Skip to content

STEP3 2003 -- Pure Mathematics

STEP3 2003 — Section A (Pure Mathematics)

Section titled “STEP3 2003 — Section A (Pure Mathematics)”

Exam: STEP3  |  Year: 2003  |  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 CalculusStandard链式法则,反函数求导,换元积分
2级数与组合数学 Series & CombinatoricsChallenging组合恒等式,广义二项式定理,幂级数求和
3微积分 CalculusStandard商法则求导,奇偶函数性质,函数图像分析
4解析几何与参数方程 Analytic Geometry & Parametric EquationsChallenging参数方程求切线,递推关系,三角恒等式
5代数与方程理论 Algebra & Equation TheoryChallenging判别式,韦达定理,极值分析,充要条件
6三角函数 TrigonometryChallenging三角恒等式证明,裂项求和,三角方程求解
7坐标几何 Coordinate GeometryChallenging坐标几何,向量投影,共线条件
8微分方程 Differential EquationsChallenging隐函数求导,变量替换,微分方程求解

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

1 Given that x+a>0x + a > 0 and x+b>0x + b > 0, and that b>ab > a, show that ddxarcsin(x+ax+b)=ba(x+b)a+b+2x\frac{\mathrm{d}}{\mathrm{d}x} \arcsin \left( \frac{x + a}{x + b} \right) = \frac{\sqrt{b - a}}{(x + b) \sqrt{a + b + 2x}} and find ddxarcosh(x+bx+a)\frac{\mathrm{d}}{\mathrm{d}x} \operatorname{arcosh} \left( \frac{x + b}{x + a} \right).

Hence, or otherwise, integrate, for x>1x > -1,

(i) 1(x+1)x+3 dx\int \frac{1}{(x + 1) \sqrt{x + 3}} \mathrm{~d}x,

(ii) 1(x+3)x+1 dx\int \frac{1}{(x + 3) \sqrt{x + 1}} \mathrm{~d}x.

[You may use the results ddxarcsinx=11x2\frac{\mathrm{d}}{\mathrm{d}x} \arcsin x = \frac{1}{\sqrt{1 - x^2}} and ddxarcoshx=1x21\frac{\mathrm{d}}{\mathrm{d}x} \operatorname{arcosh} x = \frac{1}{\sqrt{x^2 - 1}}. ]

Model Solution

Step 1: Differentiate arcsin ⁣(x+ax+b)\arcsin\!\left(\frac{x+a}{x+b}\right).

Let u=x+ax+bu = \frac{x+a}{x+b}. Then by the chain rule:

ddxarcsinu=11u2dudx.\frac{\mathrm{d}}{\mathrm{d}x}\arcsin u = \frac{1}{\sqrt{1-u^2}}\cdot\frac{\mathrm{d}u}{\mathrm{d}x}.

We compute uu and its derivative:

dudx=(x+b)(x+a)(x+b)2=ba(x+b)2.\frac{\mathrm{d}u}{\mathrm{d}x} = \frac{(x+b)-(x+a)}{(x+b)^2} = \frac{b-a}{(x+b)^2}.

For 1u21 - u^2:

1u2=1(x+a)2(x+b)2=(x+b)2(x+a)2(x+b)2=(ba)(2x+a+b)(x+b)2.1 - u^2 = 1 - \frac{(x+a)^2}{(x+b)^2} = \frac{(x+b)^2-(x+a)^2}{(x+b)^2} = \frac{(b-a)(2x+a+b)}{(x+b)^2}.

Since b>ab > a, x+a>0x+a > 0, and x+b>0x+b > 0, we have ba>0b-a > 0 and 2x+a+b>02x+a+b > 0, so:

1u2=(ba)(2x+a+b)x+b.\sqrt{1-u^2} = \frac{\sqrt{(b-a)(2x+a+b)}}{x+b}.

Therefore:

ddxarcsin ⁣(x+ax+b)=x+b(ba)(2x+a+b)ba(x+b)2=ba(x+b)(ba)(2x+a+b).\frac{\mathrm{d}}{\mathrm{d}x}\arcsin\!\left(\frac{x+a}{x+b}\right) = \frac{x+b}{\sqrt{(b-a)(2x+a+b)}}\cdot\frac{b-a}{(x+b)^2} = \frac{b-a}{(x+b)\sqrt{(b-a)(2x+a+b)}}.

Simplifying:

ddxarcsin ⁣(x+ax+b)=ba(x+b)2x+a+b.(shown)\frac{\mathrm{d}}{\mathrm{d}x}\arcsin\!\left(\frac{x+a}{x+b}\right) = \frac{\sqrt{b-a}}{(x+b)\sqrt{2x+a+b}}. \qquad \text{(shown)}

Step 2: Differentiate arcosh ⁣(x+bx+a)\operatorname{arcosh}\!\left(\frac{x+b}{x+a}\right).

Let v=x+bx+av = \frac{x+b}{x+a}. Since b>ab > a and x+a>0x+a > 0, we have v>1v > 1, so arcosh\operatorname{arcosh} is well-defined.

dvdx=(x+a)(x+b)(x+a)2=ab(x+a)2.\frac{\mathrm{d}v}{\mathrm{d}x} = \frac{(x+a)-(x+b)}{(x+a)^2} = \frac{a-b}{(x+a)^2}.

v21=(x+b)2(x+a)2(x+a)2=(ba)(2x+a+b)(x+a)2.v^2 - 1 = \frac{(x+b)^2-(x+a)^2}{(x+a)^2} = \frac{(b-a)(2x+a+b)}{(x+a)^2}.

v21=(ba)(2x+a+b)x+a.\sqrt{v^2-1} = \frac{\sqrt{(b-a)(2x+a+b)}}{x+a}.

Therefore:

ddxarcosh ⁣(x+bx+a)=1v21dvdx=x+a(ba)(2x+a+b)ab(x+a)2.\frac{\mathrm{d}}{\mathrm{d}x}\operatorname{arcosh}\!\left(\frac{x+b}{x+a}\right) = \frac{1}{\sqrt{v^2-1}}\cdot\frac{\mathrm{d}v}{\mathrm{d}x} = \frac{x+a}{\sqrt{(b-a)(2x+a+b)}}\cdot\frac{a-b}{(x+a)^2}.

=ab(x+a)(ba)(2x+a+b)=(ba)(x+a)(ba)(2x+a+b)=ba(x+a)2x+a+b.= \frac{a-b}{(x+a)\sqrt{(b-a)(2x+a+b)}} = \frac{-(b-a)}{(x+a)\sqrt{(b-a)(2x+a+b)}} = \frac{-\sqrt{b-a}}{(x+a)\sqrt{2x+a+b}}.

So ddxarcosh ⁣(x+bx+a)=ba(x+a)a+b+2x.\frac{\mathrm{d}}{\mathrm{d}x}\operatorname{arcosh}\!\left(\frac{x+b}{x+a}\right) = \frac{-\sqrt{b-a}}{(x+a)\sqrt{a+b+2x}}.

Step 3: Integral (i).

We need 1(x+1)x+3dx\int \frac{1}{(x+1)\sqrt{x+3}}\,\mathrm{d}x for x>1x > -1.

Substitute u=x+3u = \sqrt{x+3}, so x=u23x = u^2 - 3 and dx=2udu\mathrm{d}x = 2u\,\mathrm{d}u:

1(x+1)x+3dx=2u(u22)udu=2u22du.\int \frac{1}{(x+1)\sqrt{x+3}}\,\mathrm{d}x = \int \frac{2u}{(u^2 - 2)\cdot u}\,\mathrm{d}u = \int \frac{2}{u^2 - 2}\,\mathrm{d}u.

For x>1x > -1 we have u>2u > \sqrt{2}, so u2>2u^2 > 2. Using the standard result 1u2a2du=12alnuau+a\int \frac{1}{u^2 - a^2}\,\mathrm{d}u = \frac{1}{2a}\ln\left|\frac{u-a}{u+a}\right|:

2u22du=222lnu2u+2=12lnx+32x+3+2+C.\int \frac{2}{u^2-2}\,\mathrm{d}u = \frac{2}{2\sqrt{2}}\ln\frac{u-\sqrt{2}}{u+\sqrt{2}} = \frac{1}{\sqrt{2}}\ln\frac{\sqrt{x+3}-\sqrt{2}}{\sqrt{x+3}+\sqrt{2}} + C.

Step 4: Integral (ii).

We need 1(x+3)x+1dx\int \frac{1}{(x+3)\sqrt{x+1}}\,\mathrm{d}x for x>1x > -1.

Substitute u=x+1u = \sqrt{x+1}, so x=u21x = u^2 - 1 and dx=2udu\mathrm{d}x = 2u\,\mathrm{d}u:

1(x+3)x+1dx=2u(u2+2)udu=2u2+2du=22arctanu2.\int \frac{1}{(x+3)\sqrt{x+1}}\,\mathrm{d}x = \int \frac{2u}{(u^2 + 2)\cdot u}\,\mathrm{d}u = \int \frac{2}{u^2 + 2}\,\mathrm{d}u = \frac{2}{\sqrt{2}}\arctan\frac{u}{\sqrt{2}}.

=2arctanx+12+C=2arctanx+12+C.= \sqrt{2}\arctan\frac{\sqrt{x+1}}{\sqrt{2}} + C = \sqrt{2}\arctan\sqrt{\frac{x+1}{2}} + C.


Topic: 级数与组合数学 Series & Combinatorics  |  Difficulty: Challenging  |  Marks: 20

2 Show that 2rCr=1×3××(2r1)r!×2r{}^{2r}C_r = \frac{1 \times 3 \times \cdots \times (2r - 1)}{r!} \times 2^r, for r1r \geqslant 1.

(i) Give the first four terms of the binomial series for (1p)12(1 - p)^{-\frac{1}{2}}.

By choosing a suitable value for pp in this series, or otherwise, show that r=02rCr8r=2.\sum_{r=0}^{\infty} \frac{{}^{2r}C_r}{8^r} = \sqrt{2} .

