1.

The Bisection method requires:

A. Function must be differentiable
B. Function must change sign on the interval
C. Initial guess only
D. Derivative at the root

Answer: B
Explanation:
Bisection works only when f(a)·f(b) < 0, which guarantees a root in the
interval by the Intermediate Value Theorem.

2. If f(a)·f(b) < 0, the interval [a, b] contains:

A. A maximum
B. A minimum
C. At least one root
D. No root

Answer: C
Explanation:
A sign change indicates the function crosses the x-axis at least once.

3. The convergence of Newton–Raphson

method is:
A. Linear
B. Quadratic
C. Cubic
D. Logarithmic

Answer: B
Explanation:
 Newton–Raphson uses tangent-line approximation and has quadratic
convergence for simple roots.

4. The order of convergence of the Bisection

method is:
A. 1
B. 1/2
C. 2
D. 3

Answer: A
Explanation:
Bisection reduces error by half each iteration → linear convergence.

5. Newton–Raphson iteration formula is:

A. xn+1=a+b2x_{n+1} = \frac{a+b}{2}xn+1 =2a+b
B. xn+1=xn−f(xn)x_{n+1} = x_n - f(x_n)xn+1 =xn −f(xn )
C. xn+1=xn−f(xn)f′(xn)x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}xn+1 =xn −f′(xn

D. xn+1=xn+f(xn)x_{n+1} = x_n + f(x_n)xn+1 =xn +f(xn )

)f(xn )

Answer: C
Explanation:
Derived from tangent line: intersection with x-axis gives the formula.

6. Bisection method is also called:

A. Regula falsi method
B. Interval-halving method
C. Secant method
D. Newton method
Answer: B
Explanation:
At every step, the interval length is halved.

7. Newton–Raphson fails when:

A. f is continuous
B. f′(xₙ) = 0
C. f′(xₙ) ≠ 0
D. xₙ is near root

Answer: B
Explanation:
Division by zero occurs in the iteration formula → method breaks.

8. Which method always converges (if a sign

change exists)?
A. Newton method
B. Secant method
C. Fixed-point iteration
D. Bisection method

Answer: D
Explanation:
Bisection is monotonic and guaranteed to converge.

9. Newton–Raphson diverges when:

A. Initial guess is close
B. Derivative is large
C. Initial guess is far
D. Function is smooth
Answer: C
Explanation:
Far initial guesses lead to tangent lines jumping far away → divergence.

10. Bisection reduces the interval after n

iterations to:
A. (b−a)/2n(b-a)/2n(b−a)/2n
B. (b−a)/n(b-a)/n(b−a)/n
C. (b−a)/2n(b-a)/2^n(b−a)/2n
D. 2n(b−a)2^n(b-a)2n(b−a)

Answer: C
Explanation:
Each iteration halves the interval.

11. Newton–Raphson error behaves as:

A. en+1≈ene_{n+1} ≈ e_nen+1 ≈en
B. en+1≈(en)2e_{n+1} ≈ (e_n)^2en+1 ≈(en )2
C. en+1≈ene_{n+1} ≈ \sqrt{e_n}en+1 ≈en
D. None

Answer: B
Explanation:
Quadratic convergence → error squared each iteration.

12. Bisection stops when:

A. f(a) = 0
B. f(b) = 0
C. Interval width < tolerance
D. Derivative is zero
Answer: C
Explanation:
Stopping criterion is based on interval size.

13. Bisection method is:

A. Fast but unreliable
B. Slow but reliable
C. Fast and reliable
D. Unreliable and slow

Answer: B
Explanation:
Guaranteed convergence but linear (slow).

14. Newton method requires evaluation of:

A. Only f(x)
B. Only f′(x)
C. Both f and f′
D. Only second derivative

Answer: C
Explanation:
Formula depends on both function and derivative.

15. Bisection converges in:

A. Finite steps
B. Infinite steps
C. Random steps
D. Exactly 5 steps
Answer: B
Explanation:
Convergence is asymptotic (approaches but never exact unless root
endpoint).

16. Newton–Raphson works best for:

A. Discontinuous functions
B. Non-smooth functions
C. Smooth differentiable functions
D. Piecewise constant functions

Answer: C
Explanation:
Derivative-based; needs smoothness.

17. Secant method approximates derivative

using:
A. Backward difference
B. Forward difference
C. Central difference
D. Two-point finite difference

Answer: D
Explanation:
Slope of secant line substitutes for f′(x).

18. Which method does NOT require

derivatives?
A. Newton
B. Secant
C. Bisection
D. Both B and C

Answer: D
Explanation:
Both methods avoid explicit derivative computation.

19. If f is monotonic on [a,b], Bisection finds:

A. The smallest root
B. The largest root
C. A unique root
D. No root

Answer: C
Explanation:
Monotonic + sign change → one root only.

20. Newton formula is derived from:

A. Secant line
B. Tangent line
C. Linear interpolation
D. Midpoint rule

Answer: B
Explanation:
Newton method uses the tangent line to approximate next root.

21. Bisection method requires:

A. One initial value
B. Two initial values with sign change
C. Three values
D. No initial values

Answer: B
Explanation:
Root bracketing requires two endpoints.

22. Newton–Raphson may oscillate if:

A. Derivative is large
B. Derivative is small or varies
C. Initial guess is poor
D. Both B and C

Answer: D
Explanation:
Wrong initial point or small derivative causes oscillations.

23. In Bisection, approximated root is:

A. a
B. b
C. (a+b)/2
D. (3a+b)/4

Answer: C
Explanation:
Midpoint is chosen each iteration.

24. Newton–Raphson is inefficient if:

A. Derivative is easy
B. Derivative is difficult
C. Function is smooth
D. Initial guess is good

Answer: B
Explanation:
Derivative evaluation can be expensive.

25. Bisection method is preferred when:

A. Need fast convergence
B. Derivative available
C. Guaranteed convergence required
D. Function is nonlinear

Answer: C
Explanation:
Its major advantage is certainty of convergence.

