NUMERICAL METHODS

Math-01802

300 Multiple Choice Questions with Answer Key

The Islamia University of Bahawalpur — Department of Mathematics

Reference: David Kincaid, Numerical Analysis, 3rd Edition, AMS (2011)

Section Topic Questions

1 Numerical Differentiation & Interpolation Polynomials Q1 – Q60

2 Numerical Integration (Newton-Cotes, Romberg, Gaussian) Q61 – Q120

3 Difference & Differential Equations Q121 – Q180

4 Taylor Series, Euler, Runge-Kutta & Predictor-Corrector Q181 – Q240

5 Shooting Methods, BVPs & Finite Difference Methods Q241 – Q300

Total Questions: 300 Credit Hours: 03 4 Options (A/B/C/D) Answer Key: Last Page

Numerical Methods — Math-01802 | Page 1

Section 1: Numerical Differentiation & Interpolation Polynomials

Q1. Which formula uses function values at x and x+h to approximate the first derivative?
(A) Central difference
(B) Backward difference
(C) Forward difference
(D) Divided difference

Q2. The forward difference formula f'(x) ≈ [f(x+h) - f(x)]/h has a truncation error of order:
(A) O(h²)
(B) O(h)
(C) O(h³)
(D) O(1/h)

Q3. The central difference formula for f'(x) is:

(A) [f(x+h) - f(x)]/h
(B) [f(x) - f(x-h)]/h
(C) [f(x+h) - f(x-h)]/(2h)
(D) [f(x+h) + f(x-h)]/(2h)

Q4. The truncation error of the central difference formula for f'(x) is of order:
(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q5. Newton's forward interpolation formula is used when interpolating near the:
(A) End of the table
(B) Middle of the table
(C) Beginning of the table
(D) Any point in the table

Q6. Newton's backward interpolation formula is used when interpolating near the:
(A) Beginning of the table
(B) End of the table
(C) Middle of the table
(D) Any equally spaced point

Q7. The first forward difference operator ∆ is defined as:

(A) ∆f(x) = f(x) - f(x-h)
(B) ∆f(x) = f(x+h) + f(x)
(C) ∆f(x) = f(x+h) - f(x)
(D) ∆f(x) = f(x+h) × f(x)

Q8. The backward difference operator ∇ is defined as:

(A) ∇f(x) = f(x+h) - f(x)
(B) ∇f(x) = f(x) - f(x-h)
(C) ∇f(x) = f(x+h) - f(x-h)
(D) ∇f(x) = f(x) + f(x-h)

Q9. If f(x) = x², h = 1, and x = 2, the forward difference ∆f(2) is:

(A) 3
(B) 4
(C) 5
(D) 6

Q10. Lagrange interpolation polynomial of degree n passes through exactly:

Numerical Methods — Math-01802 | Page 2

 (A) n points
(B) n-1 points
(C) n+1 points
(D) 2n points

Q11. The Lagrange basis polynomial L_k(x) satisfies:

(A) L_k(x_j) = 1 for all j
(B) L_k(x_k) = 1 and L_k(x_j) = 0 for j ≠ k
(C) L_k(x_k) = 0 and L_k(x_j) = 1 for j ≠ k
(D) L_k(x_j) = 0 for all j

Q12. Divided differences are used in which interpolation method?

(A) Lagrange interpolation
(B) Newton's forward interpolation
(C) Newton's divided difference interpolation
(D) Stirling's interpolation

Q13. The zeroth-order divided difference f[x_0] equals:

(A) 0
(B) 1
(C) f(x_0)
(D) f'(x_0)

Q14. The first-order divided difference f[x_0, x_1] is:

(A) (f(x_1) + f(x_0))/(x_1 + x_0)
(B) (f(x_1) - f(x_0))/(x_1 - x_0)
(C) (f(x_1) × f(x_0))/(x_1 × x_0)
(D) f(x_1) - f(x_0)

Q15. The error term in Lagrange interpolation of degree n involves the (n+1)th derivative and is of order:
(A) O(h^n)
(B) O(h^(n+1))
(C) O(h^(n+2))
(D) O(h^(n-1))

Q16. Which of the following is the second forward difference ∆²f(x)?

(A) f(x+2h) - f(x)
(B) f(x+2h) - 2f(x+h) + f(x)
(C) f(x+2h) + 2f(x+h) - f(x)
(D) f(x+2h) - f(x+h) - f(x)

Q17. The second derivative f''(x) using central differences is approximated by:
(A) [f(x+h) - 2f(x) + f(x-h)]/h
(B) [f(x+h) - 2f(x) + f(x-h)]/h²
(C) [f(x+h) + 2f(x) + f(x-h)]/h²
(D) [f(x+h) - f(x-h)]/(2h²)

Q18. Richardson extrapolation improves the accuracy of numerical differentiation by:

(A) Reducing the step size h to zero
(B) Combining two approximations to eliminate the leading error term
(C) Using higher-order derivatives
(D) Replacing the function with a polynomial

Q19. If D(h) = f'(x) + c■h² + c■h■ + ..., then Richardson extrapolation gives a formula of order:
(A) O(h²)
(B) O(h³)
(C) O(h■)
(D) O(h■)

Numerical Methods — Math-01802 | Page 3

Q20. The phenomenon where decreasing h too much causes increased error in numerical differentiation
is due to:
(A) Truncation error
(B) Round-off error
(C) Interpolation error
(D) Discretization error

Q21. Newton's forward interpolation formula uses which type of differences?

(A) Backward differences
(B) Central differences
(C) Forward differences
(D) Divided differences

Q22. The operator E (shift operator) is related to ∆ by:

(A) E = ∆ - 1
(B) E = 1 - ∆
(C) E = 1 + ∆
(D) E = ∆ + ∆²

Q23. For equally spaced data with step h, the derivative using Newton's backward formula at x = x_n is
based on:
(A) Forward differences at x_n
(B) Backward differences at x_n
(C) Central differences at x_n
(D) Divided differences

Q24. Given f(0)=1, f(1)=2, f(2)=5, the second divided difference f[0,1,2] is:
(A) 1
(B) 2
(C) 3
(D) 4

Q25. Stirling's interpolation formula is a:

(A) Forward difference formula
(B) Backward difference formula
(C) Central difference formula
(D) Divided difference formula

Q26. The backward difference formula f'(x) ≈ [f(x) - f(x-h)]/h has truncation error:
(A) O(h²)
(B) O(h)
(C) O(h³)
(D) O(h■)

Q27. Which differentiation formula is most accurate for the same step size h?
(A) Forward difference (1st order)
(B) Backward difference (1st order)
(C) Central difference (2nd order)
(D) All have equal accuracy

Q28. The error in Newton's interpolation is called the:

(A) Round-off error
(B) Remainder term
(C) Convergence error
(D) Stability error

Q29. If f(x) = sin(x) and h = 0.1, using central difference to estimate f'(0), the result is approximately:

Numerical Methods — Math-01802 | Page 4

 (A) 0.9983
(B) 1.0000
(C) 0.9950
(D) 1.0050

Q30. The Lagrange interpolating polynomial for points (0,1), (1,3), (2,7) has degree:
(A) 1
(B) 2
(C) 3
(D) 4

Q31. Newton's divided difference interpolation reduces to Newton's forward formula when:
(A) Points are randomly spaced
(B) Points are equally spaced
(C) Points are at integers only
(D) Only two points are used

Q32. The condition for uniqueness of the interpolating polynomial of degree ≤ n through n+1 points is:
(A) Points must be equally spaced
(B) All x_i must be distinct
(C) f values must be positive
(D) h must be less than 1

Q33. Which formula approximates f''(x) with error O(h²)?

(A) [f(x+h) - f(x)]/h²
(B) [f(x+h) - 2f(x) + f(x-h)]/h²
(C) [f(x+2h) - f(x)]/h²
(D) [f(x+h) + f(x-h)]/h²

Q34. For Richardson extrapolation on first derivative with step h and h/2, the improved formula is:
(A) [4D(h/2) - D(h)]/3
(B) [D(h/2) + D(h)]/2
(C) [D(h/2) - D(h)]
(D) [2D(h/2) + D(h)]/3

Q35. The n-th order divided difference of a polynomial of degree n is:

(A) Zero
(B) A constant
(C) A linear function
(D) The leading coefficient

Q36. Newton's forward difference formula for differentiation at x = x_0 + ph gives f'(x_0) in terms of:
(A) ∇f at x_n
(B) ∆f at x_0
(C) δf at midpoints
(D) f[x_i, x_j]

Q37. The total error in numerical differentiation is:

(A) Truncation error only
(B) Round-off error only
(C) Sum of truncation and round-off errors
(D) Product of truncation and round-off errors

Q38. Runge's phenomenon refers to:

(A) Instability of Runge-Kutta methods
(B) Oscillations in high-degree polynomial interpolation at equally spaced points
(C) Divergence of iterative methods
(D) Error in Gaussian quadrature

Numerical Methods — Math-01802 | Page 5

Q39. The central difference formula for f'(x) uses function values at:
(A) x only
(B) x and x+h
(C) x+h and x-h
(D) x, x+h, and x+2h

Q40. If f(x) = x³, what is the third divided difference f[0,1,2,3]?