(ii) Show that r=0(2r+1)2rCr5r=(5)3.\sum_{r=0}^{\infty} \frac{(2r + 1) \cdot {}^{2r}C_r}{5^r} = (\sqrt{5})^3 .

[Note: nCr{}^nC_r is an alternative notation for (nr)\begin{pmatrix} n \\ r \end{pmatrix} for r1r \geqslant 1, and 0C0=1{}^0C_0 = 1.]

Model Solution

Step 1: Show 2rCr=1×3××(2r1)r!×2r{}^{2r}C_r = \frac{1\times 3\times\cdots\times(2r-1)}{r!}\times 2^r.

2rCr=(2r)!r!r!=1×2×3××(2r)(r!)(r!).{}^{2r}C_r = \frac{(2r)!}{r!\,r!} = \frac{1\times 2\times 3\times\cdots\times(2r)}{(r!)(r!)}.

Separate the numerator into odd and even factors:

1×2×3××(2r)=[1×3×5××(2r1)]×[2×4×6××2r].1\times 2\times 3\times\cdots\times(2r) = \bigl[1\times 3\times 5\times\cdots\times(2r-1)\bigr]\times\bigl[2\times 4\times 6\times\cdots\times 2r\bigr].

The even part factors as 2×4×6××2r=2r(1×2×3××r)=2rr!2\times 4\times 6\times\cdots\times 2r = 2^r(1\times 2\times 3\times\cdots\times r) = 2^r\,r!. Therefore:

2rCr=[1×3××(2r1)]×2rr!r!r!=1×3××(2r1)r!×2r.(shown){}^{2r}C_r = \frac{\bigl[1\times 3\times\cdots\times(2r-1)\bigr]\times 2^r\,r!}{r!\,r!} = \frac{1\times 3\times\cdots\times(2r-1)}{r!}\times 2^r. \qquad \text{(shown)}

Part (i): Binomial series for (1p)1/2(1-p)^{-1/2} and the sum.

The generalised binomial series gives:

(1p)1/2=r=0(1/2r)(p)r.(1-p)^{-1/2} = \sum_{r=0}^{\infty}\binom{-1/2}{r}(-p)^r.

Computing the first four terms:

(1/20)=1,(1/21)(p)=12p,\binom{-1/2}{0} = 1, \qquad \binom{-1/2}{1}(-p) = \tfrac{1}{2}p,

(1/22)(p)2=(1/2)(3/2)2!p2=38p2,(1/23)(p)3=(1/2)(3/2)(5/2)3!(p3)=516p3.\binom{-1/2}{2}(-p)^2 = \frac{(-1/2)(-3/2)}{2!}p^2 = \frac{3}{8}p^2, \qquad \binom{-1/2}{3}(-p)^3 = \frac{(-1/2)(-3/2)(-5/2)}{3!}(-p^3) = \frac{5}{16}p^3.

So the first four terms are:

(1p)1/2=1+12p+38p2+516p3+(1-p)^{-1/2} = 1 + \frac{1}{2}p + \frac{3}{8}p^2 + \frac{5}{16}p^3 + \cdots

In general, (1/2r)(1)r=1×3××(2r1)2rr!\binom{-1/2}{r}(-1)^r = \frac{1\times 3\times\cdots\times(2r-1)}{2^r\,r!}, so:

(1p)1/2=r=01×3××(2r1)2rr!pr.(1-p)^{-1/2} = \sum_{r=0}^{\infty}\frac{1\times 3\times\cdots\times(2r-1)}{2^r\,r!}\,p^r.

Using the identity from Step 1: 1×3××(2r1)r!=2rCr2r\frac{1\times 3\times\cdots\times(2r-1)}{r!} = \frac{{}^{2r}C_r}{2^r}, so each term becomes:

2rCr2r2rpr=2rCr4rpr.\frac{{}^{2r}C_r}{2^r\cdot 2^r}\,p^r = \frac{{}^{2r}C_r}{4^r}\,p^r.

Thus:

(1p)1/2=r=02rCr4rpr.(1-p)^{-1/2} = \sum_{r=0}^{\infty}\frac{{}^{2r}C_r}{4^r}\,p^r.

Setting p=12p = \frac{1}{2}:

(112)1/2=(12)1/2=2=r=02rCr4r12r=r=02rCr8r.(shown)(1-\tfrac{1}{2})^{-1/2} = \left(\tfrac{1}{2}\right)^{-1/2} = \sqrt{2} = \sum_{r=0}^{\infty}\frac{{}^{2r}C_r}{4^r}\cdot\frac{1}{2^r} = \sum_{r=0}^{\infty}\frac{{}^{2r}C_r}{8^r}. \qquad \text{(shown)}

Part (ii): Show r=0(2r+1)2rCr5r=(5)3\sum_{r=0}^{\infty}\frac{(2r+1)\cdot{}^{2r}C_r}{5^r} = (\sqrt{5})^3.

From Part (i) we have (1p)1/2=r=02rCr4rpr(1-p)^{-1/2} = \sum_{r=0}^{\infty}\frac{{}^{2r}C_r}{4^r}\,p^r.

Differentiate both sides with respect to pp:

12(1p)3/2=r=12rCr4rrpr1=r=02(r+1)Cr+14r+1(r+1)pr.\frac{1}{2}(1-p)^{-3/2} = \sum_{r=1}^{\infty}\frac{{}^{2r}C_r}{4^r}\,r\,p^{r-1} = \sum_{r=0}^{\infty}\frac{{}^{2(r+1)}C_{r+1}}{4^{r+1}}(r+1)\,p^r.

Multiply through by pp:

p2(1p)3/2=r=02r+2Cr+14r+1(r+1)pr+1=r=12rCr4rrpr.\frac{p}{2}(1-p)^{-3/2} = \sum_{r=0}^{\infty}\frac{{}^{2r+2}C_{r+1}}{4^{r+1}}(r+1)\,p^{r+1} = \sum_{r=1}^{\infty}\frac{{}^{2r}C_r}{4^r}\cdot r\cdot p^r.

This gives us a series with factor rr, but we need (2r+1)(2r+1). Instead, multiply the original series by (1p)1(1-p)^{-1} or use a different approach.

Consider (1p)3/2(1-p)^{-3/2}. By the binomial series:

(1p)3/2=r=0(3/2r)(p)r.(1-p)^{-3/2} = \sum_{r=0}^{\infty}\binom{-3/2}{r}(-p)^r.

We have (3/2r)(1)r=3×5××(2r+1)2rr!=(2r+1)!!2rr!\binom{-3/2}{r}(-1)^r = \frac{3\times 5\times\cdots\times(2r+1)}{2^r\,r!} = \frac{(2r+1)!!}{2^r\,r!}.

Now (2r+1)!!=1×3××(2r+1)=(2r+1)×[1×3××(2r1)](2r+1)!! = 1\times 3\times\cdots\times(2r+1) = (2r+1)\times[1\times 3\times\cdots\times(2r-1)], so:

(3/2r)(1)r=(2r+1)×[1×3××(2r1)]2rr!.\binom{-3/2}{r}(-1)^r = \frac{(2r+1)\times[1\times 3\times\cdots\times(2r-1)]}{2^r\,r!}.

Using the identity 1×3××(2r1)r!=2rCr2r\frac{1\times 3\times\cdots\times(2r-1)}{r!} = \frac{{}^{2r}C_r}{2^r}:

(3/2r)(1)r=(2r+1)2rCr4r.\binom{-3/2}{r}(-1)^r = \frac{(2r+1)\,{}^{2r}C_r}{4^r}.

Therefore:

(1p)3/2=r=0(2r+1)2rCr4rpr.(1-p)^{-3/2} = \sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{4^r}\,p^r.

Setting p=15p = \frac{1}{5}:

(115)3/2=(45)3/2=(54)3/2=558.(1-\tfrac{1}{5})^{-3/2} = \left(\tfrac{4}{5}\right)^{-3/2} = \left(\tfrac{5}{4}\right)^{3/2} = \frac{5\sqrt{5}}{8}.

And the series gives:

r=0(2r+1)2rCr4r15r=r=0(2r+1)2rCr20r.\sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{4^r}\cdot\frac{1}{5^r} = \sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{20^r}.

This gives (2r+1)2rCr20r=558\sum\frac{(2r+1){}^{2r}C_r}{20^r} = \frac{5\sqrt{5}}{8}, not (2r+1)2rCr5r\sum\frac{(2r+1){}^{2r}C_r}{5^r}.

We need 15r\frac{1}{5^r} not 120r\frac{1}{20^r}, so multiply both sides by 4r4^r inside the sum. Let us set p=45p = \frac{4}{5} instead:

(145)3/2=(15)3/2=53/2=55.(1-\tfrac{4}{5})^{-3/2} = \left(\tfrac{1}{5}\right)^{-3/2} = 5^{3/2} = 5\sqrt{5}.

The series gives:

r=0(2r+1)2rCr4r(45)r=r=0(2r+1)2rCr5r.\sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{4^r}\left(\frac{4}{5}\right)^r = \sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{5^r}.

Therefore:

r=0(2r+1)2rCr5r=55=(5)3.(shown)\sum_{r=0}^{\infty}\frac{(2r+1)\,{}^{2r}C_r}{5^r} = 5\sqrt{5} = (\sqrt{5})^3. \qquad \text{(shown)}


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

3 If mm is a positive integer, show that (1+x)m+(1x)m0(1+x)^m + (1-x)^m \neq 0 for any real xx.

The function ff is defined by

f(x)=(1+x)m(1x)m(1+x)m+(1x)m.f(x) = \frac{(1+x)^m - (1-x)^m}{(1+x)^m + (1-x)^m} .

Find and simplify an expression for f(x)f'(x).

In the case m=5m=5, sketch the curves y=f(x)y = f(x) and y=1f(x)y = \frac{1}{f(x)}.

Model Solution

Step 1: Show (1+x)m+(1x)m0(1+x)^m + (1-x)^m \neq 0 for any real xx.

We consider two cases based on the parity of mm.