26. Newton method uses how many

evaluations per iteration?
A. Only f
B. Only f′
C. f and f′
D. None

Answer: C
Explanation:
Requires both values each step.
27. After 10 bisection iterations, interval
shrinks by factor:
A. 10
B. 1/10
C. 1/1024
D. 1/2

Answer: C

Shrinking factor is 1/2n1/2^n1/2n.

Explanation:

For n = 10 → 1/10241/10241/1024.

28. Newton–Raphson is unreliable if root is:

A. Simple
B. Multiple
C. Real
D. Positive

Answer: B
Explanation:
Repeated roots → derivative near zero → slow/unstable.

29. If the initial guess equals root exactly,

Newton method:
A. Fails
B. Finishes in one iteration
C. Diverges
D. Takes many steps

Answer: B
Explanation:
Substituting exact root gives x₁ = x₀.
30. Major drawback of Bisection:
A. Needs derivative
B. Slow convergence
C. Not guaranteed
D. Hard to implement

Answer: B
Explanation:
Its convergence is linear, much slower than Newton’s quadratic
convergence.

The bisection method converges for any

continuous function with f(a)·f(b) < 0
because:
A. f is monotonic
B. It uses a contraction mapping
C. It uses the Intermediate Value Theorem
D. The function is differentiable

Answer: C
Explanation:
Bisection relies entirely on the IVT; no monotonicity or differentiability are
required.

2. Newton–Raphson converges quadratically

only if:
A. f(x) has no second derivative
B. f(x) has a simple root
C. f(x) has a multiple root
D. f′(x₀) = 0
Answer: B
Explanation:
Quadratic convergence requires the root to be simple; multiple roots slow it
to linear.

3. If f′(α) = 0 at root α, Newton–Raphson:

A. Diverges
B. Has cubic convergence
C. Becomes linearly convergent
D. Remains quadratic

Answer: C
Explanation:
Multiplicity >1 reduces order from 2 to 1.

4. Bisection method guarantees convergence

because:
A. It is monotonic decreasing
B. It always halves the error
C. It requires a Lipschitz condition
D. It uses derivative information

Answer: B

Error shrinks as (b−a)/2n(b-a)/2^n(b−a)/2n, a deterministic reduction.

Explanation:

5. Newton–Raphson diverges when:

A. f′(x) is small or vanishes
B. f(x) is smooth
C. f has monotonic behavior
D. Initial guess is near the root

Answer: A
Explanation:
Small derivative → tangent nearly horizontal → huge step → divergence.

6. The tangential nature of Newton–Raphson

implies:
A. It depends on global structure of f
B. Convergence is local
C. Convergence is global
D. It finds all roots

Answer: B
Explanation:
NR is locally convergent: requires initial guess close enough.

7. The condition for quadratic convergence of

Newton’s method is:
A. f′(α) ≠ 0
B. f″(α) = 0
C. |f′(α)| > |f″(α)|
D. α is a multiple root

Answer: A
Explanation:
Non-zero derivative ensures a simple root → quadratic convergence.
8. Bisection method cannot detect:
A. Existence of root
B. Multiple roots
C. Midpoint location
D. Interval sign change

Answer: B
Explanation:
Root multiplicity involves derivative behavior → bisection cannot detect it.

9. If Newton–Raphson is applied to f(x)=x³, it

will:
A. Converge quadratically
B. Converge linearly
C. Diverge
D. Converge cubically

Answer: D
Explanation:
For sufficiently smooth functions with only odd powers, convergence can be
super-quadratic.

10. For f(x)=0, the Newton update minimizes:

A. |f(x)|
B. |x−α|
C. Linearized residual
D. Second derivative

Answer: C
Explanation:
NR solves the first-order Taylor linearization exactly → minimizes local linear
residual.
11. The bisection method produces the same
error behavior regardless of:
A. Smoothness of f
B. Number of roots
C. Root multiplicity
D. All of the above

Answer: D
Explanation:
Error reduction is independent of function properties (except continuity).

12. Newton’s method may produce complex

iterates if:
A. f has real coefficients
B. f is polynomial
C. f(x) < 0
D. The tangent line intersects the x-axis in the complex plane

Answer: D
Explanation:
NR naturally extends into the complex plane for polynomials.

13. A major limitation of bisection is:

A. Slow asymptotic rate
B. Requires derivative
C. No bracketing
D. Failure on continuous functions
Answer: A
Explanation:
Linear convergence → extremely slow for tight tolerances.

14. Newton–Raphson iteration corresponds

to:
A. Steepest descent
B. Fixed-point iteration of a transformed map
C. Midpoint method
D. Secant line interpolation

Answer: B

NR is a fixed-point method x=g(x)x=g(x)x=g(x) where g(x)=x−f(x)/f′

Explanation:

(x)g(x)=x-f(x)/f'(x)g(x)=x−f(x)/f′(x).

15. The convergence domain of Newton–

Raphson depends on:
A. Only f
B. Only f′
C. Both f and f′
D. Neither

Answer: C
Explanation:
Both function and derivative shape the basins of attraction.
16. A sufficient condition for Newton
convergence is:
A. f continuous on interval
B. f′ monotonic on interval
C. |f(x)f″(x)| < K (K small)
D. f(a)=f(b)

Answer: C
Explanation:
Kantorovich-type conditions guarantee convergence if curvature is small.

17. A multiple root α of multiplicity m

modifies Newton’s iteration as:
A. xn+1=xn−f(xn)/f′(xn)x_{n+1} = x_n - f(x_n)/f'(x_n)xn+1 =xn −f(xn )/f′

B. xn+1=xn−m f(xn)/f′(xn)x_{n+1} = x_n - m\, f(x_n)/f'(x_n)xn+1 =xn

(xn )

C. xn+1=αx_{n+1} = αxn+1 =α
−mf(xn )/f′(xn )

D. None

Answer: B
Explanation:
Modified Newton restores quadratic convergence for multiple roots.

18. Newton–Raphson is equivalent to one

step of:
A. Euler Explicit
B. Euler Implicit
C. Second-order Taylor approximation
D. Newton’s second law
Answer: C
Explanation:
NR solves the linear Taylor approximation f(x+δ)=0.