(A) 0
(B) 1
(C) 2
(D) 6

Q41. In Newton's backward interpolation, the variable s is defined as:

(A) s = (x - x_0)/h
(B) s = (x - x_n)/h
(C) s = (x_n - x)/h
(D) s = (x + x_n)/h

Q42. Bessel's interpolation formula is best used for:

(A) Near beginning of table
(B) Near end of table
(C) Near middle of table
(D) Any location

Q43. The divided difference table is symmetric in the sense that:

(A) f[x_i, x_j] = -f[x_j, x_i]
(B) f[x_i, x_j] = f[x_j, x_i]
(C) f[x_i, x_j] = f[x_i] + f[x_j]
(D) f[x_i, x_j] = 0

Q44. A 5-point formula for f'(x) using central differences has truncation error of order:
(A) O(h²)
(B) O(h³)
(C) O(h■)
(D) O(h■)

Q45. The interpolation error |f(x) - P_n(x)| is bounded by a term involving:

(A) max|f(x)|
(B) max|f^(n+1)(ξ)|/(n+1)!
(C) min|f'(ξ)|
(D) max|f^(n)(ξ)|/n!

Q46. Which of the following interpolation methods does NOT require equally spaced data?
(A) Newton's forward formula
(B) Newton's backward formula
(C) Lagrange interpolation
(D) Stirling's formula

Q47. The forward difference table for f(x) = 2x has all second differences equal to:
(A) 2
(B) 1
(C) 0
(D) 4

Q48. For cubic spline interpolation, which condition ensures a unique solution?
(A) Natural boundary conditions
(B) Periodic conditions only

Numerical Methods — Math-01802 | Page 6

 (C) Zero second derivative everywhere
(D) Equal spacing only

Q49. In Richardson extrapolation, combining D(h) and D(h/2) eliminates the error term of order:
(A) O(h■)
(B) O(h²)
(C) O(h³)
(D) O(h)

Q50. The formula f'(x) ≈ [-f(x+2h) + 4f(x+h) - 3f(x)]/(2h) is a:

(A) Central difference formula
(B) Forward difference formula of order O(h²)
(C) Backward difference formula
(D) Richardson extrapolation formula

Q51. The operator δ (central difference) is defined as:

(A) δf(x) = f(x+h) - f(x)
(B) δf(x) = f(x+h/2) - f(x-h/2)
(C) δf(x) = f(x) - f(x-h)
(D) δf(x) = f(x+h) + f(x-h)

Q52. Inverse interpolation means:

(A) Finding x given f(x)
(B) Finding f(x) given x
(C) Differentiating f(x)
(D) Integrating f(x)

Q53. The mean operator µ is defined by:

(A) µf(x) = [f(x+h) - f(x-h)]/2
(B) µf(x) = [f(x+h/2) + f(x-h/2)]/2
(C) µf(x) = f(x+h) × f(x-h)
(D) µf(x) = f(x+h) + f(x-h)

Q54. Newton's general interpolation formula using divided differences is also known as:
(A) Gregory-Newton formula
(B) Gauss forward formula
(C) Newton-Gregory forward formula
(D) Newton's divided difference interpolation

Q55. If all nth forward differences of a polynomial are constant and nonzero, the polynomial has degree:
(A) n-1
(B) n
(C) n+1
(D) 2n

Q56. For f(x) = e^x, h = 0.5, the forward difference approximation of f'(0) gives approximately:
(A) 0.6487
(B) 1.2974
(C) 1.0000
(D) 0.5000

Q57. Which error is associated with replacing an exact derivative by a finite difference quotient?
(A) Round-off error
(B) Truncation error
(C) Absolute error
(D) Relative error

Q58. Newton's forward interpolation polynomial evaluated at x = x_0 gives:

Numerical Methods — Math-01802 | Page 7

 (A) 0
(B) f(x_0)
(C) f'(x_0)
(D) ∆f(x_0)

Q59. The optimal step size h that minimizes total error in numerical differentiation balances:
(A) Truncation error and interpolation error
(B) Round-off error and truncation error
(C) Absolute error and relative error
(D) Forward and backward errors

Q60. Gauss's forward interpolation formula uses differences at:

(A) x_0
(B) The point just after x_0
(C) The point just before x_0
(D) A central point x_0

Numerical Methods — Math-01802 | Page 8

Section 2: Numerical Integration

Q61. The trapezoidal rule approximates the integral by replacing the integrand with a:
(A) Quadratic polynomial
(B) Cubic polynomial
(C) Linear polynomial (trapezoid)
(D) Constant function

Q62. The trapezoidal rule formula for ∫[a,b] f(x)dx with step h = b-a is:
(A) h[f(a) + f(b)]
(B) h[f(a) + f(b)]/2
(C) h[f(a) - f(b)]/2
(D) h[f(a) + 2f(b)]/2

Q63. The error in the trapezoidal rule is of order:

(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q64. Simpson's 1/3 rule uses a __________ polynomial to approximate the integrand.
(A) Linear
(B) Quadratic
(C) Cubic
(D) Quartic

Q65. Simpson's 1/3 rule formula for ∫[a,b] f(x)dx with h = (b-a)/2 is:
(A) h[f(a) + f(b)]/2
(B) h[f(a) + 4f((a+b)/2) + f(b)]/3
(C) h[f(a) + 4f((a+b)/2) + f(b)]/6
(D) h[f(a) + 2f((a+b)/2) + f(b)]/3

Q66. The truncation error of Simpson's 1/3 rule is of order:

(A) O(h²)
(B) O(h³)
(C) O(h■)
(D) O(h■)

Q67. Simpson's 3/8 rule requires the interval to be divided into multiples of:
(A) 2 subintervals
(B) 3 subintervals
(C) 4 subintervals
(D) 5 subintervals

Q68. Simpson's 3/8 rule formula is:

(A) 3h[f(x_0) + 3f(x_1) + 3f(x_2) + f(x_3)]/8
(B) h[f(x_0) + 3f(x_1) + 3f(x_2) + f(x_3)]/8
(C) 3h[f(x_0) + f(x_1) + f(x_2) + f(x_3)]/8
(D) h[f(x_0) + 4f(x_1) + f(x_2)]/3

Q69. The Newton-Cotes formula of degree 1 is the:

(A) Simpson's rule
(B) Trapezoidal rule
(C) Boole's rule
(D) Weddle's rule

Q70. The Newton-Cotes formula of degree 2 is:

Numerical Methods — Math-01802 | Page 9

 (A) Trapezoidal rule
(B) Simpson's 3/8 rule
(C) Simpson's 1/3 rule
(D) Boole's rule

Q71. The composite trapezoidal rule with n subintervals has error of order:
(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q72. Romberg integration combines repeated applications of the trapezoidal rule using:
(A) Simpson's rule
(B) Gaussian quadrature
(C) Richardson extrapolation
(D) Newton-Cotes formulas

Q73. In Romberg's method, R(n,0) refers to:

(A) Simpson's rule estimate
(B) The trapezoidal approximation with 2^n subintervals
(C) Gaussian quadrature with n points
(D) Richardson correction term

Q74. The Romberg table entry R(n,m) is computed from R(n,m-1) and R(n-1,m-1) by:
(A) R(n,m) = [4^m R(n,m-1) - R(n-1,m-1)]/(4^m - 1)
(B) R(n,m) = [R(n,m-1) + R(n-1,m-1)]/2
(C) R(n,m) = R(n,m-1) - R(n-1,m-1)
(D) R(n,m) = R(n,m-1) × R(n-1,m-1)

Q75. Gaussian quadrature with n points integrates polynomials of degree up to ______ exactly:
(A) n
(B) n-1
(C) 2n-1
(D) 2n

Q76. Gauss-Legendre quadrature is defined on the interval:

(A) [0, 1]
(B) [0, ∞)
(C) [-1, 1]
(D) [-∞, ∞)