Case 1: mm even. Then (1+x)m0(1+x)^m \geq 0 and (1x)m0(1-x)^m \geq 0 for all real xx. For the sum to be zero, we would need both terms to be zero simultaneously, requiring x=1x = -1 and x=1x = 1 at the same time. This is impossible, so (1+x)m+(1x)m>0(1+x)^m + (1-x)^m > 0.

Case 2: mm odd. Suppose (1+x)m+(1x)m=0(1+x)^m + (1-x)^m = 0. Then (1+x)m=(1x)m(1+x)^m = -(1-x)^m. Since mm is odd, (1x)m=(x1)m-(1-x)^m = (x-1)^m, so (1+x)m=(x1)m(1+x)^m = (x-1)^m. Taking mm-th roots: 1+x=x11 + x = x - 1, giving 1=11 = -1, a contradiction.

Therefore (1+x)m+(1x)m0(1+x)^m + (1-x)^m \neq 0 for all real xx. (shown)\qquad \text{(shown)}

Step 2: Find and simplify f(x)f'(x).

Let u=(1+x)m(1x)mu = (1+x)^m - (1-x)^m and v=(1+x)m+(1x)mv = (1+x)^m + (1-x)^m. Then:

u=m(1+x)m1m(1x)m1(1)=m(1+x)m1+m(1x)m1,u' = m(1+x)^{m-1} - m(1-x)^{m-1} \cdot (-1) = m(1+x)^{m-1} + m(1-x)^{m-1},

v=m(1+x)m1+m(1x)m1(1)=m(1+x)m1m(1x)m1.v' = m(1+x)^{m-1} + m(1-x)^{m-1} \cdot (-1) = m(1+x)^{m-1} - m(1-x)^{m-1}.

By the quotient rule f(x)=uvuvv2f'(x) = \frac{u'v - uv'}{v^2}, we compute the numerator:

uv=m[(1+x)m1+(1x)m1][(1+x)m+(1x)m]u'v = m\bigl[(1+x)^{m-1} + (1-x)^{m-1}\bigr]\bigl[(1+x)^m + (1-x)^m\bigr]

=m[(1+x)2m1+(1+x)m1(1x)m+(1x)m1(1+x)m+(1x)2m1]= m\bigl[(1+x)^{2m-1} + (1+x)^{m-1}(1-x)^m + (1-x)^{m-1}(1+x)^m + (1-x)^{2m-1}\bigr]

uv=m[(1+x)m(1x)m][(1+x)m1(1x)m1]uv' = m\bigl[(1+x)^m - (1-x)^m\bigr]\bigl[(1+x)^{m-1} - (1-x)^{m-1}\bigr]

=m[(1+x)2m1(1+x)m(1x)m1(1x)m(1+x)m1+(1x)2m1]= m\bigl[(1+x)^{2m-1} - (1+x)^m(1-x)^{m-1} - (1-x)^m(1+x)^{m-1} + (1-x)^{2m-1}\bigr]

Subtracting:

uvuv=m[2(1+x)m1(1x)m+2(1x)m1(1+x)m]u'v - uv' = m\bigl[2(1+x)^{m-1}(1-x)^m + 2(1-x)^{m-1}(1+x)^m\bigr]

=2m(1+x)m1(1x)m1[(1x)+(1+x)]= 2m(1+x)^{m-1}(1-x)^{m-1}\bigl[(1-x) + (1+x)\bigr]

=2m(1+x)m1(1x)m12=4m(1+x)m1(1x)m1.= 2m(1+x)^{m-1}(1-x)^{m-1} \cdot 2 = 4m(1+x)^{m-1}(1-x)^{m-1}.

Therefore:

f(x)=4m(1+x)m1(1x)m1[(1+x)m+(1x)m]2.\boxed{f'(x) = \frac{4m\,(1+x)^{m-1}(1-x)^{m-1}}{\bigl[(1+x)^m + (1-x)^m\bigr]^2}}.

Step 3: Sketch for m=5m = 5.

With m=5m = 5, expanding (1+x)5(1+x)^5 and (1x)5(1-x)^5:

(1+x)5=1+5x+10x2+10x3+5x4+x5(1+x)^5 = 1 + 5x + 10x^2 + 10x^3 + 5x^4 + x^5

(1x)5=15x+10x210x3+5x4x5(1-x)^5 = 1 - 5x + 10x^2 - 10x^3 + 5x^4 - x^5

So (1+x)5(1x)5=10x+20x3+2x5(1+x)^5 - (1-x)^5 = 10x + 20x^3 + 2x^5 and (1+x)5+(1x)5=2+20x2+10x4(1+x)^5 + (1-x)^5 = 2 + 20x^2 + 10x^4. Thus:

f(x)=2x(x4+10x2+5)2(5x4+10x2+1)=x(x4+10x2+5)5x4+10x2+1.f(x) = \frac{2x(x^4 + 10x^2 + 5)}{2(5x^4 + 10x^2 + 1)} = \frac{x(x^4 + 10x^2 + 5)}{5x^4 + 10x^2 + 1}.

Key features of y=f(x)y = f(x):

  • Odd function: f(x)=f(x)f(-x) = -f(x), so the graph has rotational symmetry about the origin.
  • f(0)=0f(0) = 0, and the gradient at the origin is f(0)=451122=5f'(0) = \frac{4 \cdot 5 \cdot 1 \cdot 1}{2^2} = 5.
  • f(1)=1616=1f(1) = \frac{16}{16} = 1 and f(1)=1f(-1) = -1.
  • Since (1x2)40(1-x^2)^4 \geq 0 for all xx and the denominator is always positive, f(x)0f'(x) \geq 0 for all xx, with equality only at x=±1x = \pm 1. So ff is strictly increasing.
  • As xx \to \infty: f(x)2x510x4=x5f(x) \approx \frac{2x^5}{10x^4} = \frac{x}{5}, so the curve approaches the oblique asymptote y=x/5y = x/5.

For y=1f(x)y = \frac{1}{f(x)}:

1f(x)=5x4+10x2+1x(x4+10x2+5).\frac{1}{f(x)} = \frac{5x^4 + 10x^2 + 1}{x(x^4 + 10x^2 + 5)}.

Key features:

  • Also an odd function: 1f(x)=1f(x)\frac{1}{f(-x)} = -\frac{1}{f(x)}.
  • Vertical asymptote at x=0x = 0: as x0+x \to 0^+, f(x)0+f(x) \to 0^+, so 1/f(x)+1/f(x) \to +\infty.
  • 1f(1)=1\frac{1}{f(1)} = 1 and 1f(1)=1\frac{1}{f(-1)} = -1, so the two curves intersect at (±1,±1)(\pm 1, \pm 1).
  • As xx \to \infty: 1/f(x)5x42x5=52x0+1/f(x) \approx \frac{5x^4}{2x^5} = \frac{5}{2x} \to 0^+, so the positive xx-axis is a horizontal asymptote.
  • Since ff is strictly increasing and positive for x>0x > 0, 1/f1/f is strictly decreasing for x>0x > 0.

Sketch description: y=f(x)y = f(x) is a monotonically increasing S-curve through the origin with gradient 5, passing through (±1,±1)(\pm 1, \pm 1), and approaching y=x/5y = x/5 for large x|x|. y=1/f(x)y = 1/f(x) is an odd curve with a vertical asymptote at x=0x = 0, decreasing from ++\infty to 00 for x>0x > 0, crossing y=f(x)y = f(x) at (1,1)(1, 1) and (1,1)(-1, -1).


Topic: 解析几何与参数方程 Analytic Geometry & Parametric Equations  |  Difficulty: Challenging  |  Marks: 20

4 A curve is defined parametrically by

x=t2,y=t(1+t2).x = t^2, \quad y = t(1+t^2) .

The tangent at the point with parameter tt, where t0t \neq 0, meets the curve again at the point with parameter TT, where TtT \neq t. Show that

T=1t22tand3t21.T = \frac{1-t^2}{2t} \quad \text{and} \quad 3t^2 \neq 1 .

Given a point P0P_0 on the curve, with parameter t0t_0, a sequence of points P0,P1,P2,P_0, P_1, P_2, \dots on the curve is constructed such that the tangent at PiP_i meets the curve again at Pi+1P_{i+1}. If t0=tan7π18t_0 = \tan \frac{7\pi}{18}, show that P3=P0P_3 = P_0 but P1P0P_1 \neq P_0. Find a second value of t0t_0, with t0>0t_0 > 0, for which P3=P0P_3 = P_0 but P1P0P_1 \neq P_0.

Model Solution

Step 1: Find TT.

The curve is x=t2x = t^2, y=t(1+t2)y = t(1+t^2). Differentiating:

dxdt=2t,dydt=1+3t2,dydx=1+3t22t.\frac{\mathrm{d}x}{\mathrm{d}t} = 2t, \qquad \frac{\mathrm{d}y}{\mathrm{d}t} = 1 + 3t^2, \qquad \frac{\mathrm{d}y}{\mathrm{d}x} = \frac{1+3t^2}{2t}.

The tangent at parameter tt has equation:

yt(1+t2)=1+3t22t(xt2).y - t(1+t^2) = \frac{1+3t^2}{2t}(x - t^2).

To find where this meets the curve again, substitute x=T2x = T^2, y=T(1+T2)y = T(1+T^2):

T(1+T2)t(1+t2)=1+3t22t(T2t2).T(1+T^2) - t(1+t^2) = \frac{1+3t^2}{2t}(T^2 - t^2).

Multiply through by 2t2t:

2tT(1+T2)2t2(1+t2)=(1+3t2)(T2t2).2tT(1+T^2) - 2t^2(1+t^2) = (1+3t^2)(T^2 - t^2).

Expand:

2tT+2tT32t22t4=T2+3t2T2t23t4.2tT + 2tT^3 - 2t^2 - 2t^4 = T^2 + 3t^2T^2 - t^2 - 3t^4.

