Skip to content

STEP3 2001 -- Pure Mathematics

STEP3 2001 — Section A (Pure Mathematics)

Section titled “STEP3 2001 — Section A (Pure Mathematics)”

Exam: STEP3  |  Year: 2001  |  Questions: Q1—Q8  |  Total marks per question: 20

All questions are pure mathematics. Solutions and examiner commentary are included below.

QTopicDifficultyKey Techniques
1微积分 CalculusChallenging隐函数求导,数学归纳法,递推关系,Maclaurin级数展开
2微积分 CalculusChallenging反双曲函数恒等式,积分求面积,区域边界分析
3代数与方程 Algebra and EquationsStandard判别式分析,韦达定理,不等式约束,区域作图
4微积分 CalculusChallenging反三角函数恒等式,分段函数分析,函数图像描绘
5微积分 CalculusHard三次方程分析,参数方程法线,轨迹方程推导,分类讨论
6立体几何 / 3D GeometryChallenging向量法,平面方程,三角形特殊点,几何证明
7微分方程 / Differential EquationsChallenging正交轨线法,可分离变量微分方程,曲线族与包络
8复数 / Complex NumbersHard复数模和辐角,轨迹方程,分式线性变换,复平面几何

Topic: 微积分 Calculus  |  Difficulty: Challenging  |  Marks: 20

1 Given that y=ln(x+(x2+1))y = \ln (x + \sqrt{(x^2 + 1)}), show that dydx=1(x2+1)\frac{dy}{dx} = \frac{1}{\sqrt{(x^2 + 1)}}.

Prove by induction that, for n0n \geqslant 0,

(x2+1)y(n+2)+(2n+1)xy(n+1)+n2y(n)=0,(x^2 + 1) y^{(n+2)} + (2n + 1) xy^{(n+1)} + n^2y^{(n)} = 0 ,

where y(n)=dnydxny^{(n)} = \frac{d^ny}{dx^n} and y(0)=yy^{(0)} = y.

Using this result in the case x=0x = 0, or otherwise, show that the Maclaurin series for yy begins

xx36+3x540x - \frac{x^3}{6} + \frac{3x^5}{40}

and find the next non-zero term.

Model Solution

Part (a): Show that dydx=1x2+1\frac{dy}{dx} = \frac{1}{\sqrt{x^2 + 1}}

Let y=ln ⁣(x+x2+1)y = \ln\!\left(x + \sqrt{x^2 + 1}\right).

Differentiating using the chain rule:

dydx=1x+x2+1ddx ⁣(x+x2+1)\frac{dy}{dx} = \frac{1}{x + \sqrt{x^2 + 1}} \cdot \frac{d}{dx}\!\left(x + \sqrt{x^2 + 1}\right)

=1x+x2+1(1+xx2+1)= \frac{1}{x + \sqrt{x^2 + 1}} \cdot \left(1 + \frac{x}{\sqrt{x^2 + 1}}\right)

=1x+x2+1x2+1+xx2+1= \frac{1}{x + \sqrt{x^2 + 1}} \cdot \frac{\sqrt{x^2 + 1} + x}{\sqrt{x^2 + 1}}

=1x2+1= \frac{1}{\sqrt{x^2 + 1}} \qquad \blacksquare

Part (b): Prove by induction that (x2+1)y(n+2)+(2n+1)xy(n+1)+n2y(n)=0(x^2 + 1) y^{(n+2)} + (2n+1) x\, y^{(n+1)} + n^2 y^{(n)} = 0

First, from part (a), we have (x2+1)1/2dydx=1(x^2 + 1)^{1/2} \frac{dy}{dx} = 1, so:

(x2+1) ⁣(dydx)2=1(*)(x^2 + 1)\!\left(\frac{dy}{dx}\right)^2 = 1 \qquad \text{(*)}

Differentiating (*) with respect to xx:

2x ⁣(dydx)2+2(x2+1)dydxd2ydx2=02x\!\left(\frac{dy}{dx}\right)^2 + 2(x^2 + 1)\frac{dy}{dx}\frac{d^2y}{dx^2} = 0

Since dydx0\frac{dy}{dx} \neq 0:

xdydx+(x2+1)d2ydx2=0(**)x\frac{dy}{dx} + (x^2 + 1)\frac{d^2y}{dx^2} = 0 \qquad \text{(**)}

This is the case n=0n = 0 of the recurrence (since y(0)=yy^{(0)} = y does not appear, the n=0n=0 case reduces to (x2+1)y+xy=0(x^2+1)y'' + xy' = 0, which is (**)).

Base case n=0n = 0. We have just shown (x2+1)y(2)+xy(1)=0(x^2 + 1)y^{(2)} + x\,y^{(1)} = 0. (The term 02y(0)=00^2 \cdot y^{(0)} = 0 vanishes.)

Inductive step. Assume the result holds for some n0n \geqslant 0:

(x2+1)y(n+2)+(2n+1)xy(n+1)+n2y(n)=0(IH)(x^2 + 1) y^{(n+2)} + (2n+1) x\, y^{(n+1)} + n^2 y^{(n)} = 0 \qquad \text{(IH)}

Differentiate (IH) with respect to xx:

2xy(n+2)+(x2+1)y(n+3)+(2n+1)y(n+1)+(2n+1)xy(n+2)+n2y(n+1)=02x\, y^{(n+2)} + (x^2 + 1) y^{(n+3)} + (2n+1) y^{(n+1)} + (2n+1) x\, y^{(n+2)} + n^2 y^{(n+1)} = 0

Collecting terms:

(x2+1)y(n+3)+[2x+(2n+1)x]y(n+2)+[(2n+1)+n2]y(n+1)=0(x^2 + 1) y^{(n+3)} + [2x + (2n+1)x]\, y^{(n+2)} + [(2n+1) + n^2]\, y^{(n+1)} = 0

(x2+1)y(n+3)+(2n+3)xy(n+2)+(n+1)2y(n+1)=0(x^2 + 1) y^{(n+3)} + (2n+3) x\, y^{(n+2)} + (n+1)^2 y^{(n+1)} = 0

This is exactly the recurrence with nn replaced by n+1n + 1. By induction, the result holds for all n0n \geqslant 0. \blacksquare

Part (c): Maclaurin series

Setting x=0x = 0 in the recurrence:

(0+1)y(n+2)(0)+0+n2y(n)(0)=0(0 + 1) y^{(n+2)}(0) + 0 + n^2 y^{(n)}(0) = 0

y(n+2)(0)=n2y(n)(0)(***)y^{(n+2)}(0) = -n^2\, y^{(n)}(0) \qquad \text{(***)}

We compute the derivatives at x=0x = 0:

  • y(0)=ln(0+1)=0y(0) = \ln(0 + 1) = 0, so y(0)(0)=0y^{(0)}(0) = 0.
  • y(0)=10+1=1y'(0) = \frac{1}{\sqrt{0+1}} = 1.
  • From (***): y(0)=02y(0)=0y''(0) = -0^2 \cdot y(0) = 0.
  • y(3)(0)=12y(0)=1y^{(3)}(0) = -1^2 \cdot y'(0) = -1.
  • y(4)(0)=22y(0)=0y^{(4)}(0) = -2^2 \cdot y''(0) = 0.
  • y(5)(0)=32y(3)(0)=9(1)=9y^{(5)}(0) = -3^2 \cdot y^{(3)}(0) = -9 \cdot (-1) = 9.
  • y(6)(0)=42y(4)(0)=0y^{(6)}(0) = -4^2 \cdot y^{(4)}(0) = 0.
  • y(7)(0)=52y(5)(0)=259=225y^{(7)}(0) = -5^2 \cdot y^{(5)}(0) = -25 \cdot 9 = -225.