Q77. For 2-point Gauss-Legendre quadrature, the nodes are at:

(A) ±1
(B) ±1/2
(C) ±1/√3
(D) 0 and 1

Q78. For 2-point Gauss-Legendre quadrature, the weights are:

(A) 1 and 1
(B) 1/2 and 1/2
(C) 1/3 and 2/3
(D) 2/3 and 1/3

Q79. Which of the following methods has the highest accuracy for the same number of function
evaluations?
(A) Trapezoidal rule
(B) Simpson's rule
(C) Gaussian quadrature

Numerical Methods — Math-01802 | Page 10

 (D) Midpoint rule

Q80. The error formula for the composite Simpson's 1/3 rule with n subintervals is proportional to:
(A) h²
(B) h³
(C) h■
(D) h■

Q81. Richardson extrapolation applied to the trapezoidal rule gives:

(A) Boole's rule
(B) Simpson's rule
(C) Gaussian quadrature
(D) Weddle's rule

Q82. Adaptive quadrature subdivides the interval based on:

(A) Equal spacing
(B) Local error estimates
(C) The number of function evaluations
(D) The degree of the polynomial

Q83. The midpoint rule has a truncation error of order:

(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q84. To apply the composite Simpson's rule, the number of subintervals n must be:
(A) Odd
(B) Even
(C) A multiple of 3
(D) A power of 2

Q85. Using the trapezoidal rule to evaluate ∫■¹ x dx with h = 0.5, the result is:
(A) 0.25
(B) 0.5
(C) 0.75
(D) 1.0

Q86. The degree of precision of the trapezoidal rule is:

(A) 0
(B) 1
(C) 2
(D) 3

Q87. The degree of precision of Simpson's 1/3 rule is:

(A) 2
(B) 3
(C) 4
(D) 5

Q88. A Newton-Cotes formula is called closed if:

(A) It uses interior points only
(B) It includes the endpoints of the interval
(C) The weights are all equal
(D) It requires equally spaced points

Q89. Gaussian quadrature selects both ________ and ________ optimally.

(A) Intervals and functions

Numerical Methods — Math-01802 | Page 11

 (B) Nodes and weights
(C) Steps and derivatives
(D) Polynomials and errors

Q90. Using Simpson's 1/3 rule to evaluate ∫■² x² dx with h = 1, the result is:
(A) 2.333
(B) 2.667
(C) 3.000
(D) 2.000

Q91. The error bound for the trapezoidal rule over [a,b] with n subintervals is:
(A) (b-a)h²|f''(ξ)|/12
(B) (b-a)h|f'(ξ)|/2
(C) (b-a)h■|f■(ξ)|/180
(D) (b-a)h³|f'''(ξ)|/12

Q92. The error bound for composite Simpson's rule is proportional to:
(A) h²
(B) h■
(C) h■
(D) h³

Q93. Romberg's method generates estimates in a:

(A) Diagonal table
(B) Triangular table
(C) Square matrix
(D) Linear sequence

Q94. The composite midpoint rule with n subintervals has error of order:
(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q95. Gauss-Chebyshev quadrature is appropriate for integrals with weight function:

(A) 1
(B) e^(-x²)
(C) 1/√(1-x²)
(D) x^α(1-x)^β

Q96. The open Newton-Cotes formula does NOT include:

(A) Interior points
(B) Midpoints
(C) Endpoints
(D) Equally spaced points

Q97. For the 3-point Gauss-Legendre rule, the nodes include:

(A) 0, ±1
(B) 0, ±√(3/5)
(C) ±1/2, 0
(D) 0, ±1/√2

Q98. In Romberg integration, R(1,1) is the result of applying Richardson extrapolation to R(1,0) and
R(0,0), giving accuracy of order:
(A) O(h²)
(B) O(h³)
(C) O(h■)
(D) O(h■)

Numerical Methods — Math-01802 | Page 12

Q99. Which rule gives the exact answer for ∫■■¹ (3x² + 2x + 1) dx?
(A) Trapezoidal rule
(B) Simpson's rule
(C) Midpoint rule
(D) Boole's rule

Q100. The Euler-Maclaurin formula relates numerical integration to:

(A) Derivatives of the integrand
(B) Zeros of Legendre polynomials
(C) Gaussian nodes
(D) Richardson extrapolation only

Q101. To halve the error in the composite trapezoidal rule, the step size h should be:
(A) Halved
(B) Quartered
(C) Doubled
(D) Unchanged

Q102. To halve the error in the composite Simpson's rule, the step size h should be:
(A) Halved
(B) Reduced to h/2^(1/4)
(C) Quartered
(D) Reduced to h/√2

Q103. The 4-point Newton-Cotes formula (Boole's rule) has error of order:
(A) O(h■)
(B) O(h■)
(C) O(h■)
(D) O(h■)

Q104. Which integration method is derived from the trapezoidal rule by repeated halving and
Richardson extrapolation?
(A) Simpson's rule
(B) Gaussian quadrature
(C) Romberg integration
(D) Adaptive quadrature

Q105. The composite trapezoidal rule with n=4 applied to ∫■¹ dx uses h =
(A) 0.5
(B) 0.25
(C) 0.125
(D) 0.1

Q106. Which numerical integration formula integrates odd functions exactly over [-a, a]?
(A) Trapezoidal rule
(B) Gaussian quadrature (symmetric nodes)
(C) Midpoint rule
(D) Both B and C

Q107. The weights in the trapezoidal rule for the composite formula (n subintervals) are:
(A) All equal to h
(B) h for endpoints, 2h for interior
(C) h/2 for endpoints, h for interior
(D) h for all points

Q108. The Gauss-Laguerre quadrature is designed for integrals over:

(A) [-1, 1]

Numerical Methods — Math-01802 | Page 13

 (B) [0, ∞)
(C) (-∞, ∞)
(D) [0, 1]

Q109. In composite Simpson's rule with n=4, h=0.25, how many function evaluations are needed?
(A) 3
(B) 4
(C) 5
(D) 6

Q110. The Weddle's rule is a Newton-Cotes formula of degree:

(A) 4
(B) 5
(C) 6
(D) 7

Q111. For ∫■¹ e^x dx using the 2-point Gauss-Legendre rule, the approximate value is closest to:
(A) 1.7183
(B) 1.7182
(C) 1.7184
(D) 1.7180

Q112. Richardson extrapolation can be applied to which integration methods?

(A) Only to the trapezoidal rule
(B) Only to Gaussian quadrature
(C) Any method with known error expansion in h
(D) Only to Simpson's rule

Q113. The composite Simpson's rule weights for n=4 subintervals follow the pattern:
(A) 1,1,1,1,1
(B) 1,2,1
(C) 1,4,2,4,1
(D) 1,2,2,2,1

Q114. For ∫■¹ x³ dx, which rule gives the exact answer with just 3 points?
(A) Trapezoidal rule
(B) Simpson's 1/3 rule
(C) Midpoint rule
(D) Trapezoidal with Richardson

Q115. The degree of precision of Gaussian quadrature with n nodes is:

(A) n
(B) n+1
(C) 2n-1
(D) 2n

Q116. Adaptive quadrature is preferred over fixed-step methods when:

(A) The integrand is a polynomial
(B) The integrand varies rapidly in some regions
(C) High-degree polynomials are used
(D) The interval is very small

Q117. The composite trapezoidal rule formula for n subintervals is:

(A) h[f(x_0) + 2Σf(x_i) + f(x_n)]/2
(B) h[f(x_0) + 4Σf(x_i) + f(x_n)]/3
(C) h[f(x_0) + Σf(x_i) + f(x_n)]
(D) h[f(x_0) + 2Σf(x_i) + f(x_n)]

Numerical Methods — Math-01802 | Page 14

Q118. Error in Gaussian quadrature with n points involves the (2n)th derivative and is zero if:
(A) The function is continuous
(B) The function is a polynomial of degree ≤ 2n-1
(C) The function is analytic
(D) The interval is [-1,1]

Q119. The transformation used to apply Gauss-Legendre quadrature to ∫[a,b] is:

(A) x = (a+b)/2 + (b-a)t/2
(B) x = a + (b-a)t
(C) x = at + b(1-t)
(D) x = a + ht

Q120. Which integration formula has coefficients with negative values for higher-degree Newton-Cotes
rules?
(A) Trapezoidal
(B) Simpson's 1/3
(C) Simpson's 3/8
(D) Newton-Cotes for n ≥ 8

Numerical Methods — Math-01802 | Page 15

Section 3: Difference & Differential Equations

Q121. A difference equation relates values of a function at:

(A) Continuous points
(B) Discrete separated points
(C) Random points
(D) Only two points

Q122. The order of a difference equation is determined by:

(A) The number of terms
(B) The largest difference between indices of f
(C) The degree of the equation
(D) The number of initial conditions

Q123. A homogeneous linear difference equation has:

(A) A non-zero right-hand side
(B) All terms involving the unknown function and zero right-hand side
(C) Constant coefficients only
(D) No solution

Q124. The general solution of a homogeneous linear difference equation includes:

(A) Particular solution only
(B) Complementary solution only
(C) Complementary + particular solutions
(D) Any one solution

Q125. For the difference equation y(n+1) - 2y(n) = 0, the general solution is:
(A) y(n) = C·2^n
(B) y(n) = C·n
(C) y(n) = C·(-2)^n
(D) y(n) = 2C

Q126. The characteristic equation of y(n+2) - 5y(n+1) + 6y(n) = 0 is:

(A) r² + 5r + 6 = 0
(B) r² - 5r + 6 = 0
(C) r² - 5r - 6 = 0
(D) r² + 5r - 6 = 0

Q127. The roots of r² - 5r + 6 = 0 are:

(A) 2, 3
(B) -2, -3
(C) 2, -3
(D) -2, 3

Q128. For the difference equation y(n+2) - 5y(n+1) + 6y(n) = 0 with roots 2 and 3, the general solution is:
(A) y(n) = C■·2^n + C■·3^n
(B) y(n) = C■·2n + C■·3n
(C) y(n) = (C■ + C■n)·2^n
(D) y(n) = C■·2^n - C■·3^n

Q129. When the characteristic equation has a repeated root r of multiplicity 2, the general solution
includes:
(A) C·r^n
(B) (C■ + C■n)·r^n
(C) C■·r^n + C■·r^(n+1)
(D) (C■ + C■/n)·r^n

Numerical Methods — Math-01802 | Page 16

Q130. A particular solution for a non-homogeneous equation y(n+1) - 2y(n) = 3 is:
(A) y_p = 3
(B) y_p = -3
(C) y_p = n
(D) y_p = 3n

Q131. For y(n+1) - 3y(n) = 5·3^n, the form of particular solution to try is:
(A) y_p = A·3^n
(B) y_p = An·3^n
(C) y_p = A·n
(D) y_p = A·5^n

Q132. A system of difference equations involves:

(A) One equation with one unknown
(B) Multiple equations with multiple unknown sequences
(C) Equations with continuous variables
(D) Non-linear equations only

Q133. The method of undetermined coefficients for difference equations applies when the right-hand
side is:
(A) Any function
(B) A polynomial, exponential, or their products
(C) A trigonometric function only
(D) A rational function

Q134. A first-order linear difference equation y(n+1) = ay(n) + b has general solution:
(A) y(n) = Ca^n + b/(1-a), a≠1
(B) y(n) = Ca^n + b
(C) y(n) = Ca + bn
(D) y(n) = C + bn

Q135. The Fibonacci sequence satisfies the difference equation:

(A) y(n) = y(n-1) + y(n-2)
(B) y(n) = 2y(n-1)
(C) y(n) = y(n-1) - y(n-2)
(D) y(n) = y(n-1) × y(n-2)

Q136. The characteristic roots of the Fibonacci equation y(n) = y(n-1) + y(n-2) are:
(A) (1±√5)/2
(B) (1±√3)/2
(C) (1±√2)/2
(D) (±√5)/2

Q137. A difference equation is said to be stable if its solution:

(A) Grows without bound
(B) Oscillates infinitely
(C) Remains bounded as n→∞
(D) Becomes negative

Q138. For stability of a difference equation, all characteristic roots must satisfy:
(A) |r| > 1
(B) |r| < 1
(C) |r| = 1
(D) |r| ≥ 1

Q139. An equation reducible to a linear difference equation can be transformed using:

(A) Integration
(B) A substitution such as z(n) = log(y(n))

Numerical Methods — Math-01802 | Page 17

 (C) Differentiation
(D) Laplace transform

Q140. The Bernoulli difference equation has the form:

(A) y(n+1) = ay(n)
(B) y(n+1) = a(n)y(n) + b(n)[y(n)]^k
(C) y(n+1) + y(n) = 0
(D) y(n+1) = y(n) + c

Q141. For the equation y(n+2) - 4y(n) = 0, the characteristic roots are:
(A) 2, -2
(B) 4, 0
(C) 2, 2
(D) √2, -√2

Q142. Complex characteristic roots r = α ± iβ lead to solutions involving:

(A) Polynomials in n
(B) cos(nθ) and sin(nθ) multiplied by |r|^n
(C) Exponential growth only
(D) Constant sequences

Q143. The Z-transform is to difference equations what the ________ is to differential equations.
(A) Fourier transform
(B) Laplace transform
(C) Taylor series
(D) Wavelet transform

Q144. For y(n+1) - y(n) = 2^n, the particular solution is:

(A) y_p = n·2^n
(B) y_p = 2^n
(C) y_p = 2^(n+1)
(D) y_p = n·2^(n-1)

Q145. A second-order linear difference equation with constant coefficients has:

(A) One independent solution
(B) Two independent solutions
(C) Three independent solutions
(D) Infinitely many solutions

Q146. The general solution of the homogeneous equation y(n+2) + y(n) = 0 is:
(A) y(n) = C■cos(nπ/2) + C■sin(nπ/2)
(B) y(n) = C■·1^n + C■·(-1)^n
(C) y(n) = C■ + C■n
(D) y(n) = C·i^n

Q147. The superposition principle applies to:

(A) Non-linear difference equations
(B) Linear difference equations
(C) Only first-order equations
(D) Equations with variable coefficients only

Q148. A particular solution can be found using the method of:

(A) Undetermined coefficients or variation of parameters
(B) Integration by parts
(C) Taylor series
(D) Newton's method

Q149. If y_h is the homogeneous solution and y_p is a particular solution, the general solution is:

Numerical Methods — Math-01802 | Page 18

 (A) y = y_h × y_p
(B) y = y_h - y_p
(C) y = y_h + y_p
(D) y = y_h / y_p

Q150. The difference equation y(n+1) = ry(n) models:

(A) Arithmetic growth
(B) Geometric (exponential) growth
(C) Quadratic growth
(D) Logarithmic growth

Q151. For the equation y(n) - 5y(n-1) + 6y(n-2) = 0, the solution as n→∞ is dominated by:
(A) 2^n
(B) 3^n
(C) Equal contributions of both
(D) Decays to zero

Q152. A non-homogeneous difference equation with forcing function f(n) = constant c has particular
solution y_p = c/(1-a) if:
(A) a = 0
(B) a ≠ 1
(C) a = 1
(D) a > 1

Q153. An inhomogeneous difference equation y(n+1) - y(n) = n has particular solution:

(A) y_p = n(n-1)/2
(B) y_p = n²/2
(C) y_p = n
(D) y_p = n²

Q154. Linear independence of solutions to a difference equation is verified using the:

(A) Wronskian
(B) Casoratian (Casorati determinant)
(C) Jacobian
(D) Hessian

Q155. For the equation y(n+2) - 2y(n+1) + y(n) = 0, the characteristic root r = 1 has multiplicity:
(A) 1
(B) 2
(C) 3
(D) 0

Q156. With repeated root r = 1 of multiplicity 2, the general solution is:

(A) (C■ + C■n)·1^n = C■ + C■n
(B) C■·1 + C■·n²
(C) C■ + C■/n
(D) C■·n + C■·n²

Q157. A linear difference equation with variable coefficients is:

(A) Always solvable by the characteristic equation method
(B) Generally harder to solve than constant-coefficient equations
(C) Always non-homogeneous
(D) Equivalent to a differential equation

Q158. The equation y(n+1)/y(n) = 2 describes a sequence where each term is:
(A) Increased by 2
(B) Doubled
(C) Squared

Numerical Methods — Math-01802 | Page 19

 (D) Halved

Q159. The general solution for y(n+1) - y(n) = 0 is:

(A) y(n) = 0
(B) y(n) = C (constant)
(C) y(n) = Cn
(D) y(n) = C·n^n

Q160. The method of variation of parameters for difference equations determines:

(A) The characteristic roots
(B) Constants in the complementary function as functions of n
(C) The particular solution directly
(D) The stability of the system

Q161. For a 2×2 system of difference equations, the solution requires finding eigenvalues of the:
(A) Coefficient matrix
(B) Jacobian matrix
(C) Hessian matrix
(D) Identity matrix