Rearrange:

2tT3T2(1+3t2)+2tT+(t23t4+t2)=02tT^3 - T^2(1+3t^2) + 2tT + (t^2 - 3t^4 + t^2) = 0

Wait, let me redo this carefully. We have:

2tT+2tT32t22t4T23t2T2+t2+3t4=02tT + 2tT^3 - 2t^2 - 2t^4 - T^2 - 3t^2T^2 + t^2 + 3t^4 = 0

2tT3(1+3t2)T2+2tT+(t2+t4)=02tT^3 - (1+3t^2)T^2 + 2tT + (t^2 + t^4) = 0

Since T=tT = t is a solution (the tangent passes through the point itself), (Tt)(T - t) is a factor. We can factor:

2tT3(1+3t2)T2+2tT+t2(1+t2)=0.2tT^3 - (1+3t^2)T^2 + 2tT + t^2(1+t^2) = 0.

Dividing by (Tt)(T-t): we know T=tT = t satisfies this, so let us perform polynomial long division. Alternatively, note that T=tT = t is a root, so:

2tT3(1+3t2)T2+2tT+t2+t4=(Tt)Q(T).2tT^3 - (1+3t^2)T^2 + 2tT + t^2 + t^4 = (T-t)\cdot Q(T).

By polynomial division or comparing coefficients:

2tT3(1+3t2)T2+2tT+t2(1+t2)=(Tt)[2tT2(1+t2)Tt(1+t2)].2tT^3 - (1+3t^2)T^2 + 2tT + t^2(1+t^2) = (T-t)\bigl[2tT^2 - (1+t^2)T - t(1+t^2)\bigr].

Let us verify by expanding the right side:

(Tt)(2tT2(1+t2)Tt(1+t2))(T-t)(2tT^2 - (1+t^2)T - t(1+t^2)) =2tT3(1+t2)T2t(1+t2)T2t2T2+t(1+t2)T+t2(1+t2)= 2tT^3 - (1+t^2)T^2 - t(1+t^2)T - 2t^2T^2 + t(1+t^2)T + t^2(1+t^2) =2tT3(1+t2+2t2)T2+t2(1+t2)= 2tT^3 - (1+t^2+2t^2)T^2 + t^2(1+t^2) =2tT3(1+3t2)T2+t2+t4.= 2tT^3 - (1+3t^2)T^2 + t^2 + t^4.

Comparing with the original: the original also has +2tT+2tT. So let me recheck. The equation is:

2tT3(1+3t2)T2+2tT+t2(1+t2)=0.2tT^3 - (1+3t^2)T^2 + 2tT + t^2(1+t^2) = 0.

Let me re-verify the subtraction step. From:

2tT+2tT32t22t4T23t2T2+t2+3t4=02tT + 2tT^3 - 2t^2 - 2t^4 - T^2 - 3t^2T^2 + t^2 + 3t^4 = 0

Collecting:

  • T3T^3: 2t2t
  • T2T^2: (1+3t2)-(1+3t^2)
  • T1T^1: 2t2t
  • T0T^0: 2t22t4+t2+3t4=t2+t4=t2(t21)-2t^2 - 2t^4 + t^2 + 3t^4 = -t^2 + t^4 = t^2(t^2-1)

Hmm, that’s different. Let me redo. 2t2+t2=t2-2t^2 + t^2 = -t^2 and 2t4+3t4=t4-2t^4 + 3t^4 = t^4. So constant term is t4t2=t2(t21)t^4 - t^2 = t^2(t^2-1).

So: 2tT3(1+3t2)T2+2tT+t2(t21)=02tT^3 - (1+3t^2)T^2 + 2tT + t^2(t^2-1) = 0.

Factor out (Tt)(T-t). Since T=tT = t is a root:

2tt3(1+3t2)t2+2tt+t2(t21)=2t4t23t4+2t2+t4t2=0.2t\cdot t^3 - (1+3t^2)t^2 + 2t\cdot t + t^2(t^2-1) = 2t^4 - t^2 - 3t^4 + 2t^2 + t^4 - t^2 = 0. \checkmark

Dividing: 2tT3(1+3t2)T2+2tT+t2(t21)=(Tt)(2tT2+At+B)2tT^3 - (1+3t^2)T^2 + 2tT + t^2(t^2-1) = (T-t)(2tT^2 + At + B) where we need to find the quotient.

Using synthetic division with root T=tT = t:

(Tt)(2tT2+c1T+c0)(T-t)(2tT^2 + c_1 T + c_0)

Expanding: 2tT3+c1T2+c0T2t2T2c1tTc0t2tT^3 + c_1T^2 + c_0T - 2t^2T^2 - c_1tT - c_0t

=2tT3+(c12t2)T2+(c0c1t)Tc0t= 2tT^3 + (c_1 - 2t^2)T^2 + (c_0 - c_1t)T - c_0t

Matching:

  • T2T^2: c12t2=(1+3t2)c1=1t2c_1 - 2t^2 = -(1+3t^2) \Rightarrow c_1 = -1 - t^2
  • T1T^1: c0c1t=2tc0=2t+c1t=2tt(1+t2)=2ttt3=tt3=t(1t2)c_0 - c_1t = 2t \Rightarrow c_0 = 2t + c_1 t = 2t - t(1+t^2) = 2t - t - t^3 = t - t^3 = t(1-t^2)
  • T0T^0: c0t=t2(t21)-c_0t = t^2(t^2-1), so c0=t(t21)=t(1t2)c_0 = -t(t^2-1) = t(1-t^2). \checkmark

So the quadratic factor is:

2tT2(1+t2)T+t(1t2)=0.2tT^2 - (1+t^2)T + t(1-t^2) = 0.

Using the quadratic formula:

T=(1+t2)±(1+t2)242tt(1t2)4t=(1+t2)±(1+t2)28t2(1t2)4t.T = \frac{(1+t^2) \pm \sqrt{(1+t^2)^2 - 4\cdot 2t\cdot t(1-t^2)}}{4t} = \frac{(1+t^2) \pm \sqrt{(1+t^2)^2 - 8t^2(1-t^2)}}{4t}.

Expand the discriminant:

(1+t2)28t2(1t2)=1+2t2+t48t2+8t4=16t2+9t4=(13t2)2.(1+t^2)^2 - 8t^2(1-t^2) = 1 + 2t^2 + t^4 - 8t^2 + 8t^4 = 1 - 6t^2 + 9t^4 = (1-3t^2)^2.

Therefore:

T=(1+t2)±(13t2)4t.T = \frac{(1+t^2) \pm (1-3t^2)}{4t}.

Taking the ++ sign: T=22t24t=1t22tT = \frac{2-2t^2}{4t} = \frac{1-t^2}{2t}.

Taking the - sign: T=4t24t=tT = \frac{4t^2}{4t} = t (the original point).

So T=1t22tT = \frac{1-t^2}{2t}. For this to be well-defined we need t0t \neq 0, and for TtT \neq t we need 1t22tt\frac{1-t^2}{2t} \neq t, i.e., 1t22t21-t^2 \neq 2t^2, i.e., 3t213t^2 \neq 1. (shown)\qquad \text{(shown)}

Step 2: The iteration and the triple tangent property.

Define the map T(t)=1t22tT(t) = \frac{1-t^2}{2t}. We need t1=T(t0)t_1 = T(t_0), t2=T(t1)t_2 = T(t_1), t3=T(t2)t_3 = T(t_2).

Setting t=tanαt = \tan\alpha, we get:

T(tanα)=1tan2α2tanα=1tan2α=cot2α=tan ⁣(π22α).T(\tan\alpha) = \frac{1-\tan^2\alpha}{2\tan\alpha} = \frac{1}{\tan 2\alpha} = \cot 2\alpha = \tan\!\left(\frac{\pi}{2} - 2\alpha\right).

So if t=tanαt = \tan\alpha, then T(t)=tan ⁣(π22α)T(t) = \tan\!\left(\frac{\pi}{2} - 2\alpha\right).

Let t0=tanα0t_0 = \tan\alpha_0 with α0=7π18\alpha_0 = \frac{7\pi}{18}.

t1=T(t0)=tan ⁣(π214π18)=tan ⁣(9π14π18)=tan ⁣(5π18).t_1 = T(t_0) = \tan\!\left(\frac{\pi}{2} - \frac{14\pi}{18}\right) = \tan\!\left(\frac{9\pi - 14\pi}{18}\right) = \tan\!\left(-\frac{5\pi}{18}\right).

So α1=5π18\alpha_1 = -\frac{5\pi}{18} (taking the angle modulo π\pi for tan\tan).

t2=T(t1)=tan ⁣(π22(5π18))=tan ⁣(π2+10π18)=tan ⁣(9π+10π18)=tan ⁣(19π18).t_2 = T(t_1) = \tan\!\left(\frac{\pi}{2} - 2\cdot\left(-\frac{5\pi}{18}\right)\right) = \tan\!\left(\frac{\pi}{2} + \frac{10\pi}{18}\right) = \tan\!\left(\frac{9\pi + 10\pi}{18}\right) = \tan\!\left(\frac{19\pi}{18}\right).

Since tan\tan has period π\pi: tan19π18=tanπ18\tan\frac{19\pi}{18} = \tan\frac{\pi}{18}. So α2=π18\alpha_2 = \frac{\pi}{18}.

t3=T(t2)=tan ⁣(π22π18)=tan ⁣(9π2π18)=tan ⁣(7π18)=t0.t_3 = T(t_2) = \tan\!\left(\frac{\pi}{2} - \frac{2\pi}{18}\right) = \tan\!\left(\frac{9\pi - 2\pi}{18}\right) = \tan\!\left(\frac{7\pi}{18}\right) = t_0.

So P3=P0P_3 = P_0. (shown)\qquad \text{(shown)}

To show P1P0P_1 \neq P_0: we need t1t0t_1 \neq t_0, i.e., tan(5π/18)tan(7π/18)\tan(-5\pi/18) \neq \tan(7\pi/18). Since 5π/18-5\pi/18 and 7π/187\pi/18 differ by 12π/18=2π/3kπ12\pi/18 = 2\pi/3 \neq k\pi, these are not equal.

