Skip to main content

Functions — Diagnostic Tests

Unit Tests

Tests edge cases, boundary conditions, and common misconceptions for functions.

UT-1: Domain Restrictions in Composite Functions

Question:

Given f(x)=2x1f(x) = \sqrt{2x - 1} and g(x)=1x3g(x) = \frac{1}{x - 3}:

(a) Find the domain of fgf \circ g, i.e. f(g(x))f(g(x)).

(b) Find the domain of gfg \circ f, i.e. g(f(x))g(f(x)).

(c) Explain why the domains of fgf \circ g and gfg \circ f are different, identifying the specific restriction that causes the difference.

[Difficulty: hard. Tests the common error of taking the domain of a composite function as the intersection of individual domains, rather than considering the range of the inner function.]

Solution:

(a) f(g(x))=f(1x3)=LB2x31RB=LB2(x3)x3RB=LB5xx3RBf(g(x)) = f\left(\frac{1}{x-3}\right) = \sqrt◆LB◆\frac{2}{x-3} - 1◆RB◆ = \sqrt◆LB◆\frac{2 - (x-3)}{x-3}◆RB◆ = \sqrt◆LB◆\frac{5-x}{x-3}◆RB◆.

For this to be defined, we need:

  1. The expression inside the square root to be non-negative: 5xx30\frac{5-x}{x-3} \geq 0.
  2. The denominator of the original gg to be non-zero: x3x \neq 3.

Solving 5xx30\frac{5-x}{x-3} \geq 0 by sign analysis:

Critical values: x=3x = 3 (excluded, denominator zero) and x=5x = 5 (included, numerator zero).

Interval(5x)(5-x)(x3)(x-3)Ratio
x<3x < 3++--
3<x53 < x \leq 5++++++
x>5x > 5-++-

Domain of fgf \circ g: (3,5](3, 5].

(b) g(f(x))=g(2x1)=LB1RB◆◆LB2x13RBg(f(x)) = g(\sqrt{2x-1}) = \frac◆LB◆1◆RB◆◆LB◆\sqrt{2x-1} - 3◆RB◆.

For this to be defined:

  1. 2x10    x122x - 1 \geq 0 \implies x \geq \frac{1}{2} (domain of ff).
  2. 2x130    2x13    2x19    x5\sqrt{2x-1} - 3 \neq 0 \implies \sqrt{2x-1} \neq 3 \implies 2x-1 \neq 9 \implies x \neq 5.

Domain of gfg \circ f: [12,5)(5,)\left[\frac{1}{2}, 5\right) \cup (5, \infty).

(c) The domains differ because of the direction of composition:

  • For fgf \circ g: the input to ff is g(x)=1x3g(x) = \frac{1}{x-3}, and we need g(x)1/2g(x) \geq 1/2 (since ff requires 2g(x)102 \cdot g(x) - 1 \geq 0, i.e. g(x)1/2g(x) \geq 1/2). This constrains xx to a finite interval (3,5](3, 5].
  • For gfg \circ f: the input to gg is f(x)=2x1f(x) = \sqrt{2x-1}, and we need f(x)3f(x) \neq 3. Since f(x)0f(x) \geq 0 for all xx in its domain, we only exclude x=5x = 5. The domain is almost the entire domain of ff.

The key difference is that fgf \circ g requires the output of gg to fall within the domain of ff (which is [1/2,)[1/2, \infty)), while gfg \circ f requires the output of ff to avoid the single excluded value of gg (which is 33). The former is much more restrictive because g(x)=1/(x3)g(x) = 1/(x-3) only achieves values 1/2\geq 1/2 on a finite interval.


UT-2: Inverse Function Existence and Domain Restriction

Question:

The function f(x)=x2+4xf(x) = x^2 + 4x is defined on the domain x2x \leq -2.

(a) Explain why ff has an inverse on this domain.

(b) Find f1(x)f^{-1}(x), stating its domain and range.

(c) If the domain were instead x0x \geq 0, find the range of f1f^{-1} in this case.