19. Bisection method cannot be applied

when:
A. Function continuous
B. f(a)·f(b) > 0
C. Function is differentiable
D. Root is simple

Answer: B
Explanation:
If no sign change, bracketing cannot start.

20. Newton–Raphson exhibits chaotic

behavior when:
A. f′ small
B. Iterates cross an inflection region
C. f has high curvature
D. All of the above

Answer: D
Explanation:
Chaotic behavior is common for nonlinear equations with complex basins.

21. Order of convergence p is defined by:

A. en+1≤Cene_{n+1} \leq Ce_nen+1 ≤Cen
B. en+1=Cenpe_{n+1}=C e_n^pen+1 =Cenp
C. en+1≤pene_{n+1}\leq p e_nen+1 ≤pen
D. None

Answer: B
Explanation:

lim⁡n→∞en+1enp=C\lim_{n→∞} \frac{e_{n+1}}{e_{n}^p} = Climn→∞ enp

General convergence definition:

en+1 =C.

22. Newton iteration applied to f(x)=exp(x)

−1 yields:
A. Quadratic convergence
B. Linear convergence
C. No convergence
D. Infinite oscillation

Answer: A
Explanation:
Simple root at x=0 → quadratic convergence.

23. Newton iteration on periodic functions

often:
A. Always converges
B. Converges linearly
C. Produces cycles or chaos
D. Finds multiple roots simultaneously

Answer: C
Explanation:
Trigonometric functions create multiple attractors → cycles.
24. Bisection convergence rate is
independent of:
A. f′
B. f″
C. Root multiplicity
D. All of the above

Answer: D
Explanation:
Only continuity + sign change matter.

25. Newton’s method can be interpreted

geometrically as:
A. Moving along the steepest gradient
B. Sliding down the curve
C. Intersecting tangent with x-axis
D. Using horizontal projection

Answer: C
Explanation:
Graphically, next iterate is tangent’s root.

26. Bisection is not considered a Newton-

type method because it:
A. Does not use derivatives
B. Does not use Taylor expansion
C. Uses interval-based logic
D. All of the above
Answer: D
Explanation:
Newton-type methods = derivative + Taylor based.

27. Newton–Raphson fails globally but

succeeds locally because:
A. Nonlinear functions have basins of attraction
B. Tangent approximation works only near root
C. Initial guess dominates behavior
D. All of the above

Answer: D
Explanation:
NR is locally accurate but globally unpredictable.

28. In Newton's method, f′(xₙ) close to zero

produces:
A. Small steps
B. Large steps
C. No step
D. Infinite convergence

Answer: B
Explanation:
Small denominator → large update → instability.
29. For functions with flat regions near root,
Newton method:
A. Accelerates
B. Slows dramatically
C. Remains quadratic
D. Always diverges

Answer: B
Explanation:
Flat curvature → small derivative → small numerator → slow convergence.

30. The Newton–Raphson method for

polynomials is deeply connected to:
A. Fourier analysis
B. Holomorphic dynamics
C. Probability theory
D. Optimization on manifolds

Answer: B
Explanation:
Newton iteration on complex plane forms Julia sets and fractal basins.

Simultaneous Linear Algebraic

Equations
(Gauss Elimination • Crout’s Method • Gauss–Seidel •
Relaxation / SOR)

These questions test deep conceptual understanding, convergence

theory, numerical stability, and matrix properties.
✅ 30 Ph.D.-Level MCQs (with
Answers & Explanations)

1. Gauss elimination is numerically unstable

primarily when:
A. Matrix is symmetric
B. Pivot elements are small
C. Matrix is sparse
D. Matrix is diagonal dominant

Answer: B
Explanation:
Small pivots → large multipliers → amplification of round-off → instability.

2. Partial pivoting in Gaussian elimination

ensures:
A. Quadratic convergence
B. Avoiding division by very small pivots
C. Reduced computational cost
D. Exact arithmetic

Answer: B
Explanation:
Pivoting chooses the largest available element to reduce round-off
propagation.
3. Crout’s method decomposes A as:
A. A=ULA = U LA=UL
B. A=LUA = L UA=LU
C. A=LDUA = L D UA=LDU
D. A=PLUA = P L UA=PLU

Answer: B
Explanation:
Crout factorization uses unit diagonal in U and arbitrary diagonal in L.

4. Gauss–Seidel method converges if:

A. A is symmetric
B. A is diagonally dominant
C. A is singular
D. A is normal

Answer: B
Explanation:
Strict or irreducible diagonal dominance ensures convergence.

5. In SOR (Successive Over Relaxation),

convergence is fastest when ω is:
A. 0
B. 1
C. Between 1 and 2
D. >2

Answer: C
Explanation:
Optimal ω lies in (1,2) for over-relaxation.
6. In Gauss–Seidel, the iteration matrix is:
A. D−1(L+U)D^{-1}(L+U)D−1(L+U)
B. −(D+L)−1U-(D+L)^{-1}U−(D+L)−1U
C. −(D−U)−1L-(D-U)^{-1}L−(D−U)−1L
D. None

Answer: B

Splitting: A=D+L+UA = D + L + UA=D+L+U.

Explanation:

GS iteration: x(k+1)=−(D+L)−1Ux(k)+(D+L)−1bx^{(k+1)} = -(D+L)^{-1}U

x^{(k)} + (D+L)^{-1}bx(k+1)=−(D+L)−1Ux(k)+(D+L)−1b.

7. Gauss elimination without pivoting fails

when:
A. Matrix is singular
B. Zero appears on pivot position
C. Matrix is symmetric
D. Matrix is orthogonal

Answer: B
Explanation:
Zero pivot → division impossible.

8. The computational complexity of Gaussian

elimination is:
A. O(n log n)
B. O(n²)
C. O(n³)
D. Exponential

Answer: C
9. In Crout’s method, which element is set to
1?
A. Diagonal of L
B. Diagonal of U
C. Both
D. None

Answer: B
Explanation:
Crout uses unit diagonal in U.