Step 3: Find a second value of t0>0t_0 > 0 with P3=P0P_3 = P_0, P1P0P_1 \neq P_0.

The condition t3=t0t_3 = t_0 means, after three applications of the angle map απ22α\alpha \mapsto \frac{\pi}{2} - 2\alpha, we return to the same value of tan\tan.

Applying the map three times starting from α\alpha:

α    π22α    π22 ⁣(π22α)=4απ2    π22 ⁣(4απ2)=3π28α.\alpha \;\mapsto\; \frac{\pi}{2} - 2\alpha \;\mapsto\; \frac{\pi}{2} - 2\!\left(\frac{\pi}{2} - 2\alpha\right) = 4\alpha - \frac{\pi}{2} \;\mapsto\; \frac{\pi}{2} - 2\!\left(4\alpha - \frac{\pi}{2}\right) = \frac{3\pi}{2} - 8\alpha.

We need tan(3π28α)=tanα\tan(\frac{3\pi}{2} - 8\alpha) = \tan\alpha, so 3π28α=α+kπ\frac{3\pi}{2} - 8\alpha = \alpha + k\pi for some integer kk.

9α=3π2kπ    α=3π2kπ18=(32k)π18.9\alpha = \frac{3\pi}{2} - k\pi \implies \alpha = \frac{3\pi - 2k\pi}{18} = \frac{(3-2k)\pi}{18}.

For P1P0P_1 \neq P_0 we need π22αα+mπ\frac{\pi}{2} - 2\alpha \neq \alpha + m\pi, i.e., π23α+mπ\frac{\pi}{2} \neq 3\alpha + m\pi, i.e., απ2mπ6\alpha \neq \frac{\pi - 2m\pi}{6}.

The values α=(32k)π18\alpha = \frac{(3-2k)\pi}{18} with t0=tanα>0t_0 = \tan\alpha > 0 (so α(0,π/2)\alpha \in (0, \pi/2) modulo π\pi):

  • k=0k = 0: α=3π18=π6\alpha = \frac{3\pi}{18} = \frac{\pi}{6}. But P1=P0P_1 = P_0 check: 3α=π/23\alpha = \pi/2, so α=π/6\alpha = \pi/6 gives P1=P0P_1 = P_0. Reject.
  • k=1k = -1: α=5π18\alpha = \frac{5\pi}{18}. Check P1P0P_1 \neq P_0: 3α=5π/6π/2+mπ3\alpha = 5\pi/6 \neq \pi/2 + m\pi. Good.
  • k=1k = 1: α=π18\alpha = \frac{\pi}{18}. Check: 3α=π/6π/2+mπ3\alpha = \pi/6 \neq \pi/2 + m\pi. Good.
  • k=2k = -2: α=7π18\alpha = \frac{7\pi}{18}. This is the original value.
  • k=3k = -3: α=9π18=π2\alpha = \frac{9\pi}{18} = \frac{\pi}{2}. tan(π/2)\tan(\pi/2) is undefined. Reject.
  • k=2k = 2: α=π18\alpha = \frac{-\pi}{18}. Negative, tanα<0\tan\alpha < 0. Reject.

So the second positive value is t0=tan5π18t_0 = \tan\frac{5\pi}{18} (and t0=tanπ18t_0 = \tan\frac{\pi}{18} is a third).

Since the question asks for “a second value,” either works. We give t0=tan5π18t_0 = \tan\frac{5\pi}{18}.

Verification: α0=5π/18\alpha_0 = 5\pi/18. α1=π/210π/18=π/18\alpha_1 = \pi/2 - 10\pi/18 = -\pi/18. α2=π/2+2π/18=11π/18\alpha_2 = \pi/2 + 2\pi/18 = 11\pi/18. α3=π/222π/18=13π/185π/18(modπ)\alpha_3 = \pi/2 - 22\pi/18 = -13\pi/18 \equiv 5\pi/18 \pmod{\pi}. So t3=t0t_3 = t_0. And t1=tan(π/18)tan(5π/18)=t0t_1 = \tan(-\pi/18) \neq \tan(5\pi/18) = t_0. \checkmark


Topic: 代数与方程理论 Algebra & Equation Theory  |  Difficulty: Challenging  |  Marks: 20

5 Find the coordinates of the turning point on the curve y=x22bx+cy = x^2 - 2bx + c. Sketch the curve in the case that the equation x22bx+c=0x^2 - 2bx + c = 0 has two distinct real roots. Use your sketch to determine necessary and sufficient conditions on bb and cc for the equation x22bx+c=0x^2 - 2bx + c = 0 to have two distinct real roots. Determine necessary and sufficient conditions on bb and cc for this equation to have two distinct positive roots.

Find the coordinates of the turning points on the curve y=x33b2x+cy = x^3 - 3b^2x + c (with b>0b > 0) and hence determine necessary and sufficient conditions on bb and cc for the equation x33b2x+c=0x^3 - 3b^2x + c = 0 to have three distinct real roots. Determine necessary and sufficient conditions on a,ba, b and cc for the equation (xa)33b2(xa)+c=0(x-a)^3 - 3b^2(x-a) + c = 0 to have three distinct positive roots.

Show that the equation 2x39x2+7x1=02x^3 - 9x^2 + 7x - 1 = 0 has three distinct positive roots.

Model Solution

Part 1: Quadratic turning point and conditions

Completing the square: y=x22bx+c=(xb)2+cb2.y = x^2 - 2bx + c = (x - b)^2 + c - b^2 .

The turning point is at (b,  cb2)(b,\; c - b^2), and since the coefficient of x2x^2 is positive, this is a minimum.

When x22bx+c=0x^2 - 2bx + c = 0 has two distinct real roots, the parabola crosses the xx-axis at two points. The sketch shows an upward-opening parabola with its minimum below the xx-axis.

For two distinct real roots, we need the minimum value to be negative: cb2<0b2>c.c - b^2 < 0 \quad \Longleftrightarrow \quad b^2 > c .

This is necessary (the minimum must be below the axis) and sufficient (an upward parabola with minimum below the axis crosses the axis twice).

For two distinct positive roots, we need:

  1. b2>cb^2 > c (two distinct real roots),
  2. The sum of roots =2b>0= 2b > 0, so b>0b > 0,
  3. The product of roots =c>0= c > 0.

Conditions 2 and 3 ensure both roots are positive (if both were negative, their sum would be negative; if they had opposite signs, their product would be negative). Hence the necessary and sufficient conditions for two distinct positive roots are: b>0,c>0,b2>c.b > 0, \quad c > 0, \quad b^2 > c .

Part 2: Cubic turning points and three distinct real roots

For y=x33b2x+cy = x^3 - 3b^2 x + c with b>0b > 0: dydx=3x23b2=3(x2b2)=3(xb)(x+b).\frac{dy}{dx} = 3x^2 - 3b^2 = 3(x^2 - b^2) = 3(x - b)(x + b) .

The turning points are at x=bx = -b (local maximum) and x=bx = b (local minimum):

  • y(b)=b3+3b3+c=2b3+cy(-b) = -b^3 + 3b^3 + c = 2b^3 + c,
  • y(b)=b33b3+c=2b3+cy(b) = b^3 - 3b^3 + c = -2b^3 + c.

For three distinct real roots, the local maximum must be above the xx-axis and the local minimum below: 2b3+c>0and2b3+c<0,2b^3 + c > 0 \quad \text{and} \quad -2b^3 + c < 0 , 2b3<c<2b3.\Longleftrightarrow \quad -2b^3 < c < 2b^3 .

Part 3: Conditions for (xa)33b2(xa)+c=0(x-a)^3 - 3b^2(x-a) + c = 0 to have three distinct positive roots

Substituting u=xau = x - a, the equation becomes u33b2u+c=0u^3 - 3b^2 u + c = 0, with roots u1<u2<u3u_1 < u_2 < u_3. The original roots are xi=ui+ax_i = u_i + a, so we need ui+a>0u_i + a > 0 for all ii, i.e., u1>au_1 > -a (the binding constraint, since u1u_1 is the smallest).

From Part 2, three distinct real roots require 2b3<c<2b3-2b^3 < c < 2b^3.

Now consider f(a)=(a)33b2(a)+c=a3+3ab2+cf(-a) = (-a)^3 - 3b^2(-a) + c = -a^3 + 3ab^2 + c. Since ff has positive leading coefficient, f(u)f(u) \to -\infty as uu \to -\infty, and ff is increasing on (,b)(-\infty, -b). The smallest root u1u_1 is where ff first crosses zero. For u1>au_1 > -a, we need f(a)<0f(-a) < 0 (so that a-a lies to the left of u1u_1 in the region where f<0f < 0): a3+3ab2+c<0c<a33ab2.-a^3 + 3ab^2 + c < 0 \quad \Longleftrightarrow \quad c < a^3 - 3ab^2 .

For a-a to lie in the interval (,b)(-\infty, -b) where ff is increasing, we need a<b-a < -b, i.e., a>ba > b. (This also ensures a33ab2=a(a23b2)a^3 - 3ab^2 = a(a^2 - 3b^2) can be compatible with c>2b3c > -2b^3.)

For the middle root: u2(b,b)u_2 \in (-b, b), so x2=u2+a>b+a>0x_2 = u_2 + a > -b + a > 0 since a>ba > b.

For the largest root: u3>bu_3 > b, so x3=u3+a>b+a>0x_3 = u_3 + a > b + a > 0 automatically.

The necessary and sufficient conditions for three distinct positive roots are: a>b>0,2b3<c<2b3,c<a33ab2.a > b > 0, \quad -2b^3 < c < 2b^3, \quad c < a^3 - 3ab^2 .

Part 4: Show 2x39x2+7x1=02x^3 - 9x^2 + 7x - 1 = 0 has three distinct positive roots

Dividing by 2: x392x2+72x12=0x^3 - \tfrac{9}{2}x^2 + \tfrac{7}{2}x - \tfrac{1}{2} = 0.

