡2/λ is the median, not the mean.

  18. MCQStandard

    A family has two children, each equally likely to be a boy or a girl, independently. Given that at least one is a boy, the probability that both are boys is
    1. 12
    2. 14
    3. 13
    4. 23
    Answer and solution

    Answer: C

    The equally likely cases are BB, BG, GB, GG. "At least one boy" leaves BB, BG, GB, of which one is BB: 13.

    Why people get this wrong: Answering 12 treats the condition as "the elder child is a boy", which is different information.

  19. NATExam level

    Counts follow a Poisson distribution with mean 3. The probability of exactly 2 counts is ______ (to four decimal places).
    Answer and solution

    Answer: 0.224 (±0.0005)

    P(2)=32e−32!=4.5×0.04979=0.2240.

    Why people get this wrong: Forgetting the k! in the denominator.

  20. NATExam level

    For 10 tosses of a fair coin, the standard deviation of the number of heads is ______ (to three decimal places).
    Answer and solution

    Answer: 1.581 (±0.002)

    σ=np(1−p)=10×0.25=2.5=1.581.

    Why people get this wrong: Giving the variance 2.5.

  21. MCQExam level

    For a Gaussian variable, the probability of a value more than 2σ above the mean is about
    1. 5%
    2. 2.3%
    3. 0.3%
    4. 16%
    Answer and solution

    Answer: B

    P(|x−μ|>2σ)≈4.6%, shared equally between the two tails, so one tail holds about 2.3%. (One sigma above: about 16%.)

    Why people get this wrong: 5% counts both tails.

  22. MSQExam level

    Which statements about random variables X and Y are true?
    1. Var(X+Y)=VarX+VarY+2Cov(X,Y)
    2. If X and Y are independent, Cov(X,Y)=0
    3. If Cov(X,Y)=0, then X and Y are independent
    4. The correlation coefficient lies between −1 and 1
    Answer and solution

    Answer: A, B, D

    Expand ⟨(X+Y−μX−μY)2⟩ for the first. Independence gives ⟨XY⟩=⟨X⟩⟨Y⟩. The converse fails: with X symmetric about 0 and Y=X2, Cov=⟨X3⟩=0 though Y is fixed by X. Cauchy–Schwarz bounds the correlation by 1.

    Why people get this wrong: Zero covariance rules out only linear dependence.

  23. NATExam level

    With x=2.00±0.02 and y=4.0±0.08 (independent), the absolute uncertainty of f=x2/y is ______ (to three decimal places).
    Answer and solution

    Answer: 0.028 (±0.001)

    Relative errors: σx/x=1%, σy/y=2%. For f=x2y−1: σff=(2×1%)2+(1×2%)2=2.83%. With f=4/4=1: σf=0.028.

    Why people get this wrong: The power 2 multiplies the relative error of x; forgetting it gives 2.24%.

  24. MCQExam level

    The central limit theorem says that the sum of many independent, identically distributed random variables with finite variance is approximately
    1. Poisson distributed
    2. uniformly distributed
    3. distributed like each individual variable
    4. Gaussian distributed
    Answer and solution

    Answer: D

    Whatever the individual distribution (if its variance is finite), the standardised sum tends to a Gaussian, with mean Nμ and variance Nσ2. That is why measurement errors made of many small causes are Gaussian. (The Cauchy/Lorentzian distribution has no finite variance and is a famous exception.)

    Why people get this wrong: The theorem needs finite variance.

  25. NATExam level

    Two measurements of a length give 10.0±0.1 cm and 10.6±0.2 cm. Their weighted mean is ______ cm (to two decimal places).
    Answer and solution

    Answer: 10.12 cm (±0.005)

    Weights wi=1/σi2: 100 and 25. x‾=100×10.0+25×10.6125=1265125=10.12 cm. The more precise value dominates.

    Why people get this wrong: Weighting by 1/σ instead of 1/σ2 gives 10.2.

  26. MCQExam level

    For the same two measurements (10.0±0.1 cm and 10.6±0.2 cm), the uncertainty of the weighted mean is about
    1. 0.15 cm
    2. 0.09 cm
    3. 0.05 cm
    4. 0.10 cm
    Answer and solution

    Answer: B

    σx‾=(∑1/σi2)−1/2=125−1/2≈0.089 cm: smaller than either uncertainty alone. (The two values differ by 0.6/0.05≈2.7 combined sigmas, so one might also ask whether they really agree.)

    Why people get this wrong: Averaging the two uncertainties.

  27. MCQExam level

    For a Gaussian with mean 0 and σ=3, the fourth moment ⟨x4⟩ is
    1. 81
    2. 243
    3. 27
    4. 162
    Answer and solution

    Answer: B

    Gaussian moments: ⟨x2n⟩=(2n−1)!!σ2n. So ⟨x4⟩=3σ4=3×81=243. The ratio ⟨x4⟩/σ4=3 is the Gaussian kurtosis.

    Why people get this wrong: σ4=81 misses the factor 3.

  28. MCQExam level

    With the source removed, a counter records 100 counts in 100 s. With the source, it records 400 counts in 100 s. The net source rate and its uncertainty are
    1. 3.0±0.10 s⁻¹
    2. 3.0±0.17 s⁻¹
    3. 3.0±0.30 s⁻¹
    4. 3.0±0.22 s⁻¹
    Answer and solution

    Answer: D

    Net counts 400−100=300 in 100 s: 3.0 s⁻¹. The uncertainties of the two counts add in quadrature (even though we subtract): 400+100=22.4 counts, i.e. 0.22 s⁻¹.

    Why people get this wrong: Using 300 or 400−100: variances add on subtraction.

  29. MSQExam level

    Which statements about Stirling's approximation are true?
    1. ln⁡N!≈Nln⁡N−N for large N
    2. N!≈2πNNNe−N
    3. The relative error of the full formula falls roughly like 1/(12N)
    4. ln⁡N!=Nln⁡N exactly once N≥10
    Answer and solution

    Answer: A, B, C

    The full form has relative error about 112N (already 0.8% at N=10). Taking logs gives Nln⁡N−N+12ln⁡(2πN), and the last term is negligible for the huge N of statistical mechanics. No version is exact: at N=10, ln⁡10!=15.10 while 10ln⁡10=23.03.

    Why people get this wrong: Dropping the −N term is a large error, not a small one.

  30. MCQExam level

    In how many distinct ways can 3 identical bosons be placed in 3 single-particle states?
    1. 27
    2. 6
    3. 10
    4. 1
    Answer and solution

    Answer: C

    Identical bosons: only the occupation numbers matter, and any number may share a state. Stars and bars: (N+g−1N)=(53)=10. Distinguishable particles would give 33=27; fermions (33)=1.

    Why people get this wrong: 27 counts the particles as distinguishable.

← All topics · Timed mock test →