The Maclaurin series is:

y=k=0y(k)(0)k!xk=xx33!+9x55!225x77!+y = \sum_{k=0}^{\infty} \frac{y^{(k)}(0)}{k!} x^k = x - \frac{x^3}{3!} + \frac{9 x^5}{5!} - \frac{225 x^7}{7!} + \cdots

=xx36+9x5120225x75040+= x - \frac{x^3}{6} + \frac{9x^5}{120} - \frac{225 x^7}{5040} + \cdots

=xx36+3x5405x7112+= x - \frac{x^3}{6} + \frac{3x^5}{40} - \frac{5x^7}{112} + \cdots

The first three terms are xx36+3x540x - \frac{x^3}{6} + \frac{3x^5}{40} as required, and the next non-zero term is 5x7112-\dfrac{5x^7}{112}.


Topic: 微积分 Calculus  |  Difficulty: Challenging  |  Marks: 20

2 Show that cosh1x=ln(x+(x21))\cosh^{-1} x = \ln (x + \sqrt{(x^2 - 1)}).

Show that the area of the region defined by the inequalities y2x28y^2 \geqslant x^2 - 8 and x225y216x^2 \geqslant 25y^2 - 16 is (72/5)ln2(72/5) \ln 2.

Model Solution

Part (a): Show that cosh1x=ln ⁣(x+x21)\cosh^{-1} x = \ln\!\left(x + \sqrt{x^2 - 1}\right)

Let y=cosh1xy = \cosh^{-1} x, so x=coshyx = \cosh y with y0y \geqslant 0.

From coshy=ey+ey2=x\cosh y = \frac{e^y + e^{-y}}{2} = x, multiply by 2ey2e^y:

e2y2xey+1=0e^{2y} - 2x\, e^y + 1 = 0

This is quadratic in eye^y:

ey=2x±4x242=x±x21e^y = \frac{2x \pm \sqrt{4x^2 - 4}}{2} = x \pm \sqrt{x^2 - 1}

Since y0y \geqslant 0, we need ey1e^y \geqslant 1. Both roots are positive (product = 1), so x+x211x + \sqrt{x^2 - 1} \geqslant 1 and xx211x - \sqrt{x^2 - 1} \leqslant 1 (since their product is 1). Therefore ey=x+x21e^y = x + \sqrt{x^2 - 1}, giving:

cosh1x=ln ⁣(x+x21)\cosh^{-1} x = \ln\!\left(x + \sqrt{x^2 - 1}\right) \qquad \blacksquare

Part (b): Area of the region

The region is defined by y2x28y^2 \geqslant x^2 - 8 and x225y216x^2 \geqslant 25y^2 - 16.

Rewrite: y2x28y^2 \geqslant x^2 - 8 means yx28y \geqslant \sqrt{x^2 - 8} (taking the first-quadrant part, then doubling by symmetry).

And x225y216x^2 \geqslant 25y^2 - 16 means 25y2x2+1625y^2 \leqslant x^2 + 16, i.e., y15x2+16y \leqslant \frac{1}{5}\sqrt{x^2 + 16}.

The region in the upper half-plane is bounded below by y=x28y = \sqrt{x^2 - 8} and above by y=15x2+16y = \frac{1}{5}\sqrt{x^2 + 16}.

Find intersection points: x28=15x2+16\sqrt{x^2 - 8} = \frac{1}{5}\sqrt{x^2 + 16}.

Squaring: x28=x2+1625x^2 - 8 = \frac{x^2 + 16}{25}, so 25x2200=x2+1625x^2 - 200 = x^2 + 16, giving 24x2=21624x^2 = 216, hence x2=9x^2 = 9, so x=±3x = \pm 3.

The region exists for x8|x| \geqslant \sqrt{8} (where y=x28y = \sqrt{x^2 - 8} is real), and the curves meet at x=±3x = \pm 3. For x>3|x| > 3, the lower curve x28\sqrt{x^2 - 8} exceeds the upper curve 15x2+16\frac{1}{5}\sqrt{x^2 + 16}, so the region is confined to 8x3\sqrt{8} \leqslant |x| \leqslant 3.

Wait — let me check. At x=3x = 3: 98=1\sqrt{9-8} = 1 and 159+16=55=1\frac{1}{5}\sqrt{9+16} = \frac{5}{5} = 1. At x=8x = \sqrt{8}: lower =0= 0, upper =158+16=245>0= \frac{1}{5}\sqrt{8+16} = \frac{\sqrt{24}}{5} > 0. So the lower curve starts at 0 when x=8x = \sqrt{8} and rises, while the upper curve starts positive and decreases slightly. The region is 8x3\sqrt{8} \leqslant |x| \leqslant 3 with upper curve above lower curve.

By symmetry, the total area is:

A=283[15x2+16x28]dxA = 2 \int_{\sqrt{8}}^{3} \left[\frac{1}{5}\sqrt{x^2 + 16} - \sqrt{x^2 - 8}\right] dx

Wait — actually we need to check whether the region also includes a part near the origin. For x<8|x| < \sqrt{8}, the condition y2x28y^2 \geqslant x^2 - 8 is automatically satisfied (since x28<0x^2 - 8 < 0 and y20y^2 \geqslant 0), so the lower bound is y=0y = 0. The upper bound is y=15x2+16y = \frac{1}{5}\sqrt{x^2+16}, which is positive. So the region also includes the area 0x80 \leqslant |x| \leqslant \sqrt{8} bounded by y=0y = 0 and y=15x2+16y = \frac{1}{5}\sqrt{x^2+16}.

Hmm, but the region is the set of points satisfying BOTH inequalities. Let me re-examine.

The region is {(x,y):y2x28 and x225y216}\{(x,y) : y^2 \geqslant x^2 - 8 \text{ and } x^2 \geqslant 25y^2 - 16\}.

Case 1: x<8|x| < \sqrt{8}. Then x28<0x^2 - 8 < 0, so y2x28y^2 \geqslant x^2 - 8 is always true. The second condition gives y15x2+16|y| \leqslant \frac{1}{5}\sqrt{x^2+16}. So the region includes a band around the xx-axis.

Case 2: x8|x| \geqslant \sqrt{8}. Then yx28|y| \geqslant \sqrt{x^2 - 8} and y15x2+16|y| \leqslant \frac{1}{5}\sqrt{x^2+16}.

These are compatible only when x2815x2+16\sqrt{x^2-8} \leqslant \frac{1}{5}\sqrt{x^2+16}, i.e., x3|x| \leqslant 3.

So in the upper half-plane, the region is:

  • For 0x80 \leqslant x \leqslant \sqrt{8}: 0y15x2+160 \leqslant y \leqslant \frac{1}{5}\sqrt{x^2+16}.
  • For 8x3\sqrt{8} \leqslant x \leqslant 3: x28y15x2+16\sqrt{x^2-8} \leqslant y \leqslant \frac{1}{5}\sqrt{x^2+16}.

And symmetric in the lower half-plane.

Total area =2×= 2 \times (area in upper half-plane).

A2=0815x2+16dx+83[15x2+16x28]dx\frac{A}{2} = \int_0^{\sqrt{8}} \frac{1}{5}\sqrt{x^2+16}\, dx + \int_{\sqrt{8}}^{3} \left[\frac{1}{5}\sqrt{x^2+16} - \sqrt{x^2-8}\right] dx

=1503x2+16dx83x28dx= \frac{1}{5}\int_0^{3} \sqrt{x^2+16}\, dx - \int_{\sqrt{8}}^{3} \sqrt{x^2-8}\, dx

Integral 1: x2+16dx\int \sqrt{x^2 + 16}\, dx.

Using the standard formula x2+a2dx=x2x2+a2+a22ln ⁣(x+x2+a2)+C\int \sqrt{x^2 + a^2}\, dx = \frac{x}{2}\sqrt{x^2+a^2} + \frac{a^2}{2}\ln\!\left(x + \sqrt{x^2+a^2}\right) + C:

03x2+16dx=[x2x2+16+8ln ⁣(x+x2+16)]03\int_0^3 \sqrt{x^2+16}\, dx = \left[\frac{x}{2}\sqrt{x^2+16} + 8\ln\!\left(x + \sqrt{x^2+16}\right)\right]_0^3

=3225+8ln(3+5)08ln4= \frac{3}{2}\sqrt{25} + 8\ln(3+5) - 0 - 8\ln 4

=152+8ln88ln4=152+8ln2= \frac{15}{2} + 8\ln 8 - 8\ln 4 = \frac{15}{2} + 8\ln 2

Integral 2: x28dx\int \sqrt{x^2 - 8}\, dx.