10. Gauss–Seidel converges faster than

Jacobi because:
A. Uses old iterate values
B. Uses new values immediately
C. Uses relaxation factor
D. No diagonal dominance needed

Answer: B
Explanation:
Immediate substitution accelerates convergence.

11. SOR reduces to Gauss–Seidel when ω =

A. 0
B. 0.5
C. 1
D. 2

Answer: C
12. SOR fails to converge when ω ≥
A. 0
B. 1
C. 2
D. ∞

Answer: C

Convergence only for 0<ω<20 < \omega < 20<ω<2.

Explanation:

13. LU decomposition fails if:

A. A is symmetric
B. A is diagonal
C. A is singular
D. A is orthogonal

Answer: C

14. Gauss elimination round-off error is

minimized using:
A. Forward substitution
B. Crout’s method
C. Partial pivoting
D. Iterative refinement

Answer: C
15. Iterative methods (GS, Jacobi, SOR) are
preferred when A is:
A. Dense
B. Large and sparse
C. Symmetric positive definite
D. Orthogonal

Answer: B

16. For SPD matrices, which method is

guaranteed to converge?
A. Jacobi
B. Gauss–Seidel
C. SOR
D. All of the above

Answer: D

17. In Gauss–Seidel, the error reduction

factor is:
A. Spectral radius of iteration matrix
B. Condition number of A
C. Determinant of A
D. Trace of A

Answer: A
Explanation:
Convergence ⇔ ρ(T) < 1.
18. Optimal ω for SOR satisfies:
A. ω=1ω = 1ω=1
B. ω=2ω = 2ω=2
C. Depends on spectral radius of GS
D. Independent of A

Answer: C
Explanation:

ωopt=21+1−ρ(GS)2ω_{opt} = \frac{2}{1+\sqrt{1-\rho(GS)^2}}ωopt
Formula:

=1+1−ρ(GS)2 2

19. In Gaussian elimination, the growth

factor measures:
A. Increase in computational complexity
B. Increase in pivot magnitudes
C. Convergence speed
D. Error in LU Factorization

Answer: B

20. Crout’s method is numerically stable only

if:
A. A is diagonal dominant
B. Pivoting is used
C. Matrix is symmetric
D. Matrix is sparse

Answer: B
Explanation:
Without pivoting → unstable.
21. Gauss–Seidel fails when A is:
A. Diagonal dominant
B. SPD
C. Singular
D. Strictly upper triangular

Answer: C

22. Gauss elimination converts A into:

A. Upper triangular
B. Diagonal
C. Symmetric
D. Orthonormal

Answer: A

23. In iterative methods, the spectral radius

condition is:
A. ρ(T) > 1
B. ρ(T) = 1
C. ρ(T) < 1
D. ρ(T) = 0

Answer: C
24. SOR is an improvement of:
A. LU Decomposition
B. Gauss elimination
C. Gauss–Seidel
D. Jacobi

Answer: C

25. Crout’s method computes L and U with:

A. Forward elimination only
B. Backward elimination
C. Interleaved multiplication and division
D. Successive row operations

Answer: C

26. In Gauss elimination, backward

substitution cost is:
A. O(n³)
B. O(n²)
C. O(n)
D. Constant

Answer: B

27. The Gauss–Seidel update for the ith

equation uses:
A. Only previous iteration values
B. Both previous and current values
C. Only current iteration values
D. Random values

Answer: B

28. If A is strictly diagonally dominant by

rows, then:
A. Jacobi fails, GS converges
B. GS fails, Jacobi converges
C. Both converge
D. Both diverge

Answer: C

29. For SOR, as ω → 1, convergence:

A. Stops
B. Reduces to GS
C. Becomes unstable
D. Diverges

Answer: B

30. Round-off error accumulates most

severely in elimination when matrix is:
A. Well-conditioned
B. Poorly conditioned
C. Unitary
D. Diagonal
Answer: B
Explanation:
Large condition number → amplified numerical error.

30 MCQs — Interpolation
(Forward, Backward & Central
Methods)
(Difficulty: Ph.D. Entrance Level — Conceptual & Analytical)

1. Newton’s Forward Interpolation is most

accurate when the interpolation point is:
A. Near the end of the table
B. Near the middle
C. Close to the beginning
D. Far outside the table
Answer: C

2. Newton’s Backward Interpolation is

preferred when:
A. x lies near the first node
B. x lies near any node
C. x lies near the last node
D. Node spacing is nonuniform
Answer: C
3. For equally spaced nodes, forward and
backward interpolation formulas differ only
in:
A. The order of differences
B. The difference operator used
C. Step size
D. The method of least squares
Answer: B

4. The Gauss Forward method reduces to

Newton’s Forward method when:
A. x = x0
B. x = xn
C. Central point equals the first point
D. Spacing becomes zero
Answer: C

5. Central difference methods are preferred

when:
A. Interpolation point lies near the boundaries
B. Nodes are unequally spaced
C. Interpolation point lies near the middle
D. Higher-order differences are negligible
Answer: C
6. In Stirling's interpolation formula, the term
chosen first after Δ0 is:
A. ∇
B. δ
C. μ
D. Central difference Δ1/2
Answer: D

7. Bessel’s formula is preferred over

Stirling’s when:
A. Differences are erratic
B. x lies exactly at mid-node
C. x is slightly off the central point
D. Forward differences vanish
Answer: C

8. For forward interpolation, the error term

contains:
A. Δⁿ⁼ᵐ f(a)
B. f⁽ⁿ⁺¹⁾(ξ)/(n+1)! hⁿ⁺¹
C. Difference quotients only
D. Only odd-order derivatives
Answer: B
9. The interpolation polynomial of degree n
for equally spaced nodes can be expressed
using:
A. Forward differences
B. Backward differences
C. Central differences
D. Any of the above
Answer: D

10. In central difference notation, δf(x)

represents:
A. Forward difference
B. Backward difference
C. Symmetric difference
D. Second forward difference
Answer: C

