← All Units← 返回单元列表 ← Course Hub← 课程主页
U N I V E R S I T Y  C A L C U L U S
Unit D2 · Solutions第D2单元 · 解答

First-Order Models and Exact Equations · Solutions一阶模型与恰当方程 · 解答

Companion to the University-Style Practice Set大学风格练习题配套解答

MEDIUM HARD CORE PROOF APPLIED

Sections 1 to 7: exact equations, integrating factors, Bernoulli and homogeneous substitution, autonomous equations, logistic growth, mixing models, Newton cooling1 至 7 节:恰当方程、积分因子、Bernoulli 换元与齐次换元、自治方程、逻辑斯谛增长、混合模型、牛顿冷却定律CALC IV



PART I  ·  CORE TECHNIQUESComputational fluency · 28 marks计算熟练度 · 28分

Worked Solutions详细解答

Q1MEDIUMCOREtesting exactness and finding the potential验证恰当性并求势函数[8 marks]

Test exactness and find the general solution for: (a) $(y^2+2xy)\,dx+(2xy+x^2)\,dy=0$; (b) $(3x^2y)\,dx+(x^3+2y)\,dy=0$; (c) show $(2xy+y^2)\,dx+(x^2-y)\,dy=0$ is not exact.验证恰当性并求通解:(a) $(y^2+2xy)\,dx+(2xy+x^2)\,dy=0$;(b) $(3x^2y)\,dx+(x^3+2y)\,dy=0$;(c) 证明 $(2xy+y^2)\,dx+(x^2-y)\,dy=0$ 不是恰当方程。

Answers:答案:  (a) $xy^2+x^2y=C$  ·  (b) $x^3y+y^2=C$  ·  (c) $M_y-N_x=2y\ne0$

(a) Exact: integrate $M$ in $x$(a) 恰当方程:对 $M$ 关于 $x$ 积分 M1·A1·A1

Set $M=y^2+2xy$ and $N=2xy+x^2$. Compute the mixed partials: $M_y=2y+2x$ and $N_x=2y+2x$. (M1) Since $M_y=N_x$ on all of $\mathbb{R}^2$, the equation is exact.令 $M=y^2+2xy$,$N=2xy+x^2$。计算混合偏导数:$M_y=2y+2x$,$N_x=2y+2x$。(M1) 由于在整个 $\mathbb{R}^2$ 上 $M_y=N_x$,该方程是恰当方程。

Integrate $M$ in $x$ (holding $y$ fixed): $F=\displaystyle\int(y^2+2xy)\,dx=xy^2+x^2y+g(y)$. Differentiate in $y$ and match $F_y=N$: $2xy+x^2+g'(y)=2xy+x^2$, so $g'(y)=0$ and $g$ is a constant absorbed into $C$. (A1)对 $M$ 关于 $x$ 积分(固定 $y$):$F=\displaystyle\int(y^2+2xy)\,dx=xy^2+x^2y+g(y)$。对 $y$ 求偏导并匹配 $F_y=N$:$2xy+x^2+g'(y)=2xy+x^2$,故 $g'(y)=0$,$g$ 为常数可并入 $C$。(A1)

$$ F(x,y)=xy^2+x^2y=C. $$

Verification: $\nabla F=(y^2+2xy,\,2xy+x^2)=(M,N)$. (A1)验证:$\nabla F=(y^2+2xy,\,2xy+x^2)=(M,N)$。(A1)

(b) Exact: potential by the same method(b) 恰当方程:用相同方法求势函数 M1·A1·A1

$M=3x^2y$, $M_y=3x^2$; $N=x^3+2y$, $N_x=3x^2$. Equal, so exact. (M1)$M=3x^2y$,$M_y=3x^2$;$N=x^3+2y$,$N_x=3x^2$。二者相等,故为恰当方程。(M1)

$F=\displaystyle\int 3x^2y\,dx=x^3y+g(y)$. Then $F_y=x^3+g'(y)=x^3+2y$, giving $g'(y)=2y$ and $g(y)=y^2$. (A1)$F=\displaystyle\int 3x^2y\,dx=x^3y+g(y)$。则 $F_y=x^3+g'(y)=x^3+2y$,得 $g'(y)=2y$,$g(y)=y^2$。(A1)

$$ F(x,y)=x^3y+y^2=C. $$

Verification: $F_x=3x^2y=M$ and $F_y=x^3+2y=N$. (A1)验证:$F_x=3x^2y=M$,$F_y=x^3+2y=N$。(A1)