Q162. A difference equation approximation of dy/dx = f(x,y) at discrete points x_n = x_0 + nh is:
(A) y(n+1) - y(n) = 0
(B) [y(n+1) - y(n)]/h = f(x_n, y(n))
(C) y(n) = f(x_n)
(D) y(n+1) = y(n) - h·f(x_n,y(n))

Q163. The logistic difference equation y(n+1) = ry(n)(1 - y(n)) is:

(A) Linear
(B) Non-linear
(C) Homogeneous
(D) First-order linear

Q164. The equation y(n) = 2^n satisfies the difference equation:

(A) y(n+1) - y(n) = 0
(B) y(n+1) - 2y(n) = 0
(C) y(n+1) + 2y(n) = 0
(D) 2y(n+1) - y(n) = 0

Q165. For the non-homogeneous equation y(n+1) - 2y(n) = 3n, the particular solution has the form:
(A) y_p = An + B
(B) y_p = An²
(C) y_p = An
(D) y_p = A·3^n

Q166. An equilibrium point y* of y(n+1) = f(y(n)) satisfies:

(A) y* = 0
(B) y* = f(y*)
(C) f(y*) = 0
(D) f'(y*) = 0

Q167. A first-order difference equation with one initial condition y(0) = y■ has:
(A) No solution
(B) Infinitely many solutions
(C) A unique solution
(D) Two solutions

Q168. The difference equation modelling compound interest A(n+1) = (1+r)A(n) has solution:
(A) A(n) = A(0) + rn

Numerical Methods — Math-01802 | Page 20

 (B) A(n) = A(0)(1+r)^n
(C) A(n) = A(0)·r^n
(D) A(n) = A(0)·e^(rn)

Q169. The particular solution for y(n+1) - 3y(n) = 2·3^n is found using a trial solution:
(A) y_p = A·3^n
(B) y_p = An·3^n
(C) y_p = A·2^n
(D) y_p = A·n

Q170. For the system [x(n+1), y(n+1)] = A[x(n), y(n)], stability depends on eigenvalues of A satisfying:
(A) |λ| > 1 for all eigenvalues
(B) |λ| < 1 for all eigenvalues
(C) |λ| = 1 for all eigenvalues
(D) λ > 0 for all eigenvalues

Q171. The equation ∆²y(n) = 0 implies that y(n) is:

(A) A constant
(B) A linear function of n
(C) A quadratic function of n
(D) An exponential function of n

Q172. ∆³y(n) = 0 means y(n) is at most a ________ in n.

(A) Constant
(B) Linear function
(C) Quadratic function
(D) Cubic function

Q173. The equation y(n+2) - y(n+1) - 6y(n) = 0 has characteristic roots:

(A) 3, -2
(B) 3, 2
(C) -3, 2
(D) -3, -2

Q174. A non-linear difference equation z(n+1) = [z(n)]² can be linearized using substitution:
(A) y(n) = 1/z(n)
(B) y(n) = log z(n)
(C) y(n) = z(n) + 1
(D) y(n) = z(n) - 1

Q175. A difference equation of order k requires ________ initial conditions for a unique solution.
(A) k-1
(B) k
(C) k+1
(D) 2k

Q176. The general term of arithmetic progression a, a+d, a+2d,... satisfies the difference equation:
(A) y(n+1) - y(n) = d
(B) y(n+1) - y(n) = 0
(C) y(n+1) = d·y(n)
(D) y(n+1) = y(n) + n

Q177. For the equation y(n) = A·2^n + B·(-1)^n, two independent solutions are:
(A) 2^n and 1
(B) 2^n and (-1)^n
(C) n·2^n and (-1)^n
(D) 2^n and n

Numerical Methods — Math-01802 | Page 21

Q178. The equation y(n+2) + 4y(n) = 0 has characteristic roots:
(A) ±2
(B) ±2i
(C) 2, -2
(D) ±√2

Q179. The general solution for complex roots 2i and -2i is:
(A) y(n) = C■·(2i)^n + C■·(-2i)^n
(B) y(n) = 2^n [C■cos(nπ/2) + C■sin(nπ/2)]
(C) y(n) = C■cos(2n) + C■sin(2n)
(D) y(n) = C■·2^n + C■·(-2)^n

Q180. A difference equation is autonomous if:

(A) Its coefficients are functions of n
(B) Its coefficients do not depend on n
(C) It has no forcing function
(D) It has only one solution

Numerical Methods — Math-01802 | Page 22

Section 4: Taylor Series, Euler, Runge-Kutta & Predictor-Corrector

Q181. The Taylor series method for y' = f(x,y), y(x_0) = y_0 approximates y(x_0+h) using:
(A) Only f(x_0, y_0)
(B) f and its derivatives evaluated at (x_0, y_0)
(C) The trapezoidal rule
(D) Euler's method twice

Q182. Euler's method for y' = f(x,y) is:

(A) y_{n+1} = y_n + h·f(x_{n+1}, y_{n+1})
(B) y_{n+1} = y_n + h·f(x_n, y_n)
(C) y_{n+1} = y_n - h·f(x_n, y_n)
(D) y_{n+1} = y_n + h²·f(x_n, y_n)

Q183. Euler's method has a local truncation error of order:

(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q184. The global truncation error of Euler's method is:

(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q185. Modified Euler's method (Heun's method) is a ________ method.

(A) First-order
(B) Second-order
(C) Third-order
(D) Fourth-order

Q186. In Heun's method, the predictor step uses ________ and the corrector uses ________.
(A) Backward Euler; forward Euler
(B) Euler's method; the average of slopes at two endpoints
(C) Midpoint rule; trapezoidal rule
(D) RK2; RK4

Q187. The classical 4th-order Runge-Kutta (RK4) method computes:

(A) Two slope estimates
(B) Three slope estimates
(C) Four slope estimates
(D) Five slope estimates

Q188. The RK4 update formula is y_{n+1} = y_n + (h/6)(k■ + 2k■ + 2k■ + k■) where k■ is:
(A) f(x_n + h, y_n + hk■)
(B) f(x_n, y_n)
(C) f(x_n + h/2, y_n + hk■/2)
(D) f(x_n + h/2, y_n + hk■/2)

Q189. The local truncation error of RK4 is:

(A) O(h²)
(B) O(h³)
(C) O(h■)
(D) O(h■)

Q190. The global error of RK4 is:

Numerical Methods — Math-01802 | Page 23

 (A) O(h³)
(B) O(h■)
(C) O(h■)
(D) O(h²)

Q191. A predictor-corrector method uses:

(A) An explicit formula to predict, then an implicit formula to correct
(B) Two explicit formulas
(C) Two implicit formulas
(D) Only the corrector step

Q192. The Adams-Bashforth 2-step predictor formula is:

(A) y_{n+1}^P = y_n + h[3f_n - f_{n-1}]/2
(B) y_{n+1}^P = y_n + h·f_n
(C) y_{n+1}^P = y_n + h[f_n + f_{n-1}]/2
(D) y_{n+1}^P = y_{n-1} + 2h·f_n

Q193. The Adams-Moulton corrector is an ________ method.

(A) Explicit
(B) Implicit
(C) Semi-implicit
(D) Extrapolation

Q194. Adams-Moulton 2-step corrector formula is:

(A) y_{n+1} = y_n + h[f_{n+1} + f_n]/2
(B) y_{n+1} = y_n + h·f_{n+1}
(C) y_{n+1} = y_n + h[3f_{n+1} - f_n]/2
(D) y_{n+1} = y_{n-1} + 2h·f_n

Q195. A multistep method requires ________ starting values.

(A) One
(B) As many as the number of steps
(C) Two
(D) None

Q196. Which of the following is a single-step method?

(A) Adams-Bashforth
(B) Adams-Moulton
(C) Runge-Kutta
(D) Milne's method

Q197. The Taylor method of order p has local truncation error:

(A) O(h^p)
(B) O(h^(p+1))
(C) O(h^(p-1))
(D) O(h^(p+2))

Q198. Using Euler's method for y' = y, y(0) = 1 with h = 0.1, y(0.1) ≈
(A) 1.1
(B) 1.01
(C) 1.105
(D) 1.2

Q199. The midpoint method (RK2) for y' = f(x,y) is:

(A) y_{n+1} = y_n + h·f(x_n + h/2, y_n + (h/2)f(x_n,y_n))
(B) y_{n+1} = y_n + h·f(x_n, y_n)
(C) y_{n+1} = y_n + (h/2)[f(x_n,y_n) + f(x_{n+1},y_{n+1})]
(D) y_{n+1} = y_n + h·f(x_n + h, y_n + hf(x_n,y_n))

Numerical Methods — Math-01802 | Page 24

Q200. For the test equation y' = λy with Re(λ) < 0, Euler's method is stable when:
(A) |1 + hλ| ≤ 1
(B) |1 + hλ| ≥ 1
(C) h|λ| < 1
(D) h|λ| > 1

Q201. A stiff differential equation is one where:

(A) The solution changes slowly
(B) The solution has components with very different time scales
(C) The equation has no analytic solution
(D) The step size must be very large

Q202. For stiff problems, which method is preferred?