Using x2a2dx=x2x2a2a22ln ⁣x+x2a2+C\int \sqrt{x^2 - a^2}\, dx = \frac{x}{2}\sqrt{x^2-a^2} - \frac{a^2}{2}\ln\!\left|x + \sqrt{x^2-a^2}\right| + C:

83x28dx=[x2x284ln ⁣(x+x28)]83\int_{\sqrt{8}}^{3} \sqrt{x^2-8}\, dx = \left[\frac{x}{2}\sqrt{x^2-8} - 4\ln\!\left(x + \sqrt{x^2-8}\right)\right]_{\sqrt{8}}^{3}

=3214ln(3+1)0+4ln(8)= \frac{3}{2}\sqrt{1} - 4\ln(3+1) - 0 + 4\ln(\sqrt{8})

=324ln4+4ln8=324ln4+432ln2= \frac{3}{2} - 4\ln 4 + 4\ln\sqrt{8} = \frac{3}{2} - 4\ln 4 + 4 \cdot \frac{3}{2}\ln 2

=328ln2+6ln2=322ln2= \frac{3}{2} - 8\ln 2 + 6\ln 2 = \frac{3}{2} - 2\ln 2

Total area:

A2=15 ⁣(152+8ln2)(322ln2)\frac{A}{2} = \frac{1}{5}\!\left(\frac{15}{2} + 8\ln 2\right) - \left(\frac{3}{2} - 2\ln 2\right)

=32+85ln232+2ln2=85ln2+105ln2=185ln2= \frac{3}{2} + \frac{8}{5}\ln 2 - \frac{3}{2} + 2\ln 2 = \frac{8}{5}\ln 2 + \frac{10}{5}\ln 2 = \frac{18}{5}\ln 2

A=2185ln2=365ln2A = 2 \cdot \frac{18}{5}\ln 2 = \frac{36}{5}\ln 2

Hmm, the expected answer is 725ln2\frac{72}{5}\ln 2. Let me recheck.

I think I need to reconsider the region. The inequalities are y2x28y^2 \geqslant x^2 - 8 (exterior of hyperbola) and x225y216x^2 \geqslant 25y^2 - 16 (interior of another hyperbola). The region is the intersection.

Actually, I think the region might also extend into the yy-direction more broadly. Let me reconsider.

The second inequality x225y216x^2 \geqslant 25y^2 - 16 means 25y2x21625y^2 - x^2 \leqslant 16, i.e., y216/25x2161\frac{y^2}{16/25} - \frac{x^2}{16} \leqslant 1. This is the interior of a hyperbola opening in the yy-direction.

The first inequality y2x28y^2 \geqslant x^2 - 8 means y2x28y^2 - x^2 \geqslant -8, i.e., x2y28x^2 - y^2 \leqslant 8. This is the interior of a hyperbola opening in the xx-direction.

The region is the intersection: inside both hyperbolas. This is a bounded region.

In the first quadrant, the boundary of x2y2=8x^2 - y^2 = 8 is y=x28y = \sqrt{x^2 - 8} for x8x \geqslant \sqrt{8} (above this curve means y2>x28y^2 > x^2 - 8, i.e., inside). The boundary of 25y2x2=1625y^2 - x^2 = 16 is y=15x2+16y = \frac{1}{5}\sqrt{x^2 + 16} (below this curve means inside).

So in the first quadrant: the region is below y=15x2+16y = \frac{1}{5}\sqrt{x^2+16} and above y=x28y = \sqrt{x^2-8} when x8x \geqslant \sqrt{8}, or above y=0y = 0 when x<8x < \sqrt{8}.

But wait, I also need to check: is the region also above y=15x2+16y = \frac{1}{5}\sqrt{x^2+16} somewhere? No — the inequality is 25y2x2+1625y^2 \leqslant x^2 + 16, so y15x2+16y \leqslant \frac{1}{5}\sqrt{x^2+16}.

Hmm, but the region could also include points where y>15x2+16y > \frac{1}{5}\sqrt{x^2+16} as long as y2<0y^2 < 0… no, that doesn’t make sense.

Actually, let me reconsider. The region y2x28y^2 \geqslant x^2 - 8 includes all points outside or on the hyperbola y2=x28y^2 = x^2 - 8. The region x225y216x^2 \geqslant 25y^2 - 16 includes all points inside or on the hyperbola 25y2x2=1625y^2 - x^2 = 16.

The intersection is a bounded “lens” region. In the first quadrant, I had the right picture. Let me recheck my area calculation.

Actually, I realize the region in the upper half-plane also needs to account for negative yy values more carefully. By symmetry about the xx-axis, the total area is twice the upper half. But I also need to check symmetry about the yy-axis.

Both hyperbolas are symmetric about both axes. So the total area is 4 times the area in the first quadrant.

A=4[0815x2+16dx+83(15x2+16x28)dx]A = 4\left[\int_0^{\sqrt{8}} \frac{1}{5}\sqrt{x^2+16}\, dx + \int_{\sqrt{8}}^{3} \left(\frac{1}{5}\sqrt{x^2+16} - \sqrt{x^2-8}\right) dx\right]

=4[1503x2+16dx83x28dx]= 4\left[\frac{1}{5}\int_0^3 \sqrt{x^2+16}\, dx - \int_{\sqrt{8}}^3 \sqrt{x^2-8}\, dx\right]

=4[15 ⁣(152+8ln2)(322ln2)]= 4\left[\frac{1}{5}\!\left(\frac{15}{2}+8\ln 2\right) - \left(\frac{3}{2}-2\ln 2\right)\right]

=4[32+85ln232+2ln2]=4185ln2=725ln2= 4\left[\frac{3}{2}+\frac{8}{5}\ln 2 - \frac{3}{2}+2\ln 2\right] = 4 \cdot \frac{18}{5}\ln 2 = \frac{72}{5}\ln 2 \qquad \blacksquare


Topic: 代数与方程 Algebra and Equations  |  Difficulty: Standard  |  Marks: 20

3 Consider the equation

x2bx+c=0,x^2 - bx + c = 0 ,

where bb and cc are real numbers.

(i) Show that the roots of the equation are real and positive if and only if b>0b > 0 and b24c>0b^2 \geqslant 4c > 0, and sketch the region of the bb-cc plane in which these conditions hold.

(ii) Sketch the region of the bb-cc plane in which the roots of the equation are real and less than 1 in magnitude.

Model Solution

Part (i)

Let the roots be α\alpha and β\beta. By Vieta’s formulas:

α+β=b(1)\alpha + \beta = b \qquad \text{(1)}

αβ=c(2)\alpha \beta = c \qquad \text{(2)}

We prove both directions of the “if and only if”.

Forward direction. Suppose both roots are real and positive.

Since α>0\alpha > 0 and β>0\beta > 0:

  • From (1): b=α+β>0b = \alpha + \beta > 0.
  • From (2): c=αβ>0c = \alpha \beta > 0.
  • Real roots require discriminant Δ=b24c0\Delta = b^2 - 4c \geqslant 0, i.e., b24cb^2 \geqslant 4c.

Combining: b>0b > 0 and b24c>0b^2 \geqslant 4c > 0.

Reverse direction. Suppose b>0b > 0 and b24c>0b^2 \geqslant 4c > 0.

  • b24cb^2 \geqslant 4c means Δ0\Delta \geqslant 0, so the roots are real.
  • 4c>04c > 0 gives c>0c > 0, so αβ=c>0\alpha \beta = c > 0: the roots have the same sign.
  • b>0b > 0 gives α+β=b>0\alpha + \beta = b > 0: the roots are not both negative.

Since the roots have the same sign and their sum is positive, both roots are positive.