(c) Not exact: compute the obstruction(c) 非恰当方程:计算障碍项 M1·A1

$M=2xy+y^2$, $M_y=2x+2y$; $N=x^2-y$, $N_x=2x$. (M1) Then$M=2xy+y^2$,$M_y=2x+2y$;$N=x^2-y$,$N_x=2x$。(M1) 则

$$ M_y-N_x=(2x+2y)-2x=2y\ne 0 $$

in general, so the equation is not exact. (A1) The non-zero obstruction $2y$ signals that no potential exists as written; an integrating factor would be needed.一般情况下不为零,故该方程不是恰当方程。(A1) 非零障碍项 $2y$ 说明如原式所写不存在势函数,需引入积分因子。

Insight.解题要点。 Always test $M_y=N_x$ before attempting to build $F$. In (a) the mixed partials match instantly because $xy^2+x^2y$ is visibly a single potential; in (b) the test still works cleanly even though the potential is less obvious. Part (c) shows the failure mode: $M_y-N_x=2y$ depends on $y$, which prevents a $\mu(x)$-only factor (since $(M_y-N_x)/N$ would not simplify to a function of $x$ alone here). Check for a $y$-only factor as the next step.在尝试构造势函数 $F$ 之前,务必先验证 $M_y=N_x$。(a) 中混合偏导数立即吻合,因为 $xy^2+x^2y$ 显然是一个势函数;(b) 中即使势函数不那么直观,判据仍然整洁有效。(c) 展示了失败情形:$M_y-N_x=2y$ 依赖于 $y$,这使得仅关于 $x$ 的积分因子 $\mu(x)$ 不可行(因为 $(M_y-N_x)/N$ 无法化简为仅含 $x$ 的函数)。下一步应检验是否存在仅关于 $y$ 的积分因子。
Q2MEDIUMCOREexact IVP integrating in $y$ first恰当方程初值问题:先对 $y$ 积分[10 marks]

Solve $(3x^2y+2xy+y^3)\,dx+(x^3+x^2+3xy^2)\,dy=0$, $y(1)=2$. Integrate $N$ in $y$ first.求解 $(3x^2y+2xy+y^3)\,dx+(x^3+x^2+3xy^2)\,dy=0$,$y(1)=2$。先对 $N$ 关于 $y$ 积分。

Answers:答案:  (a) $M_y=N_x=3x^2+2x+3y^2$  ·  (b) $F=x^3y+x^2y+xy^3$  ·  (c) $x^3y+x^2y+xy^3=12$

(a) Verify exactness(a) 验证恰当性 M1·A1

Set $M=3x^2y+2xy+y^3$ and $N=x^3+x^2+3xy^2$. Differentiate: (M1)令 $M=3x^2y+2xy+y^3$,$N=x^3+x^2+3xy^2$。求偏导数:(M1)

$$ M_y=3x^2+2x+3y^2, \qquad N_x=3x^2+2x+3y^2. $$

They agree everywhere, so the equation is exact. (A1)二者处处相等,故该方程是恰当方程。(A1)

(b) Integrate $N$ in $y$ and find $h(x)$(b) 对 $N$ 关于 $y$ 积分并求 $h(x)$ M1·M1·M1·A1·A1

Integrate $N=x^3+x^2+3xy^2$ in $y$ (holding $x$ fixed): (M1)对 $N=x^3+x^2+3xy^2$ 关于 $y$ 积分(固定 $x$):(M1)

$$ F=\int N\,dy=x^3y+x^2y+xy^3+h(x). $$

(M1 for carrying out the integration correctly) Differentiate $F$ in $x$ and match $F_x=M$: (M1)(M1 为正确完成积分)对 $F$ 关于 $x$ 求偏导并匹配 $F_x=M$:(M1)

$$ F_x=3x^2y+2xy+y^3+h'(x)=M=3x^2y+2xy+y^3. $$

Therefore $h'(x)=0$, so $h$ is a constant absorbed into $C$. (A1) The potential is $F(x,y)=x^3y+x^2y+xy^3$. (A1)故 $h'(x)=0$,$h$ 为常数可并入 $C$。(A1) 势函数为 $F(x,y)=x^3y+x^2y+xy^3$。(A1)