We match this with (xa)33b2(xa)+c=0(x - a)^3 - 3b^2(x - a) + c = 0. Expanding: (xa)33b2(xa)+c=x33ax2+(3a23b2)x+(a3+3ab2+c).(x-a)^3 - 3b^2(x-a) + c = x^3 - 3ax^2 + (3a^2 - 3b^2)x + (-a^3 + 3ab^2 + c) .

Matching coefficients:

  • x2x^2: 3a=923a = \tfrac{9}{2}, so a=32a = \tfrac{3}{2}.
  • xx: 3a23b2=723a^2 - 3b^2 = \tfrac{7}{2}, so 3943b2=723 \cdot \tfrac{9}{4} - 3b^2 = \tfrac{7}{2}, giving 3b2=274144=1343b^2 = \tfrac{27}{4} - \tfrac{14}{4} = \tfrac{13}{4}, hence b2=1312b^2 = \tfrac{13}{12}.
  • Constant: a3+3ab2+c=12-a^3 + 3ab^2 + c = -\tfrac{1}{2}, so 278+921312+c=12-\tfrac{27}{8} + \tfrac{9}{2} \cdot \tfrac{13}{12} + c = -\tfrac{1}{2}, giving 278+398+c=12-\tfrac{27}{8} + \tfrac{39}{8} + c = -\tfrac{1}{2}, so c=12128=2c = -\tfrac{1}{2} - \tfrac{12}{8} = -2.

Now we check the conditions from Part 3 with a=32a = \tfrac{3}{2}, b2=1312b^2 = \tfrac{13}{12}, c=2c = -2:

Check a>ba > b: a2=94=2712a^2 = \tfrac{9}{4} = \tfrac{27}{12} and b2=1312b^2 = \tfrac{13}{12}, so a2>b2a^2 > b^2, hence a>ba > b. ✓

Check 2b3<c<2b3-2b^3 < c < 2b^3: We have 2b3=2(b2)3/2=2(1312)3/22b^3 = 2(b^2)^{3/2} = 2\left(\tfrac{13}{12}\right)^{3/2}. Since 1312>1\tfrac{13}{12} > 1, we get 2b3>22b^3 > 2. More precisely, b3=13131212=133972b^3 = \tfrac{13\sqrt{13}}{12\sqrt{12}} = \tfrac{13\sqrt{39}}{72}, so 2b3=1339362b^3 = \tfrac{13\sqrt{39}}{36}. Since 39>6\sqrt{39} > 6, we have 2b3>7836=136>22b^3 > \tfrac{78}{36} = \tfrac{13}{6} > 2. Thus c=2<2b3|c| = 2 < 2b^3. ✓

Check c<a33ab2c < a^3 - 3ab^2: a33ab2=278921312=27811724=8111724=3624=32a^3 - 3ab^2 = \tfrac{27}{8} - \tfrac{9}{2} \cdot \tfrac{13}{12} = \tfrac{27}{8} - \tfrac{117}{24} = \tfrac{81 - 117}{24} = -\tfrac{36}{24} = -\tfrac{3}{2}. So c=2<32c = -2 < -\tfrac{3}{2}. ✓

All conditions are satisfied, so 2x39x2+7x1=02x^3 - 9x^2 + 7x - 1 = 0 has three distinct positive roots. \blacksquare


Topic: 三角函数 Trigonometry  |  Difficulty: Challenging  |  Marks: 20

6 Show that 2sin12θcosrθ=sin(r+12)θsin(r12)θ.2 \sin \frac{1}{2} \theta \cos r\theta = \sin (r + \frac{1}{2})\theta - \sin (r - \frac{1}{2})\theta . Hence, or otherwise, find all solutions of the equation cosaθ+cos(a+1)θ++cos(b2)θ+cos(b1)θ=0,\cos a\theta + \cos(a + 1)\theta + \cdots + \cos(b - 2)\theta + \cos(b - 1)\theta = 0 , where aa and bb are positive integers with a<b1a < b - 1.

Model Solution

Step 1: Show the trigonometric identity.

Using the compound angle formulae:

sin(r+12)θ=sinrθcos12θ+cosrθsin12θ,\sin(r+\tfrac{1}{2})\theta = \sin r\theta\cos\tfrac{1}{2}\theta + \cos r\theta\sin\tfrac{1}{2}\theta,

sin(r12)θ=sinrθcos12θcosrθsin12θ.\sin(r-\tfrac{1}{2})\theta = \sin r\theta\cos\tfrac{1}{2}\theta - \cos r\theta\sin\tfrac{1}{2}\theta.

Subtracting:

sin(r+12)θsin(r12)θ=2cosrθsin12θ.(shown)\sin(r+\tfrac{1}{2})\theta - \sin(r-\tfrac{1}{2})\theta = 2\cos r\theta\sin\tfrac{1}{2}\theta. \qquad \text{(shown)}

Step 2: Apply the identity to the sum.

Let S=cosaθ+cos(a+1)θ++cos(b1)θS = \cos a\theta + \cos(a+1)\theta + \cdots + \cos(b-1)\theta.

Multiplying both sides by 2sin12θ2\sin\frac{1}{2}\theta and using the identity with each r=a,a+1,,b1r = a, a+1, \ldots, b-1:

2sin12θS=r=ab1[sin(r+12)θsin(r12)θ].2\sin\tfrac{1}{2}\theta \cdot S = \sum_{r=a}^{b-1}\bigl[\sin(r+\tfrac{1}{2})\theta - \sin(r-\tfrac{1}{2})\theta\bigr].

This is a telescoping sum. Writing out the terms:

=[sin(a+12)θsin(a12)θ]+[sin(a+32)θsin(a+12)θ]+= \bigl[\sin(a+\tfrac{1}{2})\theta - \sin(a-\tfrac{1}{2})\theta\bigr] + \bigl[\sin(a+\tfrac{3}{2})\theta - \sin(a+\tfrac{1}{2})\theta\bigr] + \cdots +[sin(b12)θsin(b32)θ].\cdots + \bigl[\sin(b-\tfrac{1}{2})\theta - \sin(b-\tfrac{3}{2})\theta\bigr].

After cancellation:

2sin12θS=sin(b12)θsin(a12)θ.2\sin\tfrac{1}{2}\theta \cdot S = \sin(b-\tfrac{1}{2})\theta - \sin(a-\tfrac{1}{2})\theta.

Therefore:

S=sin(b12)θsin(a12)θ2sin12θ,S = \frac{\sin(b-\frac{1}{2})\theta - \sin(a-\frac{1}{2})\theta}{2\sin\frac{1}{2}\theta},

provided sin12θ0\sin\frac{1}{2}\theta \neq 0, i.e., θ2kπ\theta \neq 2k\pi for integer kk.

Step 3: Solve S=0S = 0.

We need sin(b12)θ=sin(a12)θ\sin(b-\frac{1}{2})\theta = \sin(a-\frac{1}{2})\theta, with θ2kπ\theta \neq 2k\pi.

Using the identity sinA=sinB\sin A = \sin B implies A=B+2kπA = B + 2k\pi or A=πB+2kπA = \pi - B + 2k\pi:

Case 1: (b12)θ=(a12)θ+2kπ(b - \frac{1}{2})\theta = (a - \frac{1}{2})\theta + 2k\pi, giving (ba)θ=2kπ(b-a)\theta = 2k\pi, so:

θ=2kπba,kZ.\theta = \frac{2k\pi}{b-a}, \qquad k \in \mathbb{Z}.

We exclude θ=2kπ\theta = 2k'\pi (i.e., kk a multiple of bab-a), since those make sin12θ=0\sin\frac{1}{2}\theta = 0.

So θ=2kπba\theta = \frac{2k\pi}{b-a} for integer kk with (ba)k(b-a) \nmid k.

Case 2: (b12)θ=π(a12)θ+2kπ(b - \frac{1}{2})\theta = \pi - (a - \frac{1}{2})\theta + 2k\pi, giving (a+b1)θ=(2k+1)π(a + b - 1)\theta = (2k+1)\pi, so:

θ=(2k+1)πa+b1,kZ.\theta = \frac{(2k+1)\pi}{a+b-1}, \qquad k \in \mathbb{Z}.

We exclude values where sin12θ=0\sin\frac{1}{2}\theta = 0, which requires (2k+1)π2(a+b1)=mπ\frac{(2k+1)\pi}{2(a+b-1)} = m\pi, i.e., 2k+1=2m(a+b1)2k+1 = 2m(a+b-1). Since 2k+12k+1 is odd and 2m(a+b1)2m(a+b-1) is even, this is impossible. So no exclusions are needed in Case 2.

Complete solution set:

θ=2kπba(kZ,  (ba)k),orθ=(2k+1)πa+b1(kZ).\theta = \frac{2k\pi}{b-a} \quad (k \in \mathbb{Z},\; (b-a) \nmid k), \qquad \text{or} \qquad \theta = \frac{(2k+1)\pi}{a+b-1} \quad (k \in \mathbb{Z}).


Topic: 坐标几何 Coordinate Geometry  |  Difficulty: Challenging  |  Marks: 20

7 In the xxyy plane, the point AA has coordinates (a,0)(a, 0) and the point BB has coordinates (0,b)(0, b), where aa and bb are positive. The point PP, which is distinct from AA and BB, has coordinates (s,t)(s, t). XX and YY are the feet of the perpendiculars from PP to the xx-axis and yy-axis respectively, and NN is the foot of the perpendicular from PP to the line ABAB. Show that the coordinates (x,y)(x, y) of NN are given by x=ab2a(btas)a2+b2,y=a2b+b(btas)a2+b2.x = \frac{ab^2 - a(bt - as)}{a^2 + b^2} , \quad y = \frac{a^2b + b(bt - as)}{a^2 + b^2} . Show that, if (tbs)(tsa)=1\left( \frac{t - b}{s} \right) \left( \frac{t}{s - a} \right) = -1, then NN lies on the line XYXY. Give a geometrical interpretation of this result.

Model Solution

Step 1: Find the coordinates of NN.