11. The Gauss Backward formula becomes

identical to Newton Backward formula when:
A. Central node equals last node
B. First differences vanish
C. h → 0
D. x lies exactly at the center
Answer: A
12. For which method is the interpolating
polynomial symmetric in μ?
A. Newton Forward
B. Gauss Forward
C. Stirling’s
D. Newton Backward
Answer: C

13. Everett’s formula differs from Bessel’s

because Everett:
A. Uses only even differences
B. Uses only odd differences
C. Uses both Δ and ∇ symmetrically
D. Avoids central difference approximations
Answer: C

14. Which method gives best results when

the table entries are highly smooth
(analytic)?
A. Forward
B. Backward
C. Central formulas
D. Lagrange only
Answer: C
15. Which method uses μ defined as (x -
x0)/h?
A. Forward only
B. Backward only
C. All Newton and Gauss forms
D. Only Stirling
Answer: C

16. Gauss forward interpolation starts with

the central difference:
A. Δ0
B. Δ1
C. Δ−1
D. Δ1/2
Answer: D

17. For smooth data, the convergence rate of

central interpolation is:
A. Slower than forward
B. Faster than forward/backward
C. Same as forward
D. Depends only on h
Answer: B
18. The main advantage of backward
interpolation is:
A. Reduces oscillation error
B. Utilizes tabulated values at the end
C. Uses symmetric differences
D. Works with unevenly spaced points
Answer: B

19. In interpolation, zero of which operator

corresponds to equally spaced nodes?
A. D (differential)
B. Δ
C. ∇
D. δ
Answer: B

20. In Newton forward formula, the term μ(μ–

1)(μ–2)… corresponds to:
A. Central differences
B. Rising factorial
C. Falling factorial
D. Harmonic differences
Answer: C

21. Stirling’s formula is a combination of:

A. Two backward formulas
B. Two forward formulas
C. One forward and one backward formula
D. Only Gauss backward
Answer: C

22. The correct order of accuracy (local

truncation error) for central interpolation is:
A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: B

23. In Bessel’s formula, which pairs of d

differences are averaged?
A. Δ0 and Δ2
B. Δ1 and Δ−1
C. Δ1/2 and Δ−1/2
D. Δ0 and ∇0
Answer: C

24. Everett’s formula reduces truncation

error by:
A. Cancelling odd differences
B. Eliminating even differences
C. Using complementary interpolation weights
D. Using least squares
Answer: C
25. The error in Newton backward
interpolation depends on:
A. Δf(x0)
B. f⁽ⁿ⁺¹⁾(η)
C. ∇f(xn) only
D. δf(xm)
Answer: B

26. Which interpolation formula is best suited

for evaluating derivatives?
A. Forward
B. Backward
C. Central
D. Lagrange only
Answer: C

27. Central methods give inaccurate results

when:
A. Differences are smooth
B. Interpolation point is near boundary
C. x is exactly middle
D. Step h is small
Answer: B
28. The property of central interpolation
polynomials is that they are:
A. Always odd degree
B. Symmetric in even/odd difference terms
C. Require zero-order differences
D. Use backward differences only
Answer: B

29. In Newton-Gregory forward formula, Δ^n

f(x0) contributes significantly when:
A. Function is nearly linear
B. Polynomial degree is low
C. Function varies rapidly
D. Differences vanish
Answer: C

30. Gauss forward and backward formula are

primarily useful because:
A. They reduce computational effort
B. They allow polynomial degree choice
C. They allow interpolation near the center
D. They do not need difference tables
Answer: C

30 MCQs — Numerical Integration (Only

Part A)
1. The truncation error of the trapezoidal rule is
proportional to:
A. h²
B. h³
C. h⁴
D. h
Answer: B

2. Simpson’s 1/3 rule exactly integrates polynomials of

degree up to:
A. 1
B. 2
C. 3
D. 4
Answer: C

3. For Weddle’s rule, the number of subintervals must be:

A. Odd
B. Even
C. Multiple of 3
D. Multiple of 6
Answer: D

4. The error term in Simpson’s 1/3 rule involves which

derivative?
A. f′′(ξ)
B. f′′′(ξ)
C. f⁽⁴⁾(ξ)
D. f⁽⁶⁾(ξ)
Answer: C

5. Composite trapezoidal rule becomes exact when f(x) is:

A. Linear
B. Quadratic
C. Cubic
D. Quartic
Answer: A

6. Simpson’s 3/8 rule is exact for polynomials of degree:

A. ≤ 2
B. ≤ 3
C. ≤ 4
D. ≤ 5
Answer: B

7. Weddle’s rule is derived by fitting a polynomial of degree:

A. 4
B. 5
C. 6
D. 7
Answer: C

8. Simpson’s 1/3 rule uses weights:

A. 1, 2, 2, 1
B. 1, 3, 3, 1
C. 1, 4, 2, 4, 1
D. 1, 6, 1
Answer: C

9. The order of convergence of the composite Simpson’s 1/3

rule is:
A. O(h²)
B. O(h³)
C. O(h⁴)
D. O(h⁶)
Answer: C

10. Which rule is more accurate for the same number of

intervals?
A. Trapezoidal
B. Simpson’s 1/3
C. Simpson’s 3/8
D. Weddle’s
Answer: D

11. The composite trapezoidal rule error term involves:

A. f′′
B. f′
C. f⁽⁴⁾
D. f⁽⁶⁾
Answer: A
12. Newton–Cotes formulas show poor behavior due to:
A. Small h
B. Round-off error
C. High-order polynomial oscillations
D. Smoothness of f(x)
Answer: C

13. Simpson’s 3/8 rule uses how many nodes per segment?
A. 2
B. 3
C. 4
D. 5
Answer: C

14. Weddle’s rule coefficients include the set:

A. 1, 1, 1, 1, 1
B. 1, 4, 1
C. 1, 5, 1
D. 1, 5, 1, 6, 1, 5, 1
Answer: D

15. When f(x) is highly oscillatory, Newton–Cotes formulas:

A. Improve accuracy
B. Become unstable
C. Do not change
D. Become exact
Answer: B
16. The basic assumption behind Newton–Cotes rules is:
A. f(x) is discontinuous
B. f(x) is approximated by polynomials
C. f(x) must be periodic
D. Step-size must be variable
Answer: B