Verification: $F_y=x^3+x^2+3xy^2=N$ and $F_x=3x^2y+2xy+y^3=M$. Both match.验证:$F_y=x^3+x^2+3xy^2=N$,$F_x=3x^2y+2xy+y^3=M$,两者均吻合。

(c) Apply the initial condition(c) 代入初始条件 M1·A1·A1

The general solution is $x^3y+x^2y+xy^3=C$. Apply $y(1)=2$: (M1)通解为 $x^3y+x^2y+xy^3=C$。代入 $y(1)=2$:(M1)

$$ F(1,2)=1\cdot2+1\cdot2+1\cdot8=12. $$

(A1) The particular solution is (A1)(A1) 特解为 (A1)

$$ x^3y+x^2y+xy^3=12. $$
Insight.解题要点。 Integrating $N$ in $y$ instead of $M$ in $x$ is purely a strategic choice: pick whichever integral is cleaner. The check that $h'(x)=0$ (with no $x$ surviving in the expression for $h'$) is a built-in verification that $M_y=N_x$ was correct. If anything depending on $x$ remained in $h'$, it would signal a computation error or a non-exact equation.选择先对 $N$ 关于 $y$ 积分而非对 $M$ 关于 $x$ 积分,纯粹是策略性选择:选择计算更简便的那个。验证 $h'(x)=0$(即 $h'$ 的表达式中不含 $x$)是对 $M_y=N_x$ 正确性的内置检验。若 $h'$ 中仍含 $x$ 的项,则说明存在计算错误或该方程并非恰当方程。
Q3HARDCOREintegrating factor restoring exactness利用积分因子恢复恰当性[10 marks]

$(3xy+y^2)\,dx+(x^2+xy)\,dy=0$: (a) show not exact and compute $(M_y-N_x)/N$; (b) find $\mu(x)$ and verify new exactness; (c) solve.$(3xy+y^2)\,dx+(x^2+xy)\,dy=0$:(a) 证明非恰当并计算 $(M_y-N_x)/N$;(b) 求 $\mu(x)$ 并验证新方程的恰当性;(c) 求解。

Answers:答案:  (a) $(M_y-N_x)/N=1/x$  ·  (b) $\mu=x$, verified已验证  ·  (c) $2x^3y+x^2y^2=C$

(a) Compute the obstruction ratio(a) 计算障碍比 M1·A1·A1

$M=3xy+y^2$, $M_y=3x+2y$; $N=x^2+xy$, $N_x=2x+y$. (M1) The obstruction is$M=3xy+y^2$,$M_y=3x+2y$;$N=x^2+xy$,$N_x=2x+y$。(M1) 障碍项为

$$ M_y-N_x=(3x+2y)-(2x+y)=x+y. $$

Since $M_y\ne N_x$, the equation is not exact. (A1) Divide by $N$:由于 $M_y\ne N_x$,该方程不是恰当方程。(A1) 除以 $N$:

$$ \frac{M_y-N_x}{N}=\frac{x+y}{x^2+xy}=\frac{x+y}{x(x+y)}=\frac{1}{x}, $$

which depends on $x$ alone, so an integrating factor $\mu(x)$ exists. (A1)结果仅依赖于 $x$,故存在积分因子 $\mu(x)$。(A1)

(b) Compute $\mu$ and verify(b) 计算 $\mu$ 并验证 M1·A1·M1·A1

$\mu=e^{\int(1/x)\,dx}=e^{\ln x}=x$. (M1·A1) Multiply through by $\mu=x$:$\mu=e^{\int(1/x)\,dx}=e^{\ln x}=x$。(M1·A1) 方程两边乘以 $\mu=x$:

$$ \tilde{M}=x(3xy+y^2)=3x^2y+xy^2, \qquad \tilde{N}=x(x^2+xy)=x^3+x^2y. $$

Check: $\tilde{M}_y=3x^2+2xy$ and $\tilde{N}_x=3x^2+2xy$. (M1) Equal, so the multiplied equation is exact. (A1)验证:$\tilde{M}_y=3x^2+2xy$,$\tilde{N}_x=3x^2+2xy$。(M1) 二者相等,故乘以积分因子后的方程是恰当方程。(A1)