This completes the proof. \blacksquare

Sketch of the region. In the bb-cc plane, the conditions are:

  1. b>0b > 0 (right of the cc-axis, not including it),
  2. c>0c > 0 (above the bb-axis, not including it),
  3. cb24c \leqslant \frac{b^2}{4} (on or below the parabola c=b24c = \frac{b^2}{4}).

The region is bounded below by the positive bb-axis, on the left by the cc-axis (excluded), and above by the parabola c=b24c = \frac{b^2}{4} (included). The parabola passes through the origin and opens upward. The boundary c=0c = 0 (the bb-axis) is excluded since 4c>04c > 0, while the parabola c=b24c = \frac{b^2}{4} is included.

Part (ii)

We need the roots to satisfy 1<α,β<1-1 < \alpha, \beta < 1, i.e., both roots lie strictly between 1-1 and 11.

Since the coefficient of x2x^2 is positive, the parabola y=x2bx+cy = x^2 - bx + c opens upward. Both roots lie in (1,1)(-1, 1) if and only if:

(a) f(1)>0f(1) > 0: the value at x=1x = 1 is positive (so x=1x = 1 is to the right of both roots),

(b) f(1)>0f(-1) > 0: the value at x=1x = -1 is positive (so x=1x = -1 is to the left of both roots),

(c) Δ0\Delta \geqslant 0: the roots are real.

Evaluating:

f(1)=1b+c>0    c>b1(a)f(1) = 1 - b + c > 0 \implies c > b - 1 \qquad \text{(a)}

f(1)=1+b+c>0    c>b1(b)f(-1) = 1 + b + c > 0 \implies c > -b - 1 \qquad \text{(b)}

b24c(c)b^2 \geqslant 4c \qquad \text{(c)}

Note that (a) and (b) together with (c) automatically ensure b<2|b| < 2, since from (c): 4cb24c \leqslant b^2, and from (a): c>b1c > b - 1, so 4(b1)<b24(b-1) < b^2, giving b24b+4>0b^2 - 4b + 4 > 0, i.e., (b2)2>0(b-2)^2 > 0, which holds for b2b \neq 2. A symmetric argument using (b) gives b2b \neq -2. (At b=±2b = \pm 2, the conditions force c=1c = 1, giving a double root at x=±1x = \pm 1, which is excluded since we need x<1|x| < 1.)

Sketch of the region. The boundary curves are:

  • c=b1c = b - 1: a line with gradient 1 and cc-intercept 1-1.
  • c=b1c = -b - 1: a line with gradient 1-1 and cc-intercept 1-1.
  • c=b24c = \frac{b^2}{4}: the parabola from part (i).

The two lines meet at (0,1)(0, -1), forming a V-shape opening upward. The parabola intersects c=b1c = b - 1 at b=2b = 2 (giving the point (2,1)(2, 1)) and intersects c=b1c = -b - 1 at b=2b = -2 (giving the point (2,1)(-2, 1)).

The region is:

  • Above both lines: c>b1c > b - 1 and c>b1c > -b - 1, equivalently c>b1c > |b| - 1.
  • Below or on the parabola: cb24c \leqslant \frac{b^2}{4}.

This region lies between b=2b = -2 and b=2b = 2, bounded below by the V-shaped pair of lines (excluded) and above by the parabola (included). The region has a roughly lens-shaped appearance, widest near b=0b = 0 where the gap between the parabola (c=0c = 0) and the lines (c=1c = -1) is greatest.


Topic: 微积分 Calculus  |  Difficulty: Challenging  |  Marks: 20

4 In this question, the function sin1\sin^{-1} is defined to have domain 1x1-1 \leqslant x \leqslant 1 and range π2xπ2-\frac{\pi}{2} \leqslant x \leqslant \frac{\pi}{2} and the function tan1\tan^{-1} is defined to have the real numbers as its domain and range π2<x<π2-\frac{\pi}{2} < x < \frac{\pi}{2}.

(i) Let

g(x)=2x1+x2 , <x< .g(x) = \frac{2x}{1 + x^2} \text{ , } \quad -\infty < x < \infty \text{ .}

Sketch the graph of g(x)g(x) and state the range of gg.

(ii) Let

f(x)=sin1(2x1+x2) , <x< .f(x) = \sin^{-1} \left( \frac{2x}{1 + x^2} \right) \text{ , } \quad -\infty < x < \infty \text{ .}

Show that f(x)=2tan1xf(x) = 2 \tan^{-1} x for 1x1-1 \leqslant x \leqslant 1 and f(x)=π2tan1xf(x) = \pi - 2 \tan^{-1} x for x1x \geqslant 1. Sketch the graph of f(x)f(x).

Model Solution

Part (i)

We analyse g(x)=2x1+x2g(x) = \frac{2x}{1 + x^2}.

Derivative. Using the quotient rule:

g(x)=2(1+x2)2x2x(1+x2)2=2+2x24x2(1+x2)2=2(1x2)(1+x2)2g'(x) = \frac{2(1 + x^2) - 2x \cdot 2x}{(1 + x^2)^2} = \frac{2 + 2x^2 - 4x^2}{(1 + x^2)^2} = \frac{2(1 - x^2)}{(1 + x^2)^2}

Setting g(x)=0g'(x) = 0: 1x2=01 - x^2 = 0, so x=±1x = \pm 1.

  • g(x)>0g'(x) > 0 for x<1|x| < 1: gg is increasing on (1,1)(-1, 1).
  • g(x)<0g'(x) < 0 for x>1|x| > 1: gg is decreasing on (,1)(-\infty, -1) and (1,)(1, \infty).

Key values. g(1)=22=1g(-1) = \frac{-2}{2} = -1 (local minimum), g(1)=22=1g(1) = \frac{2}{2} = 1 (local maximum), g(0)=0g(0) = 0.

Behaviour at infinity. g(x)0+g(x) \to 0^+ as x+x \to +\infty, and g(x)0g(x) \to 0^- as xx \to -\infty.

Symmetry. g(x)=2x1+x2=g(x)g(-x) = \frac{-2x}{1 + x^2} = -g(x), so gg is an odd function (symmetric about the origin).

Graph. The curve passes through the origin, rises to a maximum of 11 at x=1x = 1, then decays toward 00 as xx \to \infty. By odd symmetry, it falls to a minimum of 1-1 at x=1x = -1 and approaches 00 from below as xx \to -\infty. The xx-axis is a horizontal asymptote.

Range. The maximum value is g(1)=1g(1) = 1 and the minimum value is g(1)=1g(-1) = -1. Since gg is continuous and achieves all values between 1-1 and 11:

range of g=[1,1].\text{range of } g = [-1, 1].

Part (ii)

Let θ=tan1x\theta = \tan^{-1} x, so x=tanθx = \tan\theta with θ(π2,π2)\theta \in \left(-\frac{\pi}{2}, \frac{\pi}{2}\right).

Then:

2x1+x2=2tanθ1+tan2θ=2tanθsec2θ=2sinθcosθ=sin2θ\frac{2x}{1 + x^2} = \frac{2\tan\theta}{1 + \tan^2\theta} = \frac{2\tan\theta}{\sec^2\theta} = 2\sin\theta\cos\theta = \sin 2\theta

So f(x)=sin1(sin2θ)f(x) = \sin^{-1}(\sin 2\theta).

We evaluate sin1(sin2θ)\sin^{-1}(\sin 2\theta) using the range of sin1\sin^{-1}, which is [π2,π2]\left[-\frac{\pi}{2}, \frac{\pi}{2}\right].

Case 1: 1x1-1 \leqslant x \leqslant 1.

Here θ=tan1x[π4,π4]\theta = \tan^{-1} x \in \left[-\frac{\pi}{4}, \frac{\pi}{4}\right], so 2θ[π2,π2]2\theta \in \left[-\frac{\pi}{2}, \frac{\pi}{2}\right].

Since 2θ2\theta lies in the range of sin1\sin^{-1}:

f(x)=sin1(sin2θ)=2θ=2tan1xf(x) = \sin^{-1}(\sin 2\theta) = 2\theta = 2\tan^{-1} x

Case 2: x1x \geqslant 1.