17. Simpson’s 1/3 rule can be interpreted as integrating the:

A. Tangent line
B. Quadratic interpolant
C. Cubic interpolant
D. Spline interpolant
Answer: B

18. The main limitation of using higher-order Newton–Cotes

formulas is:
A. Increased computation time
B. Polynomial oscillation near endpoints
C. Unavailability of derivatives
D. Larger h is required
Answer: B

19. The error constant in Simpson’s 3/8 rule is proportional

to:
A. h³
B. h⁴
C. h⁵
D. h⁶
Answer: B

20. Composite Weddle’s rule requires:

A. Even number of intervals
B. Odd number of intervals
C. Multiple of 6 intervals
D. No restriction
Answer: C

21. Gaussian quadrature is superior because it:

A. Uses equidistant nodes
B. Minimizes truncation error
C. Uses polynomial interpolation at endpoints
D. Avoids composite formulas
Answer: B

22. The error in the trapezoidal rule decreases when:

A. f′′(x) is small
B. f′(x) is oscillatory
C. Higher-order derivatives are large
D. h is increased
Answer: A

23. For smooth functions, the best Newton–Cotes rule is:

A. Trapezoidal
B. Simpson’s 1/3
C. Simpson’s 3/8
D. Weddle’s
Answer: D

24. Simpson’s composite rule requires the number of

subintervals to be:
A. Odd
B. Even
C. Prime
D. Multiple of 4
Answer: B

25. If f(x) is cubic, which rule integrates exactly?

A. Trapezoidal
B. Simpson’s 1/3
C. Simpson’s 3/8
D. Both B and C
Answer: D

26. Simpson’s 1/3 rule fails if:

A. Number of intervals is even
B. f(x) is linear
C. Number of intervals is odd
D. Step size is uniform
Answer: C
27. The Euler-Maclaurin formula is used to derive:
A. Taylor series
B. Newton–Cotes formulas
C. Gauss quadrature
D. Runge–Kutta methods
Answer: B

28. Romberg integration accelerates convergence of:

A. Simpson’s rule
B. Trapezoidal rule
C. Weddle’s rule
D. Gaussian quadrature
Answer: B

29. Simpson’s 3/8 rule is generally less used because:

A. It is less accurate than 1/3 rule
B. It requires more derivative information
C. It has less favorable error constant
D. It requires non-uniform spacing
Answer: C

30. For fixed n, the most accurate Newton–Cotes rule among

those listed is:
A. Trapezoidal
B. Simpson’s 1/3
C. Simpson’s 3/8
D. Weddle’s
Answer: D
Eigenvalues & Eigenvectors — 30 Advanced
MCQs (Ph.D. Level)
(Linear Algebra • Numerical Linear Algebra • Spectral Theory)

PART–B: 30 MCQs (with Answers)

1. A matrix is diagonalizable iff:

A. It has n distinct eigenvalues
B. Its characteristic polynomial splits
C. Its minimal polynomial has no repeated roots
D. Its determinant is non-zero
Answer: C

2. For a real symmetric matrix, eigenvalues are:

A. Complex only
B. Real
C. Unit magnitude
D. Repeated
Answer: B

3. Eigenvectors of a symmetric matrix corresponding to

distinct eigenvalues are:
A. Linearly dependent
B. Orthogonal
C. Complex
D. Arbitrarily oriented
Answer: B

4. The eigenvalues of an orthogonal matrix lie on the:

A. Real line
B. Imaginary axis
C. Unit circle
D. Positive axis
Answer: C

5. The determinant of a matrix equals:

A. Sum of eigenvalues
B. Product of eigenvalues
C. Spectral radius
D. Largest singular value
Answer: B

6. The power method converges when:

A. All eigenvalues are equal
B. Dominant eigenvalue is unique in magnitude
C. Matrix is defective
D. Matrix is singular
Answer: B

7. The Rayleigh quotient equals an eigenvalue when:

A. x is any vector
B. x is orthonormal
C. x is an eigenvector
D. A is orthogonal
Answer: C

8. A defective matrix has:

A. No eigenvalues
B. Geometric multiplicity < algebraic multiplicity
C. Zero determinant
D. All eigenvalues real
Answer: B

9. The eigenvalues of A⁻¹ are:

A. Same as A
B. Squares of eigenvalues of A
C. Reciprocals of eigenvalues of A
D. Negative of eigenvalues of A
Answer: C

10. Gershgorin’s theorem provides bounds on:

A. Eigenvectors
B. Singular values
C. Determinant
D. Location of eigenvalues
Answer: D

11. QR algorithm converges for:

A. Symmetric matrices only
B. Positive definite matrices only
C. All square matrices
D. Only diagonalizable matrices
Answer: C

12. A Hermitian matrix has:

A. Complex eigenvalues
B. Real eigenvalues
C. Imaginary eigenvalues
D. Zero eigenvalues only
Answer: B

13. If A is idempotent (A² = A), then eigenvalues are:

A. 0 or 2
B. 1 only
C. 0 or 1
D. ±1
Answer: C

14. For a skew-symmetric matrix, eigenvalues are:

A. Real
B. Imaginary or zero
C. Positive
D. Unit magnitude
Answer: B

15. A normal matrix satisfies:

A. AᵀA = AAᵀ
B. A² = A
C. A⁻¹ exists
D. A is diagonalizable by QR
Answer: A

16. Spectral radius ρ(A) is:

A. Largest singular value
B. Largest absolute eigenvalue
C. Trace of A
D. Frobenius norm
Answer: B

17. A matrix is orthogonally diagonalizable iff it is:

A. Symmetric
B. Diagonalizable
C. Normal
D. Invertible
Answer: A

18. For a triangular matrix, eigenvalues are:

A. Zero
B. Diagonal entries
C. Off-diagonal entries
D. Determinant
Answer: B

19. For a positive definite matrix, eigenvalues are:

A. Negative
B. Non-negative
C. > 0
D. Complex
Answer: C

20. The Jacobi method for eigenvalues is applicable to:

A. Any matrix
B. Symmetric matrices only
C. Orthogonal matrices
D. Singular matrices
Answer: B

21. Singular values of A are:

A. Absolute values of eigenvalues of A
B. Square roots of eigenvalues of AᵀA
C. Eigenvalues of A⁻¹
D. Determinants of A
Answer: B