The line ABAB passes through A(a,0)A(a,0) and B(0,b)B(0,b). Its equation is xa+yb=1\frac{x}{a} + \frac{y}{b} = 1, or equivalently bx+ay=abbx + ay = ab.

A direction vector for ABAB is (a,b)(-a, b), so a normal vector to ABAB is (b,a)(b, a).

The foot of the perpendicular from P(s,t)P(s,t) to the line bx+ay=abbx + ay = ab can be found by parametrising the perpendicular through PP. The perpendicular has direction (b,a)(b, a), so a general point on it is (s+λb,  t+λa)(s + \lambda b,\; t + \lambda a).

Substituting into bx+ay=abbx + ay = ab:

b(s+λb)+a(t+λa)=ab,b(s + \lambda b) + a(t + \lambda a) = ab,

bs+λb2+at+λa2=ab,bs + \lambda b^2 + at + \lambda a^2 = ab,

λ(a2+b2)=abbsat,\lambda(a^2 + b^2) = ab - bs - at,

λ=abbsata2+b2=a(btas)(abab)+abbsata2+b2.\lambda = \frac{ab - bs - at}{a^2 + b^2} = \frac{-a(bt - as) - (ab - ab) + ab - bs - at}{a^2 + b^2}.

More directly: λ=abbsata2+b2\lambda = \frac{ab - bs - at}{a^2 + b^2}.

The coordinates of NN are:

x=s+λb=s+b(abbsat)a2+b2=s(a2+b2)+b(abbsat)a2+b2,x = s + \lambda b = s + \frac{b(ab - bs - at)}{a^2 + b^2} = \frac{s(a^2 + b^2) + b(ab - bs - at)}{a^2 + b^2},

=a2s+b2s+ab2b2sabta2+b2=a2s+ab2abta2+b2=a(as+b2bt)a2+b2.= \frac{a^2 s + b^2 s + ab^2 - b^2 s - abt}{a^2 + b^2} = \frac{a^2 s + ab^2 - abt}{a^2 + b^2} = \frac{a(as + b^2 - bt)}{a^2 + b^2}.

=a[b2(btas)]a2+b2=ab2a(btas)a2+b2.(shown)= \frac{a\bigl[b^2 - (bt - as)\bigr]}{a^2 + b^2} = \frac{ab^2 - a(bt - as)}{a^2 + b^2}. \qquad \text{(shown)}

Similarly:

y=t+λa=t+a(abbsat)a2+b2=t(a2+b2)+a(abbsat)a2+b2,y = t + \lambda a = t + \frac{a(ab - bs - at)}{a^2 + b^2} = \frac{t(a^2 + b^2) + a(ab - bs - at)}{a^2 + b^2},

=a2t+b2t+a2babsa2ta2+b2=b2t+a2babsa2+b2=b(bt+a2as)a2+b2.= \frac{a^2 t + b^2 t + a^2 b - abs - a^2 t}{a^2 + b^2} = \frac{b^2 t + a^2 b - abs}{a^2 + b^2} = \frac{b(bt + a^2 - as)}{a^2 + b^2}.

=a2b+b(btas)a2+b2.(shown)= \frac{a^2 b + b(bt - as)}{a^2 + b^2}. \qquad \text{(shown)}

Step 2: Show NN lies on XYXY under the given condition.

XX is the foot of the perpendicular from PP to the xx-axis, so X=(s,0)X = (s, 0). YY is the foot of the perpendicular from PP to the yy-axis, so Y=(0,t)Y = (0, t).

The line XYXY passes through (s,0)(s, 0) and (0,t)(0, t), with equation:

xs+yt=1,i.e.,tx+sy=st.\frac{x}{s} + \frac{y}{t} = 1, \qquad \text{i.e.,} \quad tx + sy = st.

NN lies on XYXY if and only if txN+syN=sttx_N + sy_N = st. Substituting:

tab2a(btas)a2+b2+sa2b+b(btas)a2+b2=st.t\cdot\frac{ab^2 - a(bt - as)}{a^2 + b^2} + s\cdot\frac{a^2 b + b(bt - as)}{a^2 + b^2} = st.

Multiply through by a2+b2a^2 + b^2:

t[ab2a(btas)]+s[a2b+b(btas)]=st(a2+b2).t\bigl[ab^2 - a(bt - as)\bigr] + s\bigl[a^2 b + b(bt - as)\bigr] = st(a^2 + b^2).

Expand the left side:

tab2at(btas)+sa2b+sb(btas),tab^2 - at(bt - as) + sa^2 b + sb(bt - as),

=tab2abt2+a2st+a2bs+b2stabs2.= tab^2 - abt^2 + a^2 st + a^2 bs + b^2 st - abs^2.

The right side is sta2+stb2sta^2 + stb^2.

Subtracting sta2+stb2sta^2 + stb^2 from both sides:

tab2abt2+a2bsabs2=0,tab^2 - abt^2 + a^2 bs - abs^2 = 0,

ab(tat2+ass2)=0,ab(ta - t^2 + as - s^2) = 0,

ab[t(at)+s(as)]=0.ab\bigl[t(a - t) + s(a - s)\bigr] = 0.

Hmm, let me recheck. Actually:

tab2abt2+a2st+a2bs+b2stabs2sta2stb2tab^2 - abt^2 + a^2 st + a^2 bs + b^2 st - abs^2 - sta^2 - stb^2 =tab2abt2+a2bsabs2= tab^2 - abt^2 + a^2 bs - abs^2 =ab(tabt2+a2s/bbs2b)= ab(tab - t^2 + a^2 s/b \cdot b - s^2 \cdot b)

Wait, let me factor more carefully:

tab2abt2+a2bsabs2=ab(tbt2+ass2)=ab[t(bt)+s(as)].tab^2 - abt^2 + a^2 bs - abs^2 = ab(tb - t^2 + as - s^2) = ab\bigl[t(b-t) + s(a-s)\bigr].

For this to equal zero (with a,b>0a, b > 0):

t(bt)+s(as)=0.t(b-t) + s(a-s) = 0.

Now consider the given condition: (tbs)(tsa)=1\left(\frac{t-b}{s}\right)\left(\frac{t}{s-a}\right) = -1.

t(tb)s(sa)=1    t(tb)=s(sa)    t2bt=s2+as.\frac{t(t-b)}{s(s-a)} = -1 \implies t(t-b) = -s(s-a) \implies t^2 - bt = -s^2 + as.

t2bt+s2as=0    t(bt)+s(as)=(t2bt)(s2as)=[(t2bt)+(s2as)]=0.t^2 - bt + s^2 - as = 0 \implies t(b-t) + s(a-s) = -(t^2 - bt) - (s^2 - as) = -\bigl[(t^2 - bt) + (s^2 - as)\bigr] = 0.

Wait: t(bt)=btt2t(b-t) = bt - t^2 and s(as)=ass2s(a-s) = as - s^2. So t(bt)+s(as)=btt2+ass2t(b-t) + s(a-s) = bt - t^2 + as - s^2. The condition gives t2bt+s2as=0t^2 - bt + s^2 - as = 0, i.e., btt2+ass2=0bt - t^2 + as - s^2 = 0. Yes, so t(bt)+s(as)=0t(b-t) + s(a-s) = 0. \checkmark

Therefore NN lies on XYXY. (shown)\qquad \text{(shown)}

Step 3: Geometrical interpretation.

The condition tbstsa=1\frac{t-b}{s} \cdot \frac{t}{s-a} = -1 means tbs0t0sa=1\frac{t-b}{s-0} \cdot \frac{t-0}{s-a} = -1. The first factor is the gradient of PBPB (since B=(0,b)B = (0,b) and P=(s,t)P = (s,t)), and the second factor is the gradient of PAPA (since A=(a,0)A = (a,0)). The product of gradients being 1-1 means PAPBPA \perp PB, i.e., the angle APB=90°\angle APB = 90°.

So PP lies on the circle with diameter ABAB (by Thales’ theorem).

Geometrical interpretation: NN lies on the line XYXY if and only if PP lies on the circle with ABAB as diameter, i.e., APB\angle APB is a right angle.


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

8 (i) Show that the gradient at a point (x,y)(x, y) on the curve (y+2x)3(y4x)=c,(y + 2x)^3 (y - 4x) = c , where cc is a constant, is given by dydx=16xy2y5x.\frac{dy}{dx} = \frac{16x - y}{2y - 5x} . (ii) By considering the derivative with respect to xx of (y+ax)n(y+bx)(y + ax)^n (y + bx), or otherwise, find the general solution of the differential equation dydx=10x4y3xy.\frac{dy}{dx} = \frac{10x - 4y}{3x - y} .

Model Solution

Part (i): Show the gradient of (y+2x)3(y4x)=c(y+2x)^3(y-4x) = c is 16xy2y5x\frac{16x-y}{2y-5x}.

Differentiate implicitly with respect to xx:

3(y+2x)2(y+2)(y4x)+(y+2x)3(y4)=0.3(y+2x)^2(y'+2)\cdot(y-4x) + (y+2x)^3(y'-4) = 0.

Divide by (y+2x)2(y+2x)^2 (which is nonzero since c0c \neq 0 implies y+2x0y+2x \neq 0):

3(y+2)(y4x)+(y+2x)(y4)=0.3(y'+2)(y-4x) + (y+2x)(y'-4) = 0.

Expand:

3y(y4x)+6(y4x)+y(y+2x)4(y+2x)=0.3y'(y-4x) + 6(y-4x) + y'(y+2x) - 4(y+2x) = 0.

Collect yy' terms:

y[3(y4x)+(y+2x)]+6(y4x)4(y+2x)=0,y'\bigl[3(y-4x) + (y+2x)\bigr] + 6(y-4x) - 4(y+2x) = 0,

y(3y12x+y+2x)+6y24x4y8x=0,y'(3y - 12x + y + 2x) + 6y - 24x - 4y - 8x = 0,

y(4y10x)+2y32x=0.y'(4y - 10x) + 2y - 32x = 0.

y=32x2y4y10x=2(16xy)2(2y5x)=16xy2y5x.(shown)y' = \frac{32x - 2y}{4y - 10x} = \frac{2(16x - y)}{2(2y - 5x)} = \frac{16x - y}{2y - 5x}. \qquad \text{(shown)}

Part (ii): Solve dydx=10x4y3xy\frac{dy}{dx} = \frac{10x - 4y}{3x - y}.