Here θ=tan1x[π4,π2)\theta = \tan^{-1} x \in \left[\frac{\pi}{4}, \frac{\pi}{2}\right), so 2θ[π2,π)2\theta \in \left[\frac{\pi}{2}, \pi\right).

For 2θ[π2,π)2\theta \in \left[\frac{\pi}{2}, \pi\right), we use the identity sin(π2θ)=sin2θ\sin(\pi - 2\theta) = \sin 2\theta and note that π2θ(0,π2][π2,π2]\pi - 2\theta \in \left(0, \frac{\pi}{2}\right] \subset \left[-\frac{\pi}{2}, \frac{\pi}{2}\right]:

f(x)=sin1(sin2θ)=π2θ=π2tan1xf(x) = \sin^{-1}(\sin 2\theta) = \pi - 2\theta = \pi - 2\tan^{-1} x

Case 3: x1x \leqslant -1 (by symmetry).

Since gg is odd, we have f(x)=sin1(g(x))=sin1(g(x))=f(x)f(-x) = \sin^{-1}(-g(x)) = -\sin^{-1}(g(x)) = -f(x), so ff is odd.

For x1x \leqslant -1, using f(x)=f(x)f(x) = -f(-x) and Case 2 (since x1-x \geqslant 1):

f(x)=[π2tan1(x)]=π+2tan1(x)=π2tan1xf(x) = -[\pi - 2\tan^{-1}(-x)] = -\pi + 2\tan^{-1}(-x) = -\pi - 2\tan^{-1} x

This completes the derivation. \blacksquare

Sketch of f(x)f(x).

The piecewise definition is:

f(x)={π2tan1xx12tan1x1x1π2tan1xx1f(x) = \begin{cases} -\pi - 2\tan^{-1} x & x \leqslant -1 \\ 2\tan^{-1} x & -1 \leqslant x \leqslant 1 \\ \pi - 2\tan^{-1} x & x \geqslant 1 \end{cases}

Key features:

  • ff is continuous everywhere (check at x=±1x = \pm 1: f(1)=π2f(1) = \frac{\pi}{2}, f(1)=π2f(-1) = -\frac{\pi}{2}).
  • ff is an odd function: symmetric about the origin.
  • On [1,1][-1, 1], ff increases from π2-\frac{\pi}{2} to π2\frac{\pi}{2} with an inflection point at the origin (this portion is 2tan1x2\tan^{-1} x).
  • For x1x \geqslant 1, ff decreases from π2\frac{\pi}{2} toward 00 as xx \to \infty (horizontal asymptote y=0y = 0).
  • For x1x \leqslant -1, ff increases from π2-\frac{\pi}{2} toward 00 as xx \to -\infty (horizontal asymptote y=0y = 0).
  • Not differentiable at x=±1x = \pm 1 (the left and right derivatives differ).

The graph looks like a stretched S-curve on [1,1][-1, 1] that flattens out to the xx-axis on both sides.


Topic: 微积分 Calculus  |  Difficulty: Hard  |  Marks: 20

5 Show that the equation x3+px+q=0x^3 + px + q = 0 has exactly one real solution if p0p \geqslant 0.

A parabola CC is given parametrically by

x=at2,y=2at(a>0) .x = at^2, \quad y = 2at \quad (a > 0) \text{ .}

Find an equation which must be satisfied by tt at points on CC at which the normal passes through the point (h,k)(h, k). Hence show that, if h2ah \leqslant 2a, exactly one normal to CC will pass through (h,k)(h, k).

Find, in Cartesian form, the equation of the locus of the points from which exactly two normals can be drawn to CC. Sketch the locus.

Model Solution

Part (a): Show that x3+px+q=0x^3 + px + q = 0 has exactly one real solution if p0p \geqslant 0

Let f(x)=x3+px+qf(x) = x^3 + px + q.

f(x)=3x2+pf'(x) = 3x^2 + p

Since p0p \geqslant 0 and 3x203x^2 \geqslant 0 for all xx, we have f(x)0f'(x) \geqslant 0 for all xx. Moreover, f(x)=0f'(x) = 0 only at x=0x = 0 (when p=0p = 0), so ff is strictly increasing.

A strictly increasing continuous function can cross the xx-axis at most once. Since f(x)+f(x) \to +\infty as x+x \to +\infty and f(x)f(x) \to -\infty as xx \to -\infty, by the intermediate value theorem, ff has exactly one real root. \blacksquare

Part (b): Find the equation for tt at normal points through (h,k)(h, k)

For the parabola x=at2x = at^2, y=2aty = 2at:

dydx=dy/dtdx/dt=2a2at=1t\frac{dy}{dx} = \frac{dy/dt}{dx/dt} = \frac{2a}{2at} = \frac{1}{t}

The slope of the normal at parameter tt is t-t. The normal at (at2,2at)(at^2, 2at) is:

y2at=t(xat2)y - 2at = -t(x - at^2)

For this to pass through (h,k)(h, k):

k2at=t(hat2)k - 2at = -t(h - at^2)

k2at=th+at3k - 2at = -th + at^3

at3+(2ah)tk=0(*)at^3 + (2a - h)t - k = 0 \qquad \text{(*)}

This is the required equation. \blacksquare

Part (c): If h2ah \leqslant 2a, exactly one normal passes through (h,k)(h, k)

Equation (*) can be written as t3+2ahatka=0t^3 + \frac{2a - h}{a}\,t - \frac{k}{a} = 0.

Since h2ah \leqslant 2a, we have 2ah02a - h \geqslant 0, so the coefficient 2aha0\frac{2a - h}{a} \geqslant 0 (using a>0a > 0).

By part (a), this cubic in tt has exactly one real solution. Each real value of tt gives a distinct point on the parabola, so exactly one normal to CC passes through (h,k)(h, k). \blacksquare

Part (d): Locus of points from which exactly two normals can be drawn

Exactly two normals means the cubic (*) has exactly two distinct real roots, i.e., one root is repeated. Let the roots be s,s,t3s, s, t_3 with st3s \neq t_3.

From Vieta’s formulas for t3+2ahatka=0t^3 + \frac{2a-h}{a}\,t - \frac{k}{a} = 0:

2s+t3=0    t3=2s(1)2s + t_3 = 0 \implies t_3 = -2s \qquad \text{(1)}

s2+2st3=2aha(2)s^2 + 2st_3 = \frac{2a - h}{a} \qquad \text{(2)}

s2t3=ka(3)s^2 t_3 = \frac{k}{a} \qquad \text{(3)}

From (1) and (3):

s2(2s)=ka    k=2as3s^2(-2s) = \frac{k}{a} \implies k = -2as^3

From (1) and (2):

s2+2s(2s)=2aha    3s2=2ahas^2 + 2s(-2s) = \frac{2a - h}{a} \implies -3s^2 = \frac{2a - h}{a}

h=2a+3as2=a(2+3s2)h = 2a + 3as^2 = a(2 + 3s^2)

To eliminate ss: from k=2as3k = -2as^3, we get s3=k2as^3 = -\frac{k}{2a}, so s6=k24a2s^6 = \frac{k^2}{4a^2}.

From h=a(2+3s2)h = a(2 + 3s^2): s2=h2a3as^2 = \frac{h - 2a}{3a}, so s6=(h2a)327a3s^6 = \frac{(h-2a)^3}{27a^3}.

Equating:

k24a2=(h2a)327a3\frac{k^2}{4a^2} = \frac{(h - 2a)^3}{27a^3}

27ak2=4(h2a)327ak^2 = 4(h - 2a)^3

This is a semicubical parabola (Neile’s parabola) with a cusp at (2a,0)(2a, 0). It exists only for h2ah \geqslant 2a (since s20s^2 \geqslant 0 requires h2ah \geqslant 2a).

The locus is the curve:

27ak2=4(h2a)3\boxed{27ak^2 = 4(h - 2a)^3}

This is the evolute of the parabola CC. The cusp is at (2a,0)(2a, 0), which is the focus of the parabola. The curve extends to the right (for h>2ah > 2a), and the region inside this evolute is where three normals can be drawn to CC.