22. The principal eigenvector of a stochastic matrix

represents:
A. Eigenvalue 0
B. Stationary distribution
C. Null space
D. Largest singular value
Answer: B
23. Two matrices are similar iff they have the same:
A. Determinant only
B. Rank
C. Eigenvalues with multiplicities
D. Eigenvectors
Answer: C

24. The minimal polynomial of A:

A. Always splits
B. Has roots that are eigenvalues
C. Determines the spectral radius
D. Equals the characteristic polynomial
Answer: B

25. The eigenvalues of a projection matrix satisfy:

A. λ = 0 or λ = 1
B. λ = −1 or 1
C. λ = ±i
D. λ > 1
Answer: A

26. If A has an eigenvalue λ with eigenvector x, then for any

scalar α ≠ 0, x̂ = αx is:
A. Not an eigenvector
B. A new eigenvalue
C. Also an eigenvector
D. A singular vector
Answer: C
27. The spectral decomposition A = PDP⁻¹ exists when:
A. A is normal
B. A is diagonalizable
C. A is nilpotent
D. A is invertible
Answer: B

28. For a real symmetric matrix, the spectral norm equals:

A. Largest trace value
B. Largest eigenvalue magnitude
C. Product of eigenvalues
D. Square root of determinant
Answer: B

29. Inverse iteration method finds:

A. Largest eigenvalue
B. Smallest eigenvalue
C. All eigenvalues
D. Only real eigenvalues
Answer: B

30. A matrix has a repeated eigenvalue but is diagonalizable

iff:
A. Jordan blocks > 1
B. Geometric multiplicity = algebraic multiplicity
C. Determinant = 0
D. Trace = 0
Answer: B

30 MCQs — Numerical Solutions of ODEs

(Euler, Modified Euler, RK4)
(Ph.D.-level difficulty)

1. Euler’s method is derived by applying:

A. Forward difference to y′
B. Backward difference to y′
C. Taylor series up to first term
D. Taylor series up to second term
Answer: C

2. The local truncation error of Euler’s method is:

A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: B

3. The global truncation error of Euler’s method is:

A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: A
4. Modified Euler (Heun’s) method corresponds to which
Runge–Kutta order?
A. First-order
B. Second-order
C. Third-order
D. Fourth-order
Answer: B

5. The stability region of Euler’s method is:

A. |1 + hλ| ≤ 1
B. |1 − hλ| ≤ 1
C. |1 + hλ| ≥ 1
D. hλ = 0
Answer: A

6. RK4 is derived to match Taylor series up to order:

A. 2
B. 3
C. 4
D. 5
Answer: C

7. For Heun’s method, the predictor step is:

A. yₙ = yₙ₋₁ + hf(xₙ₋₁, yₙ₋₁)
B. y* = yₙ + hf(xₙ, yₙ)
C. y* = yₙ − hf(xₙ, yₙ)
D. y* = yₙ + h/2 f(xₙ, yₙ)
Answer: B

8. Modified Euler’s method uses how many function

evaluations per step?
A. 1
B. 2
C. 3
D. 4
Answer: B

9. RK4 uses how many function evaluations per step?

A. 2
B. 3
C. 4
D. 6
Answer: C

10. For RK4, k2k_2k2 uses which point?

A. (xₙ, yₙ)
B. (xₙ + h/2, yₙ + hk₁)
C. (xₙ + h/2, yₙ + hk₁/2)
D. (xₙ + h, yₙ + hk₃)
Answer: C
11. Euler’s method is:
A. Explicit
B. Implicit
C. Predictor-corrector
D. Multistep
Answer: A

12. Modified Euler is:

A. Always implicit
B. Always explicit
C. Predictor-corrector explicit
D. Backward method
Answer: C

13. RK4 is:

A. Explicit single-step method
B. Implicit
C. Multistep
D. Linearly implicit
Answer: A

14. Which method has highest accuracy per function

evaluation?
A. Euler
B. Modified Euler
C. RK4
D. All equal
Answer: C
15. Stability of Euler’s method fails when solving y′ = λy if:
A. hλ > 0
B. hλ < −2
C. hλ = 1
D. hλ = 0
Answer: B

16. For stiff problems, Euler’s method:

A. Is highly stable
B. Fails unless h is very small
C. Performs same as RK4
D. Is unconditionally stable
Answer: B

17. Modified Euler’s method gives exact results for which

DE?
A. Linear only
B. y′ = constant
C. y′ = ax + b
D. y′ = ay
Answer: B

18. RK4 gives exact results for:

A. All linear ODE
B. DEs whose exact solution is polynomial of degree ≤ 3
C. Polynomial solutions of degree ≤ 4
D. All nonlinear DE
Answer: B

19. Step-doubling error estimate for Euler’s method

improves order to:
A. 1
B. 2
C. 3
D. 4
Answer: B

20. In Modified Euler, the corrector uses:

A. Average of slopes
B. Slope at start only
C. Slope at end only
D. Integration formula
Answer: A

21. RK4 can be seen as weighted average of slopes with

weights:
A. 1, 1
B. 1, 2, 2, 1
C. 1, 1, 1, 1
D. 2, 1, 1, 2
Answer: B
22. Error constant of RK4 is:
A. 1/2
B. 1/6
C. 1/24
D. 1/90
Answer: D

23. Stability region of RK4 is approximately:

A. Circle
B. Parabola
C. Asymmetrical bulb shape
D. Square
Answer: C

24. Euler’s method is conditionally stable for:

A. All λ
B. λ > 0
C. λ < 0
D. h small enough
Answer: D

25. Modified Euler’s method is equivalent to applying:

A. Simpson’s rule
B. Trapezoidal rule
C. Midpoint rule
D. BDF2
Answer: B
26. RK4 becomes unstable if step size is:
A. Too large
B. Too small
C. Equal to λ
D. Zero
Answer: A