We consider F(x,y)=(y+ax)n(y+bx)F(x,y) = (y+ax)^n(y+bx) and compute dFdx=0\frac{\mathrm{d}F}{\mathrm{d}x} = 0.

dFdx=n(y+ax)n1(y+a)(y+bx)+(y+ax)n(y+b)=0.\frac{\mathrm{d}F}{\mathrm{d}x} = n(y+ax)^{n-1}(y'+a)(y+bx) + (y+ax)^n(y'+b) = 0.

Divide by (y+ax)n1(y+ax)^{n-1}:

n(y+a)(y+bx)+(y+ax)(y+b)=0.n(y'+a)(y+bx) + (y+ax)(y'+b) = 0.

Expand and collect yy':

y[n(y+bx)+(y+ax)]+na(y+bx)+b(y+ax)=0.y'\bigl[n(y+bx) + (y+ax)\bigr] + na(y+bx) + b(y+ax) = 0.

y=na(y+bx)+b(y+ax)n(y+bx)+(y+ax)=(na+b)y+nabx+bax(n+1)y+(nb+a)x.y' = -\frac{na(y+bx) + b(y+ax)}{n(y+bx) + (y+ax)} = -\frac{(na+b)y + n abx + bax}{(n+1)y + (nb+a)x}.

Hmm, let me redo more carefully:

Numerator: na(y+bx)+b(y+ax)=nay+nabx+by+abx=(na+b)y+ab(n+1)xna(y+bx) + b(y+ax) = nay + nabx + by + abx = (na+b)y + ab(n+1)x.

Denominator: n(y+bx)+(y+ax)=ny+nbx+y+ax=(n+1)y+(nb+a)xn(y+bx) + (y+ax) = ny + nbx + y + ax = (n+1)y + (nb+a)x.

So:

y=(na+b)y+ab(n+1)x(n+1)y+(nb+a)x=ab(n+1)x(na+b)y(nb+a)x+(n+1)y.y' = -\frac{(na+b)y + ab(n+1)x}{(n+1)y + (nb+a)x} = \frac{-ab(n+1)x - (na+b)y}{(nb+a)x + (n+1)y}.

We need this to equal 10x4y3xy=10x4y3xy\frac{10x - 4y}{3x - y} = \frac{10x - 4y}{3x - y}. Matching coefficients:

From the yy coefficient in the numerator: (na+b)=4-(na+b) = -4, so na+b=4na + b = 4.

From the xx coefficient in the numerator: ab(n+1)=10-ab(n+1) = 10, so ab(n+1)=10ab(n+1) = -10.

From the xx coefficient in the denominator: nb+a=3nb + a = 3.

From the yy coefficient in the denominator: n+1=1n + 1 = -1, so n=2n = -2.

Check: with n=2n = -2:

  • na+b=2a+b=4na + b = -2a + b = 4
  • nb+a=2b+a=3nb + a = -2b + a = 3
  • ab(n+1)=ab(1)=ab=10ab(n+1) = ab(-1) = -ab = -10, so ab=10ab = 10.

From 2a+b=4-2a + b = 4: b=4+2ab = 4 + 2a. Substitute into ab=10ab = 10: a(4+2a)=10a(4+2a) = 10, so 2a2+4a10=02a^2 + 4a - 10 = 0, i.e., a2+2a5=0a^2 + 2a - 5 = 0.

a=2±4+202=1±6.a = \frac{-2 \pm \sqrt{4+20}}{2} = -1 \pm \sqrt{6}.

Check with 2b+a=3-2b + a = 3: if a=1+6a = -1+\sqrt{6}, then b=4+2(1+6)=2+26b = 4 + 2(-1+\sqrt{6}) = 2+2\sqrt{6}.

2b+a=2(2+26)+(1+6)=4461+6=536-2b + a = -2(2+2\sqrt{6}) + (-1+\sqrt{6}) = -4 - 4\sqrt{6} - 1 + \sqrt{6} = -5 - 3\sqrt{6}. This should be 33. Doesn’t work.

If a=16a = -1-\sqrt{6}, then b=4+2(16)=226b = 4 + 2(-1-\sqrt{6}) = 2-2\sqrt{6}.

2b+a=2(226)+(16)=4+4616=5+36-2b + a = -2(2-2\sqrt{6}) + (-1-\sqrt{6}) = -4+4\sqrt{6}-1-\sqrt{6} = -5+3\sqrt{6}. This should be 33. So 36=83\sqrt{6} = 8, 6=8/3\sqrt{6} = 8/3. Not true.

Let me recheck the coefficient matching. We have:

y=ab(n+1)x(na+b)y(nb+a)x+(n+1)y.y' = \frac{-ab(n+1)x - (na+b)y}{(nb+a)x + (n+1)y}.

And we want y=10x4y3xyy' = \frac{10x - 4y}{3x - y}.

So: ab(n+1)nb+a=103\frac{-ab(n+1)}{nb+a} = \frac{10}{3} and (na+b)n+1=41=4\frac{-(na+b)}{n+1} = \frac{-4}{-1} = 4.

Wait, I need to be more careful. We need:

ab(n+1)x(na+b)y(nb+a)x+(n+1)y=10x4y3xy.\frac{-ab(n+1)x - (na+b)y}{(nb+a)x + (n+1)y} = \frac{10x - 4y}{3x - y}.

So the ratios must match: ab(n+1)10=(na+b)4=nb+a3=n+11\frac{-ab(n+1)}{10} = \frac{-(na+b)}{-4} = \frac{nb+a}{3} = \frac{n+1}{-1}.

From the last ratio: n+1=kn+1 = -k for some proportionality constant kk. Actually, let me just set:

ab(n+1)=10k,(na+b)=4k,nb+a=3k,n+1=k.-ab(n+1) = 10k, \quad -(na+b) = -4k, \quad nb+a = 3k, \quad n+1 = -k.

From n+1=kn+1 = -k: k=(n+1)k = -(n+1).

(na+b)=4k=4(n+1)-(na+b) = -4k = 4(n+1), so na+b=4(n+1)na+b = -4(n+1).

nb+a=3k=3(n+1)nb+a = 3k = -3(n+1).

ab(n+1)=10k=10(n+1)-ab(n+1) = 10k = -10(n+1), so ab=10ab = 10 (assuming n+10n+1 \neq 0).

From na+b=4(n+1)na + b = -4(n+1) and nb+a=3(n+1)nb + a = -3(n+1):

Subtract: na+bnba=(n+1)na + b - nb - a = -(n+1), so (n1)a(n1)b=(n+1)(n-1)a - (n-1)b = -(n+1), i.e., (n1)(ab)=(n+1)(n-1)(a-b) = -(n+1).

From the second equation: a=3(n+1)nba = -3(n+1) - nb. Substitute into the first: n(3(n+1)nb)+b=4(n+1)n(-3(n+1) - nb) + b = -4(n+1).

3n(n+1)n2b+b=4(n+1),-3n(n+1) - n^2 b + b = -4(n+1),

b(1n2)=4(n+1)+3n(n+1)=(n+1)(3n4),b(1 - n^2) = -4(n+1) + 3n(n+1) = (n+1)(3n-4),

b(1n)(1+n)=(n+1)(3n4).b(1-n)(1+n) = (n+1)(3n-4).

Assuming n+10n+1 \neq 0: b(1n)=3n4b(1-n) = 3n-4, so b=3n41n=43nn1b = \frac{3n-4}{1-n} = \frac{4-3n}{n-1}.

Then a=3(n+1)nb=3(n+1)n43nn1a = -3(n+1) - nb = -3(n+1) - n\cdot\frac{4-3n}{n-1}.

=3(n+1)n(43n)n1=3(n+1)(n1)n(43n)n1= -3(n+1) - \frac{n(4-3n)}{n-1} = \frac{-3(n+1)(n-1) - n(4-3n)}{n-1}

=3(n21)4n+3n2n1=3n2+34n+3n2n1=34nn1= \frac{-3(n^2-1) - 4n + 3n^2}{n-1} = \frac{-3n^2+3-4n+3n^2}{n-1} = \frac{3-4n}{n-1}.

And ab=10ab = 10:

(34n)(43n)(n1)2=10.\frac{(3-4n)(4-3n)}{(n-1)^2} = 10.

(34n)(43n)=129n16n+12n2=12n225n+12(3-4n)(4-3n) = 12 - 9n - 16n + 12n^2 = 12n^2 - 25n + 12.

So: 12n225n+12=10(n1)2=10n220n+1012n^2 - 25n + 12 = 10(n-1)^2 = 10n^2 - 20n + 10.

2n25n+2=0    (2n1)(n2)=0.2n^2 - 5n + 2 = 0 \implies (2n-1)(n-2) = 0.

So n=12n = \frac{1}{2} or n=2n = 2.

Case n=2n = 2: b=461=2b = \frac{4-6}{1} = -2, a=381=5a = \frac{3-8}{1} = -5. Check: ab=10ab = 10. \checkmark

So (y5x)2(y2x)=C(y - 5x)^2(y - 2x) = C.

Case n=12n = \frac{1}{2}: b=43/21/2=5/21/2=5b = \frac{4 - 3/2}{-1/2} = \frac{5/2}{-1/2} = -5, a=321/2=11/2=2a = \frac{3-2}{-1/2} = \frac{1}{-1/2} = -2. Check: ab=10ab = 10. \checkmark

So (y2x)1/2(y5x)=C(y - 2x)^{1/2}(y - 5x) = C, which is equivalent (square both sides: (y2x)(y5x)2=C2(y-2x)(y-5x)^2 = C^2).

Both give the same family of curves. The general solution is:

(y5x)2(y2x)=C\boxed{(y - 5x)^2(y - 2x) = C}

where CC is an arbitrary constant.