Topic: 立体几何 / 3D Geometry  |  Difficulty: Challenging  |  Marks: 20

6 The plane

xa+yb+zc=1\frac{x}{a} + \frac{y}{b} + \frac{z}{c} = 1

meets the co-ordinate axes at the points A,BA, B and CC. The point MM has coordinates (12a,12b,12c)(\frac{1}{2}a, \frac{1}{2}b, \frac{1}{2}c) and OO is the origin.

Show that OMOM meets the plane at the centroid (13a,13b,13c)(\frac{1}{3}a, \frac{1}{3}b, \frac{1}{3}c) of triangle ABCABC. Show also that the perpendiculars to the plane from OO and from MM meet the plane at the orthocentre and at the circumcentre of triangle ABCABC respectively.

Hence prove that the centroid of a triangle lies on the line segment joining its orthocentre and circumcentre, and that it divides this line segment in the ratio 2:12 : 1.

[The orthocentre of a triangle is the point at which the three altitudes intersect; the circumcentre of a triangle is the point equidistant from the three vertices.]

Model Solution

The plane xa+yb+zc=1\frac{x}{a} + \frac{y}{b} + \frac{z}{c} = 1 meets the coordinate axes at A=(a,0,0)A = (a, 0, 0), B=(0,b,0)B = (0, b, 0), C=(0,0,c)C = (0, 0, c).

Part (a): OMOM meets the plane at the centroid of ABC\triangle ABC

The line from O=(0,0,0)O = (0,0,0) through M=(12a,12b,12c)M = (\frac{1}{2}a, \frac{1}{2}b, \frac{1}{2}c) is:

(x,y,z)=t ⁣(12a,12b,12c)=(ta2,tb2,tc2)(x, y, z) = t\!\left(\tfrac{1}{2}a, \tfrac{1}{2}b, \tfrac{1}{2}c\right) = \left(\tfrac{ta}{2}, \tfrac{tb}{2}, \tfrac{tc}{2}\right)

Substituting into the plane equation:

ta/2a+tb/2b+tc/2c=1    3t2=1    t=23\frac{ta/2}{a} + \frac{tb/2}{b} + \frac{tc/2}{c} = 1 \implies \frac{3t}{2} = 1 \implies t = \frac{2}{3}

The intersection point is (a3,b3,c3)\left(\frac{a}{3}, \frac{b}{3}, \frac{c}{3}\right).

The centroid of ABC\triangle ABC is G=A+B+C3=(a3,b3,c3)G = \frac{A + B + C}{3} = \left(\frac{a}{3}, \frac{b}{3}, \frac{c}{3}\right).

These are the same point. \blacksquare

Part (b): The perpendicular from OO meets the plane at the orthocentre

The normal to the plane is n=(1a,1b,1c)\mathbf{n} = \left(\frac{1}{a}, \frac{1}{b}, \frac{1}{c}\right). The perpendicular from OO to the plane is:

(x,y,z)=t ⁣(1a,1b,1c)(x, y, z) = t\!\left(\frac{1}{a}, \frac{1}{b}, \frac{1}{c}\right)

Substituting into the plane equation:

ta2+tb2+tc2=1    t=11a2+1b2+1c2\frac{t}{a^2} + \frac{t}{b^2} + \frac{t}{c^2} = 1 \implies t = \frac{1}{\frac{1}{a^2} + \frac{1}{b^2} + \frac{1}{c^2}}

Let S=1a2+1b2+1c2S = \frac{1}{a^2} + \frac{1}{b^2} + \frac{1}{c^2}. The foot of the perpendicular is:

H=(1aS,1bS,1cS)H = \left(\frac{1}{aS}, \frac{1}{bS}, \frac{1}{cS}\right)

We now verify this is the orthocentre of ABC\triangle ABC by computing the feet of the altitudes.

The altitude from AA to BCBC: the foot PAP_A satisfies APABC=0\overrightarrow{AP_A} \cdot \overrightarrow{BC} = 0 and lies on BCBC. Parameterizing BCBC as B+λ(CB)=(0,b(1λ),cλ)B + \lambda(C - B) = (0, b(1-\lambda), c\lambda):

APA=(a,b(1λ),cλ),BC=(0,b,c)\overrightarrow{AP_A} = (-a, b(1-\lambda), c\lambda), \quad \overrightarrow{BC} = (0, -b, c)

APABC=b2(1λ)+c2λ=0    λ=b2b2+c2\overrightarrow{AP_A} \cdot \overrightarrow{BC} = -b^2(1-\lambda) + c^2\lambda = 0 \implies \lambda = \frac{b^2}{b^2 + c^2}

So PA=(0,bc2b2+c2,b2cb2+c2)P_A = \left(0, \frac{bc^2}{b^2+c^2}, \frac{b^2c}{b^2+c^2}\right).

The altitude from AA through PAP_A has direction APA=(a,bc2b2+c2,b2cb2+c2)\overrightarrow{AP_A} = \left(-a, \frac{bc^2}{b^2+c^2}, \frac{b^2c}{b^2+c^2}\right). The line is:

(x,y,z)=(a,0,0)+s ⁣(a,bc2b2+c2,b2cb2+c2)(x, y, z) = (a, 0, 0) + s\!\left(-a, \frac{bc^2}{b^2+c^2}, \frac{b^2c}{b^2+c^2}\right)

Let D=a2+b2+c2D = a^2 + b^2 + c^2. At s=a2Ds = \frac{a^2}{D}:

x=aa3D=a(b2+c2)D=aa2b2+c2a2Dx = a - \frac{a^3}{D} = \frac{a(b^2+c^2)}{D} = \frac{a \cdot a^2 \cdot \frac{b^2+c^2}{a^2}}{D}

Hmm, let me verify directly. We claim H=(a3D,b3D,c3D)H = \left(\frac{a^3}{D}, \frac{b^3}{D}, \frac{c^3}{D}\right) where D=a2+b2+c2D = a^2 + b^2 + c^2.

Note that 1aS=aa2S=aa2a2+a2b2+a2c2\frac{1}{aS} = \frac{a}{a^2 S} = \frac{a}{\frac{a^2}{a^2} + \frac{a^2}{b^2} + \frac{a^2}{c^2}}. Hmm, this doesn’t simplify directly.

Actually, let me verify HH is the orthocentre by checking AHBC=0\overrightarrow{AH} \cdot \overrightarrow{BC} = 0:

AH=(1aSa,1bS,1cS),BC=(0,b,c)\overrightarrow{AH} = \left(\frac{1}{aS} - a, \frac{1}{bS}, \frac{1}{cS}\right), \quad \overrightarrow{BC} = (0, -b, c)

AHBC=bbS+ccS=1S+1S=0\overrightarrow{AH} \cdot \overrightarrow{BC} = -\frac{b}{bS} + \frac{c}{cS} = -\frac{1}{S} + \frac{1}{S} = 0 \qquad \checkmark

By symmetry (cyclic permutation of a,b,ca, b, c):

BHCA=0\overrightarrow{BH} \cdot \overrightarrow{CA} = 0 \qquad \checkmark

CHAB=0\overrightarrow{CH} \cdot \overrightarrow{AB} = 0 \qquad \checkmark

So HH lies on all three altitudes, confirming it is the orthocentre. \blacksquare

Part (c): The perpendicular from MM meets the plane at the circumcentre

The perpendicular from M=(a2,b2,c2)M = (\frac{a}{2}, \frac{b}{2}, \frac{c}{2}) in direction n=(1a,1b,1c)\mathbf{n} = (\frac{1}{a}, \frac{1}{b}, \frac{1}{c}):

(x,y,z)=(a2+ta,b2+tb,c2+tc)(x, y, z) = \left(\frac{a}{2} + \frac{t}{a}, \frac{b}{2} + \frac{t}{b}, \frac{c}{2} + \frac{t}{c}\right)

Substituting into the plane:

12+ta2+12+tb2+12+tc2=1\frac{1}{2} + \frac{t}{a^2} + \frac{1}{2} + \frac{t}{b^2} + \frac{1}{2} + \frac{t}{c^2} = 1

32+tS=1    t=12S\frac{3}{2} + tS = 1 \implies t = -\frac{1}{2S}

The foot of the perpendicular is:

Oc=(a212aS,b212bS,c212cS)O_c = \left(\frac{a}{2} - \frac{1}{2aS}, \frac{b}{2} - \frac{1}{2bS}, \frac{c}{2} - \frac{1}{2cS}\right)

We verify this is equidistant from AA, BB, CC. Let α=1a2S\alpha = \frac{1}{a^2 S}, β=1b2S\beta = \frac{1}{b^2 S}, γ=1c2S\gamma = \frac{1}{c^2 S}, so α+β+γ=1\alpha + \beta + \gamma = 1.