(c) Find the potential and state the solution(c) 求势函数并写出解 M1·A1·A1

Integrate $\tilde{N}=x^3+x^2y$ in $y$: (M1)对 $\tilde{N}=x^3+x^2y$ 关于 $y$ 积分:(M1)

$$ F=\int(x^3+x^2y)\,dy=x^3y+\tfrac{1}{2}x^2y^2+h(x). $$

Match $F_x=3x^2y+xy^2+h'(x)=\tilde{M}=3x^2y+xy^2$, so $h'(x)=0$ and $h$ is constant. (A1) The general solution is匹配 $F_x=3x^2y+xy^2+h'(x)=\tilde{M}=3x^2y+xy^2$,故 $h'(x)=0$,$h$ 为常数。(A1) 通解为

$$ x^3y+\tfrac{1}{2}x^2y^2=C, \quad\text{or equivalently}\quad 2x^3y+x^2y^2=C. $$

Verification: $\nabla F=(3x^2y+xy^2,\,x^3+x^2y)=(\tilde{M},\tilde{N})$, confirming the potential. (A1)验证:$\nabla F=(3x^2y+xy^2,\,x^3+x^2y)=(\tilde{M},\tilde{N})$,势函数正确。(A1)

Insight.解题要点。 The ratio test $(M_y-N_x)/N=1/x$ is the standard diagnostic: if it simplifies to a function of $x$ only, a $\mu(x)$ factor exists. The $(x+y)$ factor cancels from numerator and denominator here; always simplify before concluding the ratio is not a function of $x$ alone. After multiplying by $\mu$, re-verify $\tilde{M}_y=\tilde{N}_x$ as a sanity check before spending time building $F$.比值检验 $(M_y-N_x)/N=1/x$ 是标准诊断方法:若能化简为仅含 $x$ 的函数,则存在积分因子 $\mu(x)$。此处 $(x+y)$ 因子在分子分母中约去;在得出比值不是 $x$ 的函数这一结论之前,务必先化简。乘以 $\mu$ 后,在花时间构造 $F$ 之前,再次验证 $\tilde{M}_y=\tilde{N}_x$ 作为合理性检验。
PART II  ·  DEFINITIONS AND PROOFRigorous arguments · 26 marks严格论证 · 26分

Worked Solutions详细解答

Q4HARDPROOFderiving the exactness criterion from Clairaut's theorem由 Clairaut 定理推导恰当性判据[8 marks]

(a) Prove $M_y=N_x$ is necessary for exactness using $F_{xy}=F_{yx}$. (b) Show $g'(y)=N-\partial_y\int M\,dx$ is independent of $x$ when $M_y=N_x$, so $F$ can be constructed.(a) 利用 $F_{xy}=F_{yx}$ 证明 $M_y=N_x$ 是方程为恰当方程的必要条件。(b) 在 $M_y=N_x$ 时,证明 $g'(y)=N-\partial_y\int M\,dx$ 与 $x$ 无关,从而 $F$ 可以被构造。

Answers:答案:  (a) $M_y=F_{xy}=F_{yx}=N_x$ via Clairaut由 Clairaut 定理得  ·  (b) $\partial_x(g'(y))=N_x-M_y=0$

(a) Necessity via Clairaut's theorem(a) 由 Clairaut 定理证明必要性 M1·A1·R1

Suppose $F$ exists with $F_x=M$ and $F_y=N$. Since $M$ and $N$ have continuous second partials, $F$ has continuous second partials, and Clairaut's theorem guarantees $F_{xy}=F_{yx}$. (M1) Therefore设 $F$ 存在,且 $F_x=M$,$F_y=N$。由于 $M$ 和 $N$ 具有连续的二阶偏导数,$F$ 也具有连续的二阶偏导数,Clairaut 定理保证 $F_{xy}=F_{yx}$。(M1) 因此

$$ M_y=\frac{\partial M}{\partial y}=\frac{\partial F_x}{\partial y}=F_{xy}=F_{yx}=\frac{\partial F_y}{\partial x}=\frac{\partial N}{\partial x}=N_x. $$

(A1) Hence $M_y=N_x$ is a necessary condition: if a potential exists, the mixed partials must agree. (R1)(A1) 故 $M_y=N_x$ 是必要条件:若势函数存在,则混合偏导数必须相等。(R1)