(A) Explicit Euler
(B) RK4
(C) Implicit methods (e.g., backward Euler)
(D) Adams-Bashforth

Q203. Backward Euler's method is:

(A) y_{n+1} = y_n + h·f(x_n, y_n)
(B) y_{n+1} = y_n + h·f(x_{n+1}, y_{n+1})
(C) y_{n+1} = y_n - h·f(x_n, y_n)
(D) y_{n+1} = y_{n-1} + 2h·f(x_n, y_n)

Q204. The stability region of RK4 is ________ than that of Euler's method.
(A) Smaller
(B) Larger
(C) The same
(D) Depends on the problem

Q205. For y'' = f(x, y, y'), the system can be converted to a first-order system by substituting:
(A) u = y, v = y'
(B) u = y', v = y''
(C) u = y + y'
(D) u = y/y'

Q206. The 4-step Adams-Bashforth method has order:

(A) 2
(B) 3
(C) 4
(D) 5

Q207. Local truncation error is defined as the error introduced in:

(A) The entire computation
(B) One single step assuming previous values are exact
(C) The initial conditions
(D) The stability analysis

Q208. Global truncation error after n steps with step size h is typically:
(A) n × (local truncation error)
(B) (local truncation error)/n
(C) O(h) if local error is O(h²)
(D) O(h²) if local error is O(h³)

Q209. Milne's predictor-corrector method is a ________ step method.

(A) One
(B) Two

Numerical Methods — Math-01802 | Page 25

 (C) Three
(D) Four

Q210. A consistent numerical method for ODEs has local truncation error that:
(A) Grows with n
(B) Approaches zero as h→0
(C) Remains constant
(D) Is always O(h■)

Q211. A numerical method is convergent if:

(A) It is consistent only
(B) It is stable only
(C) It is both consistent and stable
(D) It has small truncation error

Q212. The Dahlquist equivalence theorem states that a method is convergent if and only if it is:
(A) Consistent
(B) Stable
(C) Both consistent and zero-stable
(D) Explicit

Q213. Using RK4 with h = 0.1 for y' = y, y(0) = 1, the value y(0.1) ≈
(A) 1.10517
(B) 1.10000
(C) 1.10500
(D) 1.11000

Q214. The 4th-order Taylor method requires computing up to the ________ derivative of f.
(A) 2nd
(B) 3rd
(C) 4th
(D) 5th

Q215. The main advantage of RK methods over Taylor methods is:

(A) Higher accuracy
(B) No need to compute higher derivatives of f
(C) Larger stability region
(D) Better for stiff equations

Q216. In predictor-corrector methods, how many times is the corrector typically applied?
(A) Once (PECE mode)
(B) Twice
(C) Until convergence
(D) n times for n steps

Q217. The 2nd-order Runge-Kutta method with parameter α = 1 gives Heun's method, with α = 1/2 gives:
(A) Euler's method
(B) Midpoint method
(C) RK4
(D) Backward Euler

Q218. For the system y' = Ay where A has eigenvalues λ_i, RK4 stability requires:
(A) Re(λ_i) < 0 only
(B) All hλ_i lie in the stability region of RK4
(C) h < 1/max|λ_i|
(D) h > max|λ_i|

Q219. The Adams 4-step predictor-corrector (P4EC4E) has order:

Numerical Methods — Math-01802 | Page 26

 (A) 2
(B) 3
(C) 4
(D) 5

Q220. The local error estimate in adaptive step size control for ODE solvers is usually obtained by:
(A) Comparing solutions from two different order methods
(B) Using the exact solution
(C) Differentiating the numerical solution
(D) Applying Richardson extrapolation to differentiation

Q221. Euler's method applied to y' = -100y with h = 0.1 is:

(A) Stable
(B) Unstable
(C) Marginally stable
(D) Always convergent

Q222. The Runge-Kutta-Fehlberg method (RKF45) uses ________ function evaluations per step.
(A) 4
(B) 5
(C) 6
(D) 8

Q223. For a 2nd-order ODE y'' + p(x)y' + q(x)y = r(x), to convert to first-order system, we need:
(A) 1 new variable
(B) 2 new variables
(C) 3 new variables
(D) No new variables

Q224. The leapfrog (midpoint) method y_{n+1} = y_{n-1} + 2h·f(x_n, y_n) is:
(A) A 1-step method
(B) A 2-step explicit method
(C) An implicit method
(D) A predictor-corrector method

Q225. A numerical method has order p if its global error behaves as:
(A) O(h^(p-1))
(B) O(h^p)
(C) O(h^(p+1))
(D) O(1/h^p)

Q226. The trapezoidal method (Crank-Nicolson for ODEs) y_{n+1} = y_n + h[f_n + f_{n+1}]/2 is:
(A) First-order explicit
(B) Second-order implicit
(C) Fourth-order explicit
(D) First-order implicit

Q227. The difference between predictor and corrector values gives an estimate of:
(A) The exact error
(B) The local truncation error
(C) The global error
(D) The stability

Q228. For y' = f(x,y) solved by Euler's method, if h is halved, the global error is approximately:
(A) Doubled
(B) Halved
(C) Quartered
(D) Unchanged

Numerical Methods — Math-01802 | Page 27

Q229. For y' = f(x,y) solved by RK4, if h is halved, the global error is approximately:
(A) Halved
(B) Divided by 16
(C) Divided by 4
(D) Doubled

Q230. The test equation y' = λy (λ < 0 real) is used to analyze:

(A) Consistency
(B) Order of accuracy
(C) Absolute stability
(D) Convergence

Q231. Zero stability of a multistep method refers to:

(A) Stability for the test equation only
(B) Stability as h → 0 for fixed interval
(C) Stability for all step sizes
(D) Stability for stiff equations

Q232. Absolute stability region of a numerical method is the set of hλ values for which:
(A) The method is convergent
(B) The numerical solution does not grow unboundedly
(C) The truncation error is zero
(D) The method is consistent

Q233. An A-stable method has an absolute stability region that includes:

(A) Only negative real axis
(B) The entire left half of the complex hλ-plane
(C) The entire complex plane
(D) Only imaginary axis

Q234. Backward Euler is ________ stable.

(A) A
(B) Conditionally
(C) Not
(D) Marginally

Q235. The Adams-Bashforth 4-step method formula involves function values at:
(A) x_n only
(B) x_n, x_{n-1}
(C) x_n, x_{n-1}, x_{n-2}, x_{n-3}
(D) x_{n+1}, x_n, x_{n-1}

Q236. The characteristic polynomial of a multistep method must satisfy the root condition for:
(A) High accuracy
(B) Zero stability
(C) Absolute stability
(D) Consistency

Q237. The 2nd-order Taylor method for y' = xy, y(0)=1 requires computing y'' = :
(A) y
(B) y + xy'
(C) xy' + y = xy·x + y = x²y + y
(D) xy

Q238. Implicit methods solve a ________ equation at each step.