[Difficulty: hard. Tests understanding that a function must be one-to-one to have an inverse, and how domain restriction affects the inverse.]

Solution:

(a) To show ff is one-to-one on x2x \leq -2, we show it is strictly monotonic (strictly decreasing) on this domain.

f(x)=2x+4f'(x) = 2x + 4

For x2x \leq -2: 2x+402x + 4 \leq 0, with equality only at x=2x = -2.

For x<2x < -2: f(x)<0f'(x) < 0, so ff is strictly decreasing on (,2](-\infty, -2].

Since ff is strictly decreasing, it is one-to-one, and therefore has an inverse.

(b) Let y=x2+4xy = x^2 + 4x. Completing the square: y=(x+2)24y = (x+2)^2 - 4.

Solving for xx: (x+2)2=y+4(x+2)^2 = y + 4, so x+2=±y+4x + 2 = \pm\sqrt{y+4}.

Since x2x \leq -2, we have x+20x + 2 \leq 0, so we take the negative root:

x=2y+4x = -2 - \sqrt{y+4}

Therefore f1(x)=2x+4f^{-1}(x) = -2 - \sqrt{x+4}.

Domain of f1f^{-1}: We need x+40x + 4 \geq 0, so x4x \geq -4. Also, the range of ff on x2x \leq -2: since ff is decreasing, as xx \to -\infty, f(x)+f(x) \to +\infty, and f(2)=0f(-2) = 0. So the range of ff is [0,)[0, \infty).

Wait: completing the square gives f(x)=(x+2)24f(x) = (x+2)^2 - 4. At x=2x = -2: f(2)=4f(-2) = -4. As xx \to -\infty: f(x)+f(x) \to +\infty. Since ff is decreasing on (,2](-\infty, -2], the range is [4,)[-4, \infty).

Therefore domain of f1f^{-1} is [4,)[-4, \infty), confirming x+40x + 4 \geq 0.

Range of f1f^{-1} equals domain of ff: (,2](-\infty, -2].

(c) If the domain were x0x \geq 0: f(x)=2x+4>0f'(x) = 2x + 4 > 0 for all x0x \geq 0, so ff is strictly increasing.

Range of ff: f(0)=0f(0) = 0 and f(x)+f(x) \to +\infty as x+x \to +\infty. So range is [0,)[0, \infty).

The inverse would be f1(x)=2+x+4f^{-1}(x) = -2 + \sqrt{x+4} (taking the positive root since x+22>0x + 2 \geq 2 > 0).

Range of f1f^{-1} equals domain of ff: [0,)[0, \infty).


UT-3: Three-Layer Composite Function

Question:

Let f(x)=2x+1f(x) = \frac{2}{x+1}, g(x)=x21g(x) = x^2 - 1, and h(x)=xh(x) = \sqrt{x}.

(a) Find fghf \circ g \circ h in its simplest form, stating its domain.

(b) Solve the equation (fgh)(x)=1(f \circ g \circ h)(x) = 1.

(c) Show that (fgh)(x)=(gfh)(x)(f \circ g \circ h)(x) = (g \circ f \circ h)(x) and explain algebraically why this identity holds.

[Difficulty: hard. Tests multi-layer composition with domain tracking and algebraic identity verification.]

Solution:

(a) Working from the inside out:

h(x)=xh(x) = \sqrt{x} (domain: x0x \geq 0)

g(h(x))=(x)21=x1g(h(x)) = (\sqrt{x})^2 - 1 = x - 1 (domain: x0x \geq 0, since we need h(x)h(x) defined first)

f(g(h(x)))=f(x1)=2(x1)+1=2xf(g(h(x))) = f(x - 1) = \frac{2}{(x-1)+1} = \frac{2}{x} (domain: x11    x0x - 1 \neq -1 \implies x \neq 0, combined with x0x \geq 0)

(fgh)(x)=2x,domain: x>0(f \circ g \circ h)(x) = \frac{2}{x}, \quad \text{domain: } x > 0

(b) 2x=1    x=2\frac{2}{x} = 1 \implies x = 2.

Check: x=2>0x = 2 > 0, so it is in the domain. Also verify through each layer:

  • h(2)=2h(2) = \sqrt{2}
  • g(2)=21=1g(\sqrt{2}) = 2 - 1 = 1
  • f(1)=22=1f(1) = \frac{2}{2} = 1. Confirmed.

(c) Compute gfhg \circ f \circ h:

h(x)=xh(x) = \sqrt{x}

f(h(x))=LB2RB◆◆LBx+1RBf(h(x)) = \frac◆LB◆2◆RB◆◆LB◆\sqrt{x}+1◆RB◆ (domain: x0x \geq 0, x0x \neq 0, so x>0x > 0)

g(f(h(x)))=(LB2RB◆◆LBx+1RB)21=LB4RB◆◆LB(x+1)2RB1=LB4(x+1)2RB◆◆LB(x+1)2RBg(f(h(x))) = \left(\frac◆LB◆2◆RB◆◆LB◆\sqrt{x}+1◆RB◆\right)^2 - 1 = \frac◆LB◆4◆RB◆◆LB◆(\sqrt{x}+1)^2◆RB◆ - 1 = \frac◆LB◆4 - (\sqrt{x}+1)^2◆RB◆◆LB◆(\sqrt{x}+1)^2◆RB◆

=LB4(x+2x+1)RB◆◆LB(x+1)2RB=LB3x2xRB◆◆LB(x+1)2RB= \frac◆LB◆4 - (x + 2\sqrt{x} + 1)◆RB◆◆LB◆(\sqrt{x}+1)^2◆RB◆ = \frac◆LB◆3 - x - 2\sqrt{x}◆RB◆◆LB◆(\sqrt{x}+1)^2◆RB◆

This does not equal 2x\frac{2}{x} in general. Let me verify with x=4x = 4:

  • (fgh)(4)=24=12(f \circ g \circ h)(4) = \frac{2}{4} = \frac{1}{2}
  • (gfh)(4)=(23)21=491=59(g \circ f \circ h)(4) = \left(\frac{2}{3}\right)^2 - 1 = \frac{4}{9} - 1 = -\frac{5}{9}

These are not equal. The claim in part (c) is false. The identity (fgh)(x)=(gfh)(x)(f \circ g \circ h)(x) = (g \circ f \circ h)(x) does not hold in general. Function composition is not commutative.

Let me re-examine: (fgh)(x)=f(g(h(x)))(f \circ g \circ h)(x) = f(g(h(x))) versus (gfh)(x)=g(f(h(x)))(g \circ f \circ h)(x) = g(f(h(x))). The inner function hh is the same, but ff and gg are applied in different orders. Since ff and gg do not commute, the results differ.


Integration Tests

Tests synthesis of functions with other topics. Requires combining concepts from multiple units.

IT-1: Exponential Function Composition with Logarithmic Inequality (with Exponentials and Logarithms)

Question:

The function ff is defined by f(x)=e2x+e2xf(x) = e^{2x} + e^{-2x} for xRx \in \mathbb{R}.

(a) Show that f(x)2f(x) \geq 2 for all xRx \in \mathbb{R}, and find the value of xx for which equality holds.

(b) The function gg is defined by g(x)=ln(x+x2+1)g(x) = \ln(x + \sqrt{x^2 + 1}) for xRx \in \mathbb{R}. Find f(g(x))f(g(x)) in terms of xx and simplify your answer.

(c) Solve the inequality f(g(x))10f(g(x)) \leq 10.

[Difficulty: hard. Combines exponential properties, hyperbolic cosine identities, and logarithmic manipulation.]

Solution:

(a) By the AM-GM inequality, for a=e2x>0a = e^{2x} > 0 and b=e2x>0b = e^{-2x} > 0:

a+b2ab=LBe2xe2xRB=1=1\frac{a + b}{2} \geq \sqrt{ab} = \sqrt◆LB◆e^{2x} \cdot e^{-2x}◆RB◆ = \sqrt{1} = 1

So f(x)=e2x+e2x2f(x) = e^{2x} + e^{-2x} \geq 2.

Equality holds when a=ba = b, i.e. e2x=e2xe^{2x} = e^{-2x}, giving 4x=04x = 0, so x=0x = 0.

Alternatively, f(x)=2cosh(2x)f(x) = 2\cosh(2x), and since cosh(u)1\cosh(u) \geq 1 for all uu with equality at u=0u = 0, we get f(x)2f(x) \geq 2 with equality at x=0x = 0.

(b) Let u=g(x)=ln(x+x2+1)u = g(x) = \ln(x + \sqrt{x^2 + 1}).

First, note that eu=x+x2+1e^u = x + \sqrt{x^2+1} and eu=LB1RB◆◆LBx+x2+1RB=LBx2+1xRB◆◆LB(x+x2+1)(x2+1x)RB=x2+1xe^{-u} = \frac◆LB◆1◆RB◆◆LB◆x + \sqrt{x^2+1}◆RB◆ = \frac◆LB◆\sqrt{x^2+1}-x◆RB◆◆LB◆(x+\sqrt{x^2+1})(\sqrt{x^2+1}-x)◆RB◆ = \sqrt{x^2+1} - x.

Therefore: e2u=(x+x2+1)2=x2+2xx2+1+x2+1=2x2+1+2xx2+1e^{2u} = (x + \sqrt{x^2+1})^2 = x^2 + 2x\sqrt{x^2+1} + x^2 + 1 = 2x^2 + 1 + 2x\sqrt{x^2+1}

e2u=(x2+1x)2=x2+12xx2+1+x2=2x2+12xx2+1e^{-2u} = (\sqrt{x^2+1} - x)^2 = x^2 + 1 - 2x\sqrt{x^2+1} + x^2 = 2x^2 + 1 - 2x\sqrt{x^2+1}

f(g(x))=e2u+e2u=(2x2+1+2xx2+1)+(2x2+12xx2+1)=4x2+2f(g(x)) = e^{2u} + e^{-2u} = (2x^2 + 1 + 2x\sqrt{x^2+1}) + (2x^2 + 1 - 2x\sqrt{x^2+1}) = 4x^2 + 2

(c) 4x2+210    4x28    x22    2x24x^2 + 2 \leq 10 \implies 4x^2 \leq 8 \implies x^2 \leq 2 \implies -\sqrt{2} \leq x \leq \sqrt{2}.

x[2,2]x \in [-\sqrt{2}, \sqrt{2}]


IT-2: Trigonometric Composition in a Geometry Context (with Trigonometry)

Question:

The function ff is defined by f(θ)=sin(2θ)f(\theta) = \sin(2\theta) and the function gg is defined by g(θ)=sinθ+cosθg(\theta) = \sin\theta + \cos\theta for θ[0,2π)\theta \in [0, 2\pi).

(a) Express g(θ)g(\theta) in the form Rsin(θ+α)R\sin(\theta + \alpha) where R>0R > 0 and 0<α<LBπRB◆◆LB2RB0 < \alpha < \frac◆LB◆\pi◆RB◆◆LB◆2◆RB◆.

(b) Solve the equation f(θ)=g(θ)f(\theta) = g(\theta) for θ[0,2π)\theta \in [0, 2\pi).

(c) A triangle has sides a=3a = 3, b=4b = 4, and the angle between them is CC. The area of the triangle is A=12absinC=f(C2)g(C2)A = \frac{1}{2}ab\sin C = f\left(\frac{C}{2}\right) \cdot g\left(\frac{C}{2}\right). Find the exact value of CC.

[Difficulty: hard. Combines trigonometric identities, harmonic form, equation solving, and geometric application.]

Solution:

(a) g(θ)=sinθ+cosθg(\theta) = \sin\theta + \cos\theta.

R=12+12=2R = \sqrt{1^2 + 1^2} = \sqrt{2}

α=arctan(11)=LBπRB◆◆LB4RB\alpha = \arctan\left(\frac{1}{1}\right) = \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆

g(θ)=2sin(θ+LBπRB◆◆LB4RB)g(\theta) = \sqrt{2}\sin\left(\theta + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆\right)

(b) sin(2θ)=2sin(θ+LBπRB◆◆LB4RB)\sin(2\theta) = \sqrt{2}\sin\left(\theta + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆\right).

Using the double angle formula sin(2θ)=2sinθcosθ\sin(2\theta) = 2\sin\theta\cos\theta:

2sinθcosθ=sinθ+cosθ2\sin\theta\cos\theta = \sin\theta + \cos\theta

2sinθcosθsinθcosθ=02\sin\theta\cos\theta - \sin\theta - \cos\theta = 0

Let u=sinθu = \sin\theta, v=cosθv = \cos\theta. Then 2uvuv=02uv - u - v = 0, i.e. (2u1)(v)u=0(2u - 1)(v) - u = 0, which gives v(2u1)=uv(2u-1) = u.

Alternatively, add 12\frac{1}{2} to both sides:

2sinθcosθsinθcosθ+12=122\sin\theta\cos\theta - \sin\theta - \cos\theta + \frac{1}{2} = \frac{1}{2}

(2sinθ1)(cosθ12)=0(2\sin\theta - 1)(\cos\theta - \frac{1}{2}) = 0

Wait: (2u1)(v1/2)=2uvuv+1/2(2u-1)(v - 1/2) = 2uv - u - v + 1/2. So 2uvuv=(2u1)(v1/2)1/22uv - u - v = (2u-1)(v - 1/2) - 1/2. That doesn't help directly.

Let me factor differently: 2uvuv=0    u(2v1)=v    u=v2v12uv - u - v = 0 \implies u(2v - 1) = v \implies u = \frac{v}{2v-1} (when 2v12v \neq 1).

Also u2+v2=1u^2 + v^2 = 1. This gives v2(2v1)2+v2=1\frac{v^2}{(2v-1)^2} + v^2 = 1, which is a quartic in vv.

A cleaner approach: let t=θ+π/4t = \theta + \pi/4. Then sinθ+cosθ=2sint\sin\theta + \cos\theta = \sqrt{2}\sin t and sin(2θ)=sin(2tπ/2)=cos(2t)\sin(2\theta) = \sin(2t - \pi/2) = -\cos(2t).

cos(2t)=2sint-\cos(2t) = \sqrt{2}\sin t (12sin2t)=2sint-(1 - 2\sin^2 t) = \sqrt{2}\sin t 2sin2t2sint1=02\sin^2 t - \sqrt{2}\sin t - 1 = 0

Let s=sints = \sin t:

s=LB2±2+8RB◆◆LB4RB=LB2±10RB◆◆LB4RBs = \frac◆LB◆\sqrt{2} \pm \sqrt{2 + 8}◆RB◆◆LB◆4◆RB◆ = \frac◆LB◆\sqrt{2} \pm \sqrt{10}◆RB◆◆LB◆4◆RB◆

Since sint1|\sin t| \leq 1: LB2+10RB◆◆LB4RB1.414+3.16241.144>1\frac◆LB◆\sqrt{2} + \sqrt{10}◆RB◆◆LB◆4◆RB◆ \approx \frac{1.414 + 3.162}{4} \approx 1.144 > 1. Not valid.

LB210RB◆◆LB4RB1.4143.16240.437\frac◆LB◆\sqrt{2} - \sqrt{10}◆RB◆◆LB◆4◆RB◆ \approx \frac{1.414 - 3.162}{4} \approx -0.437. Valid.

So sint=LB210RB◆◆LB4RB\sin t = \frac◆LB◆\sqrt{2} - \sqrt{10}◆RB◆◆LB◆4◆RB◆.

t=θ+LBπRB◆◆LB4RBt = \theta + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆, so sin(θ+LBπRB◆◆LB4RB)=LB210RB◆◆LB4RB\sin\left(\theta + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆\right) = \frac◆LB◆\sqrt{2} - \sqrt{10}◆RB◆◆LB◆4◆RB◆.

θ+LBπRB◆◆LB4RB=arcsin(LB210RB◆◆LB4RB)0.452 or π+0.4523.594\theta + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆ = \arcsin\left(\frac◆LB◆\sqrt{2}-\sqrt{10}◆RB◆◆LB◆4◆RB◆\right) \approx -0.452 \text{ or } \pi + 0.452 \approx 3.594

θ0.4520.785=1.237orθ3.5940.785=2.809\theta \approx -0.452 - 0.785 = -1.237 \quad \text{or} \quad \theta \approx 3.594 - 0.785 = 2.809

In [0,2π)[0, 2\pi): θ2.809\theta \approx 2.809 and θ1.237+2π5.046\theta \approx -1.237 + 2\pi \approx 5.046.

The exact solutions are:

θ=arcsin(LB210RB◆◆LB4RB)LBπRB◆◆LB4RB+2πorθ=πarcsin(LB210RB◆◆LB4RB)LBπRB◆◆LB4RB\theta = \arcsin\left(\frac◆LB◆\sqrt{2}-\sqrt{10}◆RB◆◆LB◆4◆RB◆\right) - \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆ + 2\pi \quad \text{or} \quad \theta = \pi - \arcsin\left(\frac◆LB◆\sqrt{2}-\sqrt{10}◆RB◆◆LB◆4◆RB◆\right) - \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆

(c) The area condition gives:

12(3)(4)sinC=sin(2C2)(sinC2+cosC2)\frac{1}{2}(3)(4)\sin C = \sin(2 \cdot \frac{C}{2}) \cdot \left(\sin\frac{C}{2} + \cos\frac{C}{2}\right)

6sinC=sinC(sinC2+cosC2)6\sin C = \sin C \cdot \left(\sin\frac{C}{2} + \cos\frac{C}{2}\right)

If sinC0\sin C \neq 0:

6=sinC2+cosC2=2sin(C2+LBπRB◆◆LB4RB)6 = \sin\frac{C}{2} + \cos\frac{C}{2} = \sqrt{2}\sin\left(\frac{C}{2} + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆\right)

sin(C2+LBπRB◆◆LB4RB)=LB6RB◆◆LB2RB=32\sin\left(\frac{C}{2} + \frac◆LB◆\pi◆RB◆◆LB◆4◆RB◆\right) = \frac◆LB◆6◆RB◆◆LB◆\sqrt{2}◆RB◆ = 3\sqrt{2}

Since 324.24>13\sqrt{2} \approx 4.24 > 1, this is impossible. Therefore sinC=0\sin C = 0, giving C=0C = 0 or C=πC = \pi. Since C=0C = 0 gives a degenerate triangle, C=πC = \pi.

But C=πC = \pi also gives a degenerate triangle (collinear points). This suggests the original problem parameters may need adjustment. In a valid triangle, 0<C<π0 < C < \pi and the area formula 6sinC=sinC(sin(C/2)+cos(C/2))6\sin C = \sin C(\sin(C/2) + \cos(C/2)) requires sin(C/2)+cos(C/2)=6\sin(C/2) + \cos(C/2) = 6, which has no solution since sin(C/2)+cos(C/2)2\sin(C/2) + \cos(C/2) \leq \sqrt{2}.

The problem as stated has no solution for a non-degenerate triangle. This itself is a useful diagnostic insight: recognising when a problem has no valid solution is an important skill.


IT-3: Function Iteration Producing a Sequence (with Sequences)

Question:

The function ff is defined by f(x)=12(x+3x)f(x) = \frac{1}{2}\left(x + \frac{3}{x}\right) for x>0x > 0.

A sequence (an)(a_n) is defined by a1=1a_1 = 1 and an+1=f(an)a_{n+1} = f(a_n) for n1n \geq 1.

(a) Find a2a_2, a3a_3, and a4a_4 as exact fractions.

(b) Prove by induction that an>0a_n > 0 for all n1n \geq 1.

(c) Prove that if an>3a_n > \sqrt{3} then an+1<ana_{n+1} < a_n, and if an<3a_n < \sqrt{3} then an+1>ana_{n+1} > a_n.

(d) State the limit of the sequence (an)(a_n) as nn \to \infty and justify your answer.

[Difficulty: hard. Combines function iteration, proof by induction, and convergence analysis.]

Solution:

(a)

a2=f(1)=12(1+3)=2a_2 = f(1) = \frac{1}{2}\left(1 + 3\right) = 2

a3=f(2)=12(2+32)=1272=74a_3 = f(2) = \frac{1}{2}\left(2 + \frac{3}{2}\right) = \frac{1}{2} \cdot \frac{7}{2} = \frac{7}{4}

a4=f(74)=12(74+LB34RB◆◆LB7RB)=12(74+127)=1249+4828=9756a_4 = f\left(\frac{7}{4}\right) = \frac{1}{2}\left(\frac{7}{4} + \frac◆LB◆3 \cdot 4◆RB◆◆LB◆7◆RB◆\right) = \frac{1}{2}\left(\frac{7}{4} + \frac{12}{7}\right) = \frac{1}{2} \cdot \frac{49 + 48}{28} = \frac{97}{56}

(b) Base case: a1=1>0a_1 = 1 > 0. True.

Inductive step: Assume ak>0a_k > 0 for some k1k \geq 1. Then ak+1=12(ak+3/ak)a_{k+1} = \frac{1}{2}(a_k + 3/a_k). Since ak>0a_k > 0, both aka_k and 3/ak3/a_k are positive, so their sum is positive, and ak+1>0a_{k+1} > 0.

By induction, an>0a_n > 0 for all n1n \geq 1.

(c) Consider an+1an=12(an+3an)an=12(3anan)=3an22ana_{n+1} - a_n = \frac{1}{2}\left(a_n + \frac{3}{a_n}\right) - a_n = \frac{1}{2}\left(\frac{3}{a_n} - a_n\right) = \frac{3 - a_n^2}{2a_n}.

Since an>0a_n > 0 (by part (b)), the sign of an+1ana_{n+1} - a_n is determined by 3an23 - a_n^2:

  • If an>3a_n > \sqrt{3}: an2>3a_n^2 > 3, so 3an2<03 - a_n^2 < 0, giving an+1an<0a_{n+1} - a_n < 0, i.e. an+1<ana_{n+1} < a_n.
  • If an<3a_n < \sqrt{3}: an2<3a_n^2 < 3, so 3an2>03 - a_n^2 > 0, giving an+1an>0a_{n+1} - a_n > 0, i.e. an+1>ana_{n+1} > a_n.

(d) The limit LL must satisfy L=f(L)=12(L+3/L)L = f(L) = \frac{1}{2}(L + 3/L):

2L=L+3L2L = L + \frac{3}{L} L=3LL = \frac{3}{L} L2=3L^2 = 3 L=3L = \sqrt{3}

(We take the positive root since an>0a_n > 0 for all nn.)

Justification: By part (c), if an>3a_n > \sqrt{3} then the sequence decreases, and if an<3a_n < \sqrt{3} then the sequence increases. Since a2=2>31.732a_2 = 2 > \sqrt{3} \approx 1.732, the sequence decreases from n=2n = 2 onwards. The sequence is bounded below by 3\sqrt{3} (since terms above 3\sqrt{3} decrease towards it, and terms below 3\sqrt{3} increase towards it). By the monotone convergence theorem, the sequence converges, and the only possible limit is 3\sqrt{3}.

This function is the Babylonian method (Newton's method) for computing 3\sqrt{3}.