(b) Sufficiency: the key independence argument(b) 充分性:关键的独立性论证 M1·M1·A1·R1·A1

Now assume $M_y=N_x$ on a simply connected region. Integrate $M$ in $x$ (holding $y$ fixed): (M1)现设 $M_y=N_x$ 在单连通区域上成立。对 $M$ 关于 $x$ 积分(固定 $y$):(M1)

$$ F(x,y)=\int M(x,y)\,dx+g(y). $$

The unknown $g(y)$ arises because $y$ was held fixed during integration. Differentiate $F$ in $y$ and require $F_y=N$: (M1)未知函数 $g(y)$ 的出现是因为积分时 $y$ 被固定。对 $F$ 关于 $y$ 求偏导并要求 $F_y=N$:(M1)

$$ F_y=\frac{\partial}{\partial y}\int M\,dx+g'(y)=N \implies g'(y)=N-\frac{\partial}{\partial y}\int M\,dx. $$

To integrate $g'(y)$ in $y$ we must confirm the right side does not depend on $x$. Differentiate with respect to $x$: (A1)为了对 $g'(y)$ 关于 $y$ 积分,须确认右侧不依赖于 $x$。对 $x$ 求偏导:(A1)

$$ \frac{\partial}{\partial x}g'(y)=\frac{\partial N}{\partial x}-\frac{\partial}{\partial x}\frac{\partial}{\partial y}\int M\,dx=N_x-\frac{\partial^2}{\partial x\,\partial y}\int M\,dx=N_x-M_y=0, $$

where the last step uses the hypothesis $M_y=N_x$ and the fact that differentiation and integration interchange (by continuity). (R1) Since $\partial_x(g'(y))=0$, the expression for $g'$ truly depends only on $y$, so $g(y)=\displaystyle\int g'(y)\,dy$ is well-defined by a single integration in $y$, and $F$ is constructed. (A1)最后一步利用了假设 $M_y=N_x$ 以及微分与积分可交换(由连续性保证)。(R1) 由于 $\partial_x(g'(y))=0$,$g'$ 的表达式确实仅依赖于 $y$,故 $g(y)=\displaystyle\int g'(y)\,dy$ 由单次关于 $y$ 的积分确定,$F$ 的构造完成。(A1)

Insight.解题要点。 The proof is the same idea as recovering a scalar potential from a conservative vector field: $M_y=N_x$ is the curl-free condition, and on a simply connected region curl-free implies conservative. The argument in (b) is the constructive half: it does not just assert $F$ exists, it builds it and shows the procedure terminates. The independence check $\partial_x(g'(y))=0$ is where $M_y=N_x$ actually gets used; without it, $g$ could depend on $x$ and the method would be undefined.该证明与从保守向量场恢复标量势函数的思路相同:$M_y=N_x$ 是无旋条件,而在单连通区域上无旋意味着保守。(b) 中的论证是构造性的一半:它不仅断言 $F$ 存在,而且将其构造出来并说明该过程终止。独立性检验 $\partial_x(g'(y))=0$ 正是 $M_y=N_x$ 发挥作用的关键所在;若没有这一步,$g$ 可能依赖于 $x$,从而该方法将无法定义。
Q5HARDPROOFBernoulli equation: derive the substitution and solveBernoulli 方程:推导换元并求解[10 marks]

$y'+\frac{1}{x}y=xy^3$, $y(1)=1$, $x>0$: (a) derive the linear $v$-equation from $v=y^{-2}$; (b) solve by integrating factor; (c) revert to $y$ and find the particular solution with its domain.$y'+\frac{1}{x}y=xy^3$,$y(1)=1$,$x>0$:(a) 由 $v=y^{-2}$ 推导关于 $v$ 的线性方程;(b) 用积分因子法求解;(c) 回代得 $y$ 并求特解及其定义域。

Answers:答案:  (a) $v'-\tfrac{2}{x}v=-2x$  ·  (b) $v=x^2(C-2\ln x)$  ·  (c) $y=\dfrac{1}{x\sqrt{1-2\ln x}}$, valid for $x\in(0,\sqrt{e}\,)$有效定义域 $x\in(0,\sqrt{e}\,)$