27. Local truncation error of Modified Euler is:

A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: C

28. Global error of Modified Euler is:

A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: B

29. RK4 global error is:

A. O(h²)
B. O(h³)
C. O(h⁴)
D. O(h⁵)
Answer: C
30. Most suitable method for highly nonlinear stiff systems:
A. Euler
B. Modified Euler
C. RK4
D. None of these (need implicit methods)
Answer: D

30 MCQs — Numerical PDEs (Parabolic,

Elliptic, Hyperbolic)
(Ph.D.–level difficulty)

1. The FTCS scheme for the heat equation is stable only if:
A. r ≤ 1
B. r ≤ 1/2
C. r ≤ 1/4
D. r ≤ 2

where r=αΔtΔx2r = \alpha \frac{\Delta t}{\Delta x^{2}}r=αΔx2Δt

Answer: B

2. The Crank–Nicolson method applied to the heat equation

is:
A. Conditionally stable
B. Unconditionally stable
C. Unstable for small Δt
D. Explicit
Answer: B
3. The backward-time central-space scheme (BTCS) is:
A. Explicit
B. Conditionally stable
C. Unconditionally stable
D. 2nd-order in time
Answer: C

4. The classification of PDEs is based on:

A. Order of derivatives

C. Discriminant B2−4ACB^2 - 4ACB2−4AC

B. Boundary conditions

D. Smoothness of solution
Answer: C

5. Laplace's equation is:

A. Parabolic
B. Elliptic
C. Hyperbolic
D. Mixed
Answer: B

6. The wave equation is an example of:

A. Parabolic PDE
B. Hyperbolic PDE
C. Elliptic PDE
D. Integral equation
Answer: B
7. The CFL condition for the 1D wave equation (explicit
scheme) is:
A. c Δt / Δx ≤ 1
B. c Δt / Δx ≥ 1
C. c Δt = Δx
D. c Δt / Δx = 0
Answer: A

8. The standard 5-point Laplacian formula is:

A. First-order accurate
B. Second-order accurate
C. Third-order accurate
D. Fourth-order accurate
Answer: B

9. The Crank–Nicolson scheme is derived using:

A. Forward Euler + backward Euler
B. Midpoint averaging
C. Centered time differencing
D. Taylor expansion
Answer: A

10. The ADI (Alternating Direction Implicit) method is used

mainly for:
A. 1D wave equation
B. 1D heat equation
C. Multi-dimensional parabolic problems
D. Poisson equation only
Answer: C

11. Gauss–Seidel method converges faster than Jacobi for

elliptic problems because:
A. Fewer iterations
B. Implicit smoothing
C. Uses updated values immediately
D. Uses larger step sizes
Answer: C

12. Successive over-relaxation (SOR) introduces factor ω to:

A. Improve stability
B. Accelerate convergence
C. Reduce error order
D. Increase truncation error
Answer: B

13. For Laplace equation, boundary conditions are typically:

A. Initial conditions only
B. Mixed type
C. Dirichlet and/or Neumann
D. Integral constraints
Answer: C
14. Hyperbolic PDEs propagate information along:
A. Grid lines
B. Boundary surfaces
C. Characteristics
D. Normal directions
Answer: C

15. Parabolic PDE solutions exhibit:

A. Instantaneous spatial smoothing
B. Oscillatory behavior
C. Static equilibrium
D. Singular behavior
Answer: A

16. Discretization of Poisson equation leads to:

A. Tridiagonal matrix
B. Block tridiagonal system
C. Uncoupled linear system
D. Eigenvalue problem
Answer: B

17. Lax–Wendroff method is used for:

A. Elliptic PDEs
B. Hyperbolic PDEs
C. Parabolic PDEs
D. Mixed PDEs
Answer: B
18. FTCS is unstable for advection equation because:
A. Diffusion is dominant
B. Central differencing introduces phase error
C. No time-centering
D. Higher-order truncation error
Answer: B

19. A PDE of form utt=c2uxxu_{tt} = c^2 u_{xx}utt

=c2uxx requires:
A. One initial condition
B. Two initial conditions
C. No initial condition
D. Boundary condition only
Answer: B

20. The Crank–Nicolson scheme has which accuracy?

A. 1st order time, 2nd space
B. 2nd order time, 1st space
C. 2nd order time, 2nd space
D. 4th order both
Answer: C

21. The 9-point Laplacian improves order to:

A. O(h)
B. O(h²)
C. O(h³)
D. O(h⁴)
Answer: D
22. Explicit schemes for hyperbolic PDEs fail when:
A. CFL condition is violated
B. Δx = Δt
C. CFL = 0
D. Coefficients are constant
Answer: A

23. In ADI method, splitting is possible because:

A. Operators commute approximately
B. Grid is uniform
C. Boundary conditions are constant
D. Time step is small
Answer: A

24. Convergence of SOR is fastest when ω is:

A. 0 < ω < 1
B. ω = 1
C. 1 < ω < 2
D. ω > 2
Answer: C

25. Discretization of 2D heat equation using FTCS needs

stability:
A. r < 1/3
B. r < 1/4
C. r < 1/2
D. r < 1/5
Answer: B

26. The Laplacian operator is:

A. Divergence of gradient
B. Curl of gradient
C. Sum of first derivatives
D. Product of gradients
Answer: A

27. Hyperbolic PDE numerical solutions often require:

A. Artificial viscosity
B. Larger Δt
C. Boundary-only information
D. Neumann-only boundary conditions
Answer: A

28. Elliptic PDEs describe:

A. Steady-state processes
B. Dynamic diffusion
C. Wave motion
D. Shock formation
Answer: A
29. For Poisson equation, iterative methods converge if the
spectral radius ρ < 1. The role of relaxation ω is to:
A. Maximize ρ
B. Minimize ρ
C. Make ρ = 1
D. Remove eigenmodes
Answer: B

30. McCormack scheme is used for:

A. Elliptic PDEs
B. Parabolic PDEs
C. Hyperbolic conservation laws
D. Integral PDEs
Answer: C