(A) Linear algebraic
(B) Nonlinear algebraic (in general)

Numerical Methods — Math-01802 | Page 28

 (C) Differential
(D) Integral

Q239. Which starting method is commonly used to provide the required initial values for a multistep
method?
(A) Euler's method
(B) Runge-Kutta method
(C) Taylor series method
(D) All of the above

Q240. The Hamming predictor-corrector method has order:

(A) 2
(B) 3
(C) 4
(D) 5

Numerical Methods — Math-01802 | Page 29

Section 5: Shooting Methods, BVPs & Finite Difference Methods

Q241. A boundary value problem (BVP) for a 2nd-order ODE specifies conditions at:
(A) Only the initial point
(B) Two different points (endpoints of the interval)
(C) Three points
(D) Any interior point

Q242. The shooting method converts a BVP into:

(A) A finite difference problem
(B) An initial value problem (IVP)
(C) A system of algebraic equations
(D) A variational problem

Q243. In the simple shooting method, the unknown is:

(A) The solution at the midpoint
(B) The missing initial condition (slope at x = a)
(C) The boundary value at x = b
(D) The step size h

Q244. The simple shooting method requires solving a ________ equation to find the missing initial
condition.
(A) Algebraic
(B) Differential
(C) Integral
(D) Partial

Q245. For BVP y'' = f(x,y,y'), y(a) = α, y(b) = β, if we guess y'(a) = s, we need to find s such that:
(A) y(a,s) = α
(B) y(b,s) = β
(C) y'(b,s) = 0
(D) y(a,s) = 0

Q246. The secant method or Newton's method is used in shooting to:

(A) Compute the IVP solution
(B) Update the guess for the missing initial condition
(C) Solve the finite difference system
(D) Estimate the truncation error

Q247. Multiple shooting method divides the interval [a,b] into subintervals to:
(A) Reduce the order of the ODE
(B) Improve stability and handle long intervals
(C) Avoid solving IVPs
(D) Apply finite differences

Q248. The finite difference method replaces derivatives with:

(A) Integrals
(B) Difference quotients
(C) Taylor polynomials
(D) Fourier series

Q249. The finite difference approximation of y''(x) using central differences is:
(A) [y(x+h) - y(x-h)]/(2h)
(B) [y(x+h) - 2y(x) + y(x-h)]/h²
(C) [y(x+h) + y(x-h)]/h²
(D) [y(x+h) - y(x-h)]/h²

Numerical Methods — Math-01802 | Page 30

Q250. Applying finite differences to y'' + p(x)y' + q(x)y = r(x) on a grid produces a:
(A) Single equation
(B) System of linear algebraic equations
(C) System of ODEs
(D) Single integral equation

Q251. The finite difference system for a BVP with n interior points results in a ________ system.
(A) n×n tridiagonal
(B) n×n full matrix
(C) (n+2)×(n+2) diagonal
(D) n×(n+2) rectangular

Q252. The explicit finite difference method for the heat equation u_t = c²u_xx uses:
(A) Future time levels only
(B) Current time level to compute next time level directly
(C) Implicit equations at each time step
(D) Crank-Nicolson averaging

Q253. The explicit (Forward Euler in time) finite difference scheme for heat equation is stable when:
(A) r = c²k/h² ≤ 1/2
(B) r = c²k/h² ≤ 1
(C) r = c²k/h² > 1
(D) Any r > 0

Q254. The stability parameter r = α∆t/∆x² for the heat equation must satisfy r ≤ 1/2 for:
(A) Implicit scheme
(B) Crank-Nicolson scheme
(C) Explicit scheme
(D) All schemes

Q255. The implicit finite difference method for the heat equation requires solving at each time step:
(A) An explicit formula
(B) A tridiagonal system
(C) A full matrix system
(D) A single equation

Q256. The Crank-Nicolson method for the heat equation is:

(A) Explicit and 1st-order in time
(B) Implicit, 2nd-order in both time and space
(C) Explicit and 2nd-order in time
(D) Implicit and 1st-order in time

Q257. The Crank-Nicolson method averages the ________ at time levels n and n+1.
(A) Initial conditions
(B) Spatial difference operator
(C) Boundary conditions
(D) Time derivatives

Q258. The Crank-Nicolson scheme is unconditionally:

(A) Unstable
(B) Stable
(C) Convergent but not stable
(D) Divergent

Q259. Von Neumann stability analysis examines how ________ errors are amplified.
(A) Truncation
(B) Boundary
(C) Fourier mode (oscillatory)

Numerical Methods — Math-01802 | Page 31

 (D) Initial

Q260. For the wave equation u_tt = c²u_xx, the explicit scheme is stable when:
(A) c∆t/∆x ≤ 1 (CFL condition)
(B) c∆t/∆x ≥ 1
(C) ∆t < ∆x
(D) c < 1

Q261. The CFL (Courant-Friedrichs-Lewy) condition is a requirement for:

(A) Consistency of the scheme
(B) Stability of explicit schemes for hyperbolic equations
(C) Accuracy of implicit schemes
(D) Convergence of iterative solvers

Q262. For the BVP -y'' = f(x), y(0)=0, y(1)=0, the finite difference matrix is:
(A) A diagonal matrix
(B) A tridiagonal matrix with -2 on diagonal and 1 on off-diagonals
(C) An upper triangular matrix
(D) An identity matrix

Q263. The finite difference method converges to the exact solution as:
(A) h → ∞
(B) h → 0
(C) n → 0
(D) The boundary conditions change

Q264. The local truncation error of the central difference approximation for y'' is:
(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q265. A BVP has a unique solution if the Green's function exists, which requires:
(A) f(x) = 0
(B) The homogeneous problem has only the trivial solution
(C) The boundary conditions are periodic
(D) The interval is short

Q266. For nonlinear BVPs, the shooting method requires ________ to solve the resulting equation.
(A) Direct methods
(B) Root-finding methods (e.g., Newton's, secant)
(C) Analytical methods
(D) Eigenvalue methods

Q267. The finite element method differs from finite differences in that it:
(A) Uses difference quotients
(B) Uses weighted residuals and basis functions
(C) Only applies to linear problems
(D) Requires uniform grids

Q268. The backward difference in time (backward Euler) applied to the heat equation produces the:
(A) Explicit scheme
(B) Fully implicit scheme
(C) Crank-Nicolson scheme
(D) Leapfrog scheme

Q269. For the explicit heat equation scheme, the number of time steps needed to advance to time T with
step k is:

Numerical Methods — Math-01802 | Page 32

 (A) T/h
(B) T/k
(C) T×k
(D) T×h

Q270. Thomas algorithm (LU for tridiagonal) solves a tridiagonal system of n equations in ________
operations.
(A) O(n³)
(B) O(n²)
(C) O(n log n)
(D) O(n)

Q271. A Dirichlet boundary condition specifies:

(A) The derivative of u at the boundary
(B) The value of u at the boundary
(C) A combination of u and its derivative
(D) The flux through the boundary

Q272. A Neumann boundary condition specifies:

(A) The value of u at the boundary
(B) The normal derivative of u at the boundary
(C) The second derivative
(D) Periodic conditions

Q273. For the Laplace equation on a 2D grid, the standard 5-point stencil gives:
(A) u_ij = (u_{i+1,j} + u_{i-1,j} + u_{i,j+1} + u_{i,j-1})/4
(B) u_ij = (u_{i+1,j} + u_{i-1,j})/2
(C) u_ij = u_{i+1,j} + u_{i-1,j} + u_{i,j+1} + u_{i,j-1}
(D) u_ij = u_{i+1,j} - u_{i-1,j}

Q274. In the simple shooting method for a linear BVP, only ________ IVP solutions are needed.
(A) One
(B) Two
(C) Three
(D) Four

Q275. For linear BVP y'' + p(x)y' + q(x)y = r(x), the shooting method solves two IVPs: one for ________
and one for ________
(A) y1 (homogeneous); y2 (particular)
(B) y(a); y(b)
(C) y'; y''
(D) p(x); q(x)

Q276. The consistency condition for finite difference schemes means the truncation error goes to:
(A) Infinity as h → 0
(B) Zero as h → 0
(C) A constant as h → 0
(D) The exact solution as h → 0

Q277. The Lax equivalence theorem states: for a well-posed linear PDE problem, consistency + stability
■
(A) Accuracy
(B) Convergence
(C) Efficiency
(D) Uniqueness

Q278. For the explicit heat equation with ∆x = 0.1 and α = 1, the maximum stable ∆t is:
(A) 0.01

Numerical Methods — Math-01802 | Page 33

 (B) 0.005
(C) 0.05
(D) 0.02

Q279. A mixed (Robin) boundary condition has the form:

(A) u = g
(B) ∂u/∂n = g
(C) αu + β∂u/∂n = g
(D) ∂²u/∂n² = g

Q280. The local truncation error (LTE) of a finite difference scheme is the residual obtained by:
(A) Substituting the exact solution into the scheme
(B) Substituting the numerical solution
(C) Taking the limit h → 0
(D) Differentiating the scheme

Q281. The second-order finite difference scheme for y' using central differences on a non-uniform grid
requires:
(A) Adjusting for different h values on left and right
(B) Uniform step size
(C) More grid points
(D) A higher-order scheme

Q282. The heat equation u_t = u_xx has exact solution u(x,t) = e^(-π²t)sin(πx). The explicit scheme with
r=1/2 and ∆x=0.1 gives errors of order:
(A) O(∆x)
(B) O(∆x² + ∆t)
(C) O(∆x² + ∆t²)
(D) O(∆t²)

Q283. In multiple shooting with m subintervals, the unknowns are:

(A) m+1 values of y
(B) The values of y and y' at each internal shooting point
(C) m values of y'
(D) The step size at each subinterval

Q284. The order of the finite difference scheme for the BVP -y'' = f using central differences is:
(A) O(h)
(B) O(h²)
(C) O(h³)
(D) O(h■)

Q285. For the Poisson equation -∇²u = f on [0,1]×[0,1] with uniform grid (M+1)×(N+1) interior points, the
system matrix has size:
(A) M×N
(B) MN×MN
(C) (M+2)×(N+2)
(D) M²×N²

Q286. The shooting method fails (or becomes unreliable) when:

(A) The BVP has a unique solution
(B) The solution is sensitive to the initial conditions (ill-conditioned)
(C) The ODE is linear
(D) The interval is short

Q287. Finite difference methods for BVPs converge at rate O(h^p) where p is determined by:
(A) The order of the ODE
(B) The order of consistency of the finite difference approximations used

Numerical Methods — Math-01802 | Page 34

 (C) The number of grid points
(D) The boundary condition type

Q288. The implicit scheme for the wave equation using central differences in both space and time is
stable when:
(A) CFL number c·∆t/∆x ≤ 1
(B) For all ∆t, ∆x > 0
(C) Only when ∆t = ∆x
(D) CFL number ≥ 1

Q289. In the finite difference method, increasing the number of grid points (decreasing h) generally:
(A) Increases the error
(B) Decreases the accuracy
(C) Increases accuracy but also computational cost
(D) Has no effect on accuracy

Q290. For a 2nd-order BVP with n interior grid points, the finite difference approach leads to a system
of:
(A) n/2 equations
(B) n equations
(C) 2n equations
(D) n² equations

Q291. Which of the following is an unconditionally stable scheme for the heat equation?
(A) Explicit (Forward Euler)
(B) DuFort-Frankel
(C) Crank-Nicolson
(D) Leapfrog

Q292. The D'Yakonov (ADI) method for 2D heat equation splits the time step into:
(A) Two 1D implicit problems
(B) Four 1D explicit problems
(C) One 2D implicit problem
(D) Two 2D explicit problems

Q293. For the BVP y'' = 6x, y(0) = 0, y(1) = 1, the exact solution is y = x³. A 2nd-order FD scheme with h =
0.5 gives error of order:
(A) O(h) = O(0.5)
(B) O(h²) = O(0.25)
(C) O(h■) = O(0.0625)
(D) Zero (exact)

Q294. The stability of the explicit heat scheme is best described as:
(A) Unconditionally stable
(B) Conditionally stable
(C) Unconditionally unstable
(D) Asymptotically stable

Q295. In the simple shooting method for a nonlinear BVP, Newton's method iteration for the slope s
gives:
(A) s_{k+1} = s_k - F(s_k)/F'(s_k)
(B) s_{k+1} = s_k + F(s_k)
(C) s_{k+1} = s_k × F(s_k)
(D) s_{k+1} = F(s_k)/s_k

Q296. The finite difference formula for the first derivative using backward differences at grid point i is:
(A) (y_{i+1} - y_i)/h
(B) (y_i - y_{i-1})/h

Numerical Methods — Math-01802 | Page 35

 (C) (y_{i+1} - y_{i-1})/(2h)
(D) (y_{i+1} + y_i)/h

Q297. For the heat equation u_t = u_xx, the explicit scheme with r = c²∆t/(∆x)² has amplification factor:
(A) 1 - 4r·sin²(βh/2)
(B) 1 + 4r·sin²(βh/2)
(C) 1/(1 + 4r·sin²(βh/2))
(D) e^(-rβ²h²)

Q298. The two-point boundary value problem y'' - y = 0, y(0) = 1, y(1) = e has exact solution y = e^x. With
finite differences and n=3 interior points, the system is:
(A) 2×2
(B) 3×3
(C) 4×4
(D) 5×5

Q299. The shooting parameter s (initial slope guess) is updated using the secant method as:
(A) s_{k+1} = s_k - F(s_k)(s_k - s_{k-1})/(F(s_k) - F(s_{k-1}))
(B) s_{k+1} = s_k + F(s_k)
(C) s_{k+1} = F(s_k)/s_k
(D) s_{k+1} = s_k - hF(s_k)

Q300. The maximum principle for the discrete Laplacian states that the maximum of the numerical
solution occurs:
(A) At interior grid points
(B) On the boundary
(C) At the center of the domain
(D) At saddle points

Numerical Methods — Math-01802 | Page 36

 ANSWER KEY

Q# Ans Q# Ans Q# Ans Q# Ans Q# Ans

Q1 C Q2 B Q3 C Q4 B Q5 C

Q6 B Q7 C Q8 B Q9 C Q10 C

Q11 B Q12 C Q13 C Q14 B Q15 B

Q16 B Q17 B Q18 B Q19 C Q20 B

Q21 C Q22 C Q23 B Q24 A Q25 C

Q26 B Q27 C Q28 B Q29 A Q30 B

Q31 B Q32 B Q33 B Q34 A Q35 B

Q36 B Q37 C Q38 B Q39 C Q40 B

Q41 B Q42 C Q43 B Q44 C Q45 B

Q46 C Q47 C Q48 A Q49 B Q50 B

Q51 B Q52 A Q53 B Q54 D Q55 B

Q56 B Q57 B Q58 B Q59 B Q60 D

Q61 C Q62 B Q63 B Q64 B Q65 B

Q66 C Q67 B Q68 A Q69 B Q70 C

Q71 B Q72 C Q73 B Q74 A Q75 C

Q76 C Q77 C Q78 A Q79 C Q80 C

Q81 B Q82 B Q83 B Q84 B Q85 B

Q86 B Q87 B Q88 B Q89 B Q90 B

Q91 A Q92 B Q93 B Q94 B Q95 C

Q96 C Q97 B Q98 C Q99 B Q100 A

Q101 A Q102 B Q103 C Q104 C Q105 B

Q106 D Q107 C Q108 B Q109 C Q110 C

Q111 B Q112 C Q113 C Q114 B Q115 C

Q116 B Q117 A Q118 B Q119 A Q120 D

Q121 B Q122 B Q123 B Q124 B Q125 A

Q126 B Q127 A Q128 A Q129 B Q130 B

Q131 B Q132 B Q133 B Q134 A Q135 A

Q136 A Q137 C Q138 B Q139 B Q140 B

Q141 A Q142 B Q143 B Q144 B Q145 B

Q146 A Q147 B Q148 A Q149 C Q150 B

Q151 B Q152 B Q153 A Q154 B Q155 B

Q156 A Q157 B Q158 B Q159 B Q160 B

Q161 A Q162 B Q163 B Q164 B Q165 A

Q166 B Q167 C Q168 B Q169 B Q170 B

Q171 B Q172 C Q173 A Q174 B Q175 B

Q176 A Q177 B Q178 B Q179 B Q180 B

Q181 B Q182 B Q183 B Q184 A Q185 B

Numerical Methods — Math-01802 | Page 37

Q186 B Q187 C Q188 B Q189 D Q190 B

Q191 A Q192 A Q193 B Q194 A Q195 B

Q196 C Q197 B Q198 A Q199 A Q200 A

Q201 B Q202 C Q203 B Q204 B Q205 A

Q206 C Q207 B Q208 C Q209 D Q210 B

Q211 C Q212 C Q213 A Q214 B Q215 B

Q216 A Q217 B Q218 B Q219 C Q220 A

Q221 B Q222 C Q223 A Q224 B Q225 B

Q226 B Q227 B Q228 B Q229 B Q230 C

Q231 B Q232 B Q233 B Q234 A Q235 C

Q236 B Q237 C Q238 B Q239 D Q240 C

Q241 B Q242 B Q243 B Q244 A Q245 B

Q246 B Q247 B Q248 B Q249 B Q250 B

Q251 A Q252 B Q253 A Q254 C Q255 B

Q256 B Q257 B Q258 B Q259 C Q260 A

Q261 B Q262 B Q263 B Q264 B Q265 B

Q266 B Q267 B Q268 B Q269 B Q270 D

Q271 B Q272 B Q273 A Q274 B Q275 A

Q276 B Q277 B Q278 B Q279 C Q280 A

Q281 A Q282 B Q283 B Q284 B Q285 B

Q286 B Q287 B Q288 B Q289 C Q290 B

Q291 C Q292 A Q293 B Q294 B Q295 A

Q296 B Q297 A Q298 B Q299 A Q300 B

Prepared for: Math-01802 Numerical Methods | The Islamia University of Bahawalpur

Numerical Methods — Math-01802 | Page 38