(a) Derive the linearised equation step by step(a) 逐步推导线性化方程 M1·M1·A1·A1

Here $n=3$. Divide the equation $y'+\tfrac{1}{x}y=xy^3$ through by $y^3$ (assuming $y\ne0$): (M1)此处 $n=3$。方程 $y'+\tfrac{1}{x}y=xy^3$ 两边除以 $y^3$(设 $y\ne0$):(M1)

$$ y^{-3}y'+\frac{1}{x}y^{-2}=x. $$

Set $v=y^{1-n}=y^{-2}$. By the chain rule, $v'=-2y^{-3}y'$, so $y^{-3}y'=v'/(-2)$. (M1) Substitute: $v'/(-2)+(1/x)v=x$, which gives (A1)令 $v=y^{1-n}=y^{-2}$。由链式法则,$v'=-2y^{-3}y'$,故 $y^{-3}y'=v'/(-2)$。(M1) 代入:$v'/(-2)+(1/x)v=x$,得 (A1)

$$ v'-\frac{2}{x}v=-2x. $$

This is linear in $v$; the factor $(1-n)=1-3=-2$ appears both as the coefficient of the $p$-term and on the right side, as the Bernoulli formula predicts. (A1)该方程关于 $v$ 是线性的;因子 $(1-n)=1-3=-2$ 同时出现在 $p$ 项的系数和右侧,与 Bernoulli 公式的预测一致。(A1)

(b) Solve by integrating factor(b) 用积分因子法求解 M1·A1·M1·A1

The equation $v'-(2/x)v=-2x$ has $p=-2/x$. The integrating factor is (M1)方程 $v'-(2/x)v=-2x$ 中 $p=-2/x$。积分因子为 (M1)

$$ \mu=e^{\int(-2/x)\,dx}=e^{-2\ln x}=x^{-2}. $$

Multiply through by $x^{-2}$: $\displaystyle\frac{d}{dx}(x^{-2}v)=x^{-2}\cdot(-2x)=-\frac{2}{x}$. (A1) Integrate both sides: (M1)方程两边乘以 $x^{-2}$:$\displaystyle\frac{d}{dx}(x^{-2}v)=x^{-2}\cdot(-2x)=-\frac{2}{x}$。(A1) 对两边积分:(M1)

$$ x^{-2}v=\int-\frac{2}{x}\,dx=-2\ln x+C. $$

Therefore $v=x^2(C-2\ln x)$. (A1)故 $v=x^2(C-2\ln x)$。(A1)

(c) Revert to $y$ and apply the initial condition(c) 回代得 $y$ 并代入初始条件 M1·A1

Since $v=y^{-2}$, we have $y^{-2}=x^2(C-2\ln x)$, so $y^2=\dfrac{1}{x^2(C-2\ln x)}$. (M1) Apply $y(1)=1$: $1=\dfrac{1}{1^2(C-0)}$ gives $C=1$. Therefore由 $v=y^{-2}$,得 $y^{-2}=x^2(C-2\ln x)$,故 $y^2=\dfrac{1}{x^2(C-2\ln x)}$。(M1) 代入 $y(1)=1$:$1=\dfrac{1}{1^2(C-0)}$,得 $C=1$。因此

$$ y(x)=\frac{1}{x\sqrt{1-2\ln x}}. $$

The solution is valid where $1-2\ln x>0$, i.e. $\ln x<\tfrac12$, i.e. $x0$, the domain is $x\in(0,\sqrt{e}\,)$. (A1)解的有效区域为 $1-2\ln x>0$,即 $\ln x<\tfrac12$,即 $x0$,定义域为 $x\in(0,\sqrt{e}\,)$。(A1)

Insight.解题要点。 The Bernoulli substitution $v=y^{1-n}$ is the only substitution that linearises $y^n$ on the right; do not confuse it with $v=y^n$ or $v=1/y$. The factor $(1-n)=-2$ on both sides of the linearised equation is a sign-reversal trap: the equation in (a) has a negative coefficient on $v$ and a negative right side, which are both correct. The finite blow-up at $x=\sqrt{e}$ (where the denominator vanishes) is not a flaw in the method; it reflects the true behavior of the solution, and Picard's theorem guarantees uniqueness only within $(0,\sqrt{e}\,)$.Bernoulli 换元 $v=y^{1-n}$ 是唯一能线性化右侧 $y^n$ 的换元;不要与 $v=y^n$ 或 $v=1/y$ 混淆。线性化方程两侧的因子 $(1-n)=-2$ 是一个符号陷阱:(a) 中的方程在 $v$ 的系数和右侧均为负,这是正确的。$x=\sqrt{e}$ 处的有限爆破(分母消失处)并非方法的缺陷,它反映了解的真实行为,Picard 定理仅在 $(0,\sqrt{e}\,)$ 内保证唯一性。
Q6HARDPROOFautonomous equations and phase-line stability自治方程与相线稳定性[8 marks]

$y'=y(y-2)(4-y)$: (a) find and classify equilibria via $f'(y^*)$; (b) draw a phase line; (c) give $\lim_{t\to\infty}y(t)$ for $y(0)=3$.$y'=y(y-2)(4-y)$:(a) 用 $f'(y^*)$ 求平衡点并分类;(b) 画相线;(c) 给出 $y(0)=3$ 时 $\lim_{t\to\infty}y(t)$ 的值。

Answers:答案:  (a) $y^*=0$ stable稳定, $y^*=2$ unstable不稳定, $y^*=4$ stable稳定  ·  (b) arrows: up on $(-\infty,0)$, down on $(0,2)$, up on $(2,4)$, down on $(4,\infty)$箭头方向:$(-\infty,0)$ 上,$(0,2)$ 下,$(2,4)$ 上,$(4,\infty)$ 下  ·  (c) $\lim_{t\to\infty}y(t)=4$

(a) Find equilibria and apply the linearised stability test(a) 求平衡点并应用线性化稳定性判据 M1·A1·M1·A1

Set $f(y)=y(y-2)(4-y)=0$. The equilibria are $y^*=0$, $y^*=2$, and $y^*=4$. (M1)令 $f(y)=y(y-2)(4-y)=0$。平衡点为 $y^*=0$、$y^*=2$ 和 $y^*=4$。(M1)

Expand to differentiate: $f(y)=y(y-2)(4-y)=y(4y-y^2-8+2y)=6y^2-y^3-8y$, so $f'(y)=12y-3y^2-8$. (M1) Evaluate at each equilibrium: (A1)展开后求导:$f(y)=y(y-2)(4-y)=y(4y-y^2-8+2y)=6y^2-y^3-8y$,故 $f'(y)=12y-3y^2-8$。(M1) 在各平衡点处求值:(A1)

$$ f'(0)=-8<0\;(\text{stable}), \quad f'(2)=24-12-8=4>0\;(\text{unstable}), \quad f'(4)=48-48-8=-8<0\;(\text{stable}). $$

So $y^*=0$ and $y^*=4$ are sinks; $y^*=2$ is a source. (A1)故 $y^*=0$ 和 $y^*=4$ 是稳定平衡点(汇);$y^*=2$ 是不稳定平衡点(源)。(A1)

(b) Phase line(b) 相线 M1·A1

Check the sign of $f$ on each interval using a test point: (M1)用测试点检验 $f$ 在各区间上的符号:(M1)

  • $y<0$: test $y=-1$, $f(-1)=(-1)(-3)(5)=15>0$. Arrow pointing up.$y<0$:取 $y=-1$,$f(-1)=(-1)(-3)(5)=15>0$,箭头向上。
  • $0$0
  • $20$. Arrow pointing up.$20$,箭头向上。
  • $y>4$: test $y=5$, $f(5)=(5)(3)(-1)=-15<0$. Arrow pointing down.$y>4$:取 $y=5$,$f(5)=(5)(3)(-1)=-15<0$,箭头向下。

Phase line: arrows converge toward $0$ from $(-\infty,0)\cup(0,2)$ (note: from below $0$ arrows go up toward $0$, from above $0$ until $2$ arrows go down toward $0$), and arrows converge toward $4$ from $(2,4)\cup(4,\infty)$. The unstable equilibrium $y^*=2$ separates the two basins of attraction. (A1)相线:来自 $(-\infty,0)\cup(0,2)$ 的箭头收敛于 $0$(注意:从 $0$ 以下箭头向上指向 $0$,从 $0$ 到 $2$ 之间箭头向下指向 $0$),来自 $(2,4)\cup(4,\infty)$ 的箭头收敛于 $4$。不稳定平衡点 $y^*=2$ 将两个吸引盆分隔开。(A1)

(c) Long-term behavior from $y(0)=3$(c) 从 $y(0)=3$ 出发的长期行为 M1·A1

Since $y(0)=3\in(2,4)$ and $f>0$ on this interval, $y$ is increasing and cannot cross the equilibrium $y^*=4$ (by uniqueness). (M1) All solutions starting in $(2,4)$ are attracted to the stable equilibrium $y^*