Then Oc=(a(1α)2,b(1β)2,c(1γ)2)O_c = \left(\frac{a(1-\alpha)}{2}, \frac{b(1-\beta)}{2}, \frac{c(1-\gamma)}{2}\right).

Distance to A=(a,0,0)A = (a, 0, 0):

OcA2=a2(1+α)24+b2(1β)24+c2(1γ)24|O_c A|^2 = \frac{a^2(1+\alpha)^2}{4} + \frac{b^2(1-\beta)^2}{4} + \frac{c^2(1-\gamma)^2}{4}

Distance to B=(0,b,0)B = (0, b, 0):

OcB2=a2(1α)24+b2(1+β)24+c2(1γ)24|O_c B|^2 = \frac{a^2(1-\alpha)^2}{4} + \frac{b^2(1+\beta)^2}{4} + \frac{c^2(1-\gamma)^2}{4}

Distance to C=(0,0,c)C = (0, 0, c):

OcC2=a2(1α)24+b2(1β)24+c2(1+γ)24|O_c C|^2 = \frac{a^2(1-\alpha)^2}{4} + \frac{b^2(1-\beta)^2}{4} + \frac{c^2(1+\gamma)^2}{4}

Computing OcA2OcB2|O_c A|^2 - |O_c B|^2:

=a2[(1+α)2(1α)2]b2[(1+β)2(1β)2]4= \frac{a^2[(1+\alpha)^2 - (1-\alpha)^2] - b^2[(1+\beta)^2 - (1-\beta)^2]}{4}

=4a2α4b2β4=a2αb2β=1S1S=0= \frac{4a^2\alpha - 4b^2\beta}{4} = a^2\alpha - b^2\beta = \frac{1}{S} - \frac{1}{S} = 0

Similarly OcB2OcC2=0|O_c B|^2 - |O_c C|^2 = 0. So OcO_c is equidistant from all three vertices, confirming it is the circumcentre. \blacksquare

Part (d): The centroid divides the orthocentre—circumcentre segment in the ratio 2:12:1

The centroid, orthocentre, and circumcentre all lie in the plane of ABC\triangle ABC. We show G=2Oc+H3G = \frac{2O_c + H}{3}.

2Oc+H3=13(a(1α)+1aS,  b(1β)+1bS,  c(1γ)+1cS)\frac{2O_c + H}{3} = \frac{1}{3}\left(a(1-\alpha) + \frac{1}{aS}, \; b(1-\beta) + \frac{1}{bS}, \; c(1-\gamma) + \frac{1}{cS}\right)

Since 1aS=aα\frac{1}{aS} = a\alpha:

a(1α)+aα=aa(1-\alpha) + a\alpha = a

So the first component is a3\frac{a}{3}. By symmetry, the second and third components are b3\frac{b}{3} and c3\frac{c}{3}.

Therefore 2Oc+H3=(a3,b3,c3)=G\frac{2O_c + H}{3} = \left(\frac{a}{3}, \frac{b}{3}, \frac{c}{3}\right) = G.

This means GG lies on the segment HOcHO_c and divides it in the ratio HG:GOc=2:1HG : GO_c = 2 : 1 (with GG closer to OcO_c). \blacksquare


Topic: 微分方程 / Differential Equations  |  Difficulty: Challenging  |  Marks: 20

7 Sketch the graph of the function lnx12x2\ln x - \frac{1}{2}x^2.

Show that the differential equation dydx=2xyx21\frac{\mathrm{d}y}{\mathrm{d}x} = \frac{2xy}{x^2 - 1} describes a family of parabolas each of which passes through the points (1,0)(1, 0) and (1,0)(-1, 0) and has its vertex on the yy-axis.

Hence find the equation of the curve that passes through the point (1,1)(1, 1) and intersects each of the above parabolas orthogonally. Sketch this curve.

[Two curves intersect orthogonally if their tangents at the point of intersection are perpendicular.]

Model Solution

Part (a): Sketch of f(x)=lnx12x2f(x) = \ln x - \frac{1}{2}x^2

Domain: x>0x > 0.

Derivative. f(x)=1xx=1x2xf'(x) = \frac{1}{x} - x = \frac{1 - x^2}{x}.

Setting f(x)=0f'(x) = 0: x=1x = 1 (since x>0x > 0).

  • f(x)>0f'(x) > 0 for 0<x<10 < x < 1: ff is increasing.
  • f(x)<0f'(x) < 0 for x>1x > 1: ff is decreasing.

So x=1x = 1 is a maximum. f(1)=012=12f(1) = 0 - \frac{1}{2} = -\frac{1}{2}.

Behaviour at boundaries. f(x)f(x) \to -\infty as x0+x \to 0^+ (since lnx\ln x \to -\infty) and f(x)f(x) \to -\infty as x+x \to +\infty (since 12x2-\frac{1}{2}x^2 dominates).

Root. Since f(1)=12<0f(1) = -\frac{1}{2} < 0 and f(x)<0f(x) < 0 at both endpoints, ff is negative everywhere. The graph has a single maximum at (1,12)(1, -\frac{1}{2}) and lies entirely below the xx-axis.

The curve is an inverted bell shape, rising from -\infty near x=0x = 0, reaching a peak of 12-\frac{1}{2} at x=1x = 1, then falling back to -\infty.

Part (b): The differential equation describes a family of parabolas

dydx=2xyx21\frac{dy}{dx} = \frac{2xy}{x^2 - 1}

Separating variables:

dyy=2xx21dx\frac{dy}{y} = \frac{2x}{x^2 - 1}\,dx

Integrating:

lny=lnx21+C\ln|y| = \ln|x^2 - 1| + C

y=eCx21|y| = e^C |x^2 - 1|

y=A(x21)y = A(x^2 - 1)

where AA is an arbitrary constant. This is a family of parabolas y=A(x21)y = A(x^2 - 1), each with vertex at (0,A)(0, -A) (on the yy-axis) and each passing through (1,0)(1, 0) and (1,0)(-1, 0). \blacksquare

Part (c): Orthogonal trajectory through (1,1)(1, 1)

For the family y=A(x21)y = A(x^2 - 1), the slope is dydx=2Ax\frac{dy}{dx} = 2Ax. The orthogonal trajectory has slope 12Ax=x212xy-\frac{1}{2Ax} = -\frac{x^2 - 1}{2xy} (using A=yx21A = \frac{y}{x^2 - 1}).

So the orthogonal trajectory satisfies:

dydx=x212xy\frac{dy}{dx} = -\frac{x^2 - 1}{2xy}

Separating variables:

2ydy=x21xdx=(x+1x)dx2y\,dy = -\frac{x^2 - 1}{x}\,dx = \left(-x + \frac{1}{x}\right) dx

Integrating:

y2=x22+lnx+Cy^2 = -\frac{x^2}{2} + \ln|x| + C

Since we need the curve to pass through (1,1)(1, 1):

1=12+ln1+C    C=321 = -\frac{1}{2} + \ln 1 + C \implies C = \frac{3}{2}

The orthogonal trajectory is:

y2+x22lnx=32(x>0)y^2 + \frac{x^2}{2} - \ln x = \frac{3}{2} \qquad (x > 0)

or equivalently:

y=±32x22+lnxy = \pm\sqrt{\frac{3}{2} - \frac{x^2}{2} + \ln x}

Verification. At (1,1)(1, 1): 1+120=321 + \frac{1}{2} - 0 = \frac{3}{2} \checkmark. The slope of the orthogonal trajectory at (1,1)(1, 1) is 11211=0-\frac{1 - 1}{2 \cdot 1 \cdot 1} = 0, while the slope of the parabola y=x21y = x^2 - 1 at (1,1)(1, 1) is 22; the product is 00, confirming orthogonality at this point.

Sketch. The curve is symmetric about the xx-axis. It exists for x>0x > 0 where 32x22+lnx0\frac{3}{2} - \frac{x^2}{2} + \ln x \geqslant 0. As x0+x \to 0^+, lnx\ln x \to -\infty, so y2+y^2 \to +\infty and the curve has vertical asymptotes near x=0x = 0. The curve has a closed loop passing through (1,±1)(1, \pm 1), bounded on the right by the condition 32x22+lnx=0\frac{3}{2} - \frac{x^2}{2} + \ln x = 0.


Topic: 复数 / Complex Numbers  |  Difficulty: Hard  |  Marks: 20

8 (i) Prove that the equations z(1+i)2=2(*)|z - (1 + i)|^2 = 2 \qquad \text{(*)} and z(1i)2=2z12|z - (1 - i)|^2 = 2|z - 1|^2 describe the same locus in the complex zz-plane. Sketch this locus.

(ii) Prove that the equation arg(z2z)=π4(**)\arg\left(\frac{z - 2}{z}\right) = \frac{\pi}{4} \qquad \text{(**)} describes part of this same locus, and show on your sketch which part.

(iii) The complex number ww is related to zz by w=2z.w = \frac{2}{z} . Determine the locus produced in the complex ww-plane if zz satisfies (*). Sketch this locus and indicate the part of this locus that corresponds to (**).

Model Solution

Part (i): The two equations describe the same locus

Let z=x+iyz = x + iy.

Equation (*): z(1+i)2=2|z - (1+i)|^2 = 2

(x1)2+(y1)2=2(x - 1)^2 + (y - 1)^2 = 2

x2+y22x2y=0(I)x^2 + y^2 - 2x - 2y = 0 \qquad \text{(I)}

Second equation: z(1i)2=2z12|z - (1-i)|^2 = 2|z - 1|^2

(x1)2+(y+1)2=2(x1)2+2y2(x - 1)^2 + (y + 1)^2 = 2(x - 1)^2 + 2y^2

x22x+1+y2+2y+1=2x24x+2+2y2x^2 - 2x + 1 + y^2 + 2y + 1 = 2x^2 - 4x + 2 + 2y^2

0=x2+y22x2y0 = x^2 + y^2 - 2x - 2y

This is identical to (I). Therefore both equations describe the same locus: a circle centred at (1,1)(1, 1) with radius 2\sqrt{2}. \blacksquare

Sketch. The circle passes through the origin and through (2,0)(2, 0), (0,2)(0, 2), and (2,2)(2, 2).

Part (ii): The arg equation describes part of this locus

Let z=x+iyz = x + iy. We compute:

z2z=(x2)+iyx+iy\frac{z - 2}{z} = \frac{(x - 2) + iy}{x + iy}

Multiplying numerator and denominator by the conjugate of the denominator:

=[(x2)+iy](xiy)x2+y2= \frac{[(x-2) + iy](x - iy)}{x^2 + y^2}

Numerator: (x2)x+y2+i[yx(x2)y]=(x22x+y2)+i(2y)(x-2)x + y^2 + i[yx - (x-2)y] = (x^2 - 2x + y^2) + i(2y)

z2z=x2+y22xx2+y2+i2yx2+y2\frac{z - 2}{z} = \frac{x^2 + y^2 - 2x}{x^2 + y^2} + i\frac{2y}{x^2 + y^2}

The condition arg(z2z)=π4\arg\left(\frac{z-2}{z}\right) = \frac{\pi}{4} requires:

Im=Re>0\text{Im} = \text{Re} > 0

Setting Im == Re:

2yx2+y2=x2+y22xx2+y2\frac{2y}{x^2 + y^2} = \frac{x^2 + y^2 - 2x}{x^2 + y^2}

2y=x2+y22x2y = x^2 + y^2 - 2x

(x1)2+(y1)2=2(x - 1)^2 + (y - 1)^2 = 2

This is exactly the circle from part (i). \blacksquare

The additional constraint Re >0> 0 gives x2+y22x>0x^2 + y^2 - 2x > 0, i.e., (x1)2+y2>1(x - 1)^2 + y^2 > 1. Combined with the circle equation, this restricts the locus to the arc of the circle lying outside the circle centred at (1,0)(1, 0) with radius 11. This arc runs from (2,0)(2, 0) (where z2=0z - 2 = 0, so the argument is undefined) through (0,2)(0, 2) and back to (2,2)(2, 2), excluding the point (2,0)(2, 0) itself.

Part (iii): Locus in the ww-plane

Given w=2zw = \frac{2}{z}, so z=2w=2wˉw2z = \frac{2}{w} = \frac{2\bar{w}}{|w|^2}. With w=u+ivw = u + iv:

z=2(uiv)u2+v2=2uu2+v2i2vu2+v2z = \frac{2(u - iv)}{u^2 + v^2} = \frac{2u}{u^2 + v^2} - i\frac{2v}{u^2 + v^2}

Substituting into equation (*): z(1+i)2=2|z - (1+i)|^2 = 2, i.e., (x1)2+(y1)2=2(x-1)^2 + (y-1)^2 = 2 where x=2uu2+v2x = \frac{2u}{u^2+v^2}, y=2vu2+v2y = \frac{-2v}{u^2+v^2}:

(2uu2+v21)2+(2vu2+v21)2=2\left(\frac{2u}{u^2+v^2} - 1\right)^2 + \left(\frac{-2v}{u^2+v^2} - 1\right)^2 = 2

Let r2=u2+v2r^2 = u^2 + v^2. Expanding:

(2ur2)2+(2v+r2)2r4=2\frac{(2u - r^2)^2 + (2v + r^2)^2}{r^4} = 2

(2ur2)2+(2v+r2)2=2r4(2u - r^2)^2 + (2v + r^2)^2 = 2r^4

Expanding the left side:

4u24ur2+r4+4v2+4vr2+r4=2r44u^2 - 4ur^2 + r^4 + 4v^2 + 4vr^2 + r^4 = 2r^4

4(u2+v2)+r4(4u+4v)+2r4=2r44(u^2 + v^2) + r^4(-4u + 4v) + 2r^4 = 2r^4

Wait, let me redo this carefully:

4u24ur2+r4+4v2+4vr2+r4=2r44u^2 - 4ur^2 + r^4 + 4v^2 + 4vr^2 + r^4 = 2r^4

4(u2+v2)4r2(uv)+2r4=2r44(u^2 + v^2) - 4r^2(u - v) + 2r^4 = 2r^4

4r24r2(uv)=04r^2 - 4r^2(u - v) = 0

Since r20r^2 \neq 0 (as z0z \neq 0):

1(uv)=0    uv=11 - (u - v) = 0 \implies u - v = 1

The locus is the line uv=1u - v = 1, i.e., Re(w)Im(w)=1\operatorname{Re}(w) - \operatorname{Im}(w) = 1. \blacksquare

Sketch. This is a straight line in the ww-plane passing through (1,0)(1, 0) and (0,1)(0, -1) with gradient 11.

Which part corresponds to () :**

The arc from part (ii) satisfies (x1)2+y2>1(x-1)^2 + y^2 > 1 (the point (2,0)(2, 0) is excluded). The boundary point (2,0)(2, 0) maps to w=22=1w = \frac{2}{2} = 1, i.e., (u,v)=(1,0)(u, v) = (1, 0), which lies on the line uv=1u - v = 1.

The corresponding part of the line uv=1u - v = 1 is the portion where the arc maps to, which is the line excluding the point (1,0)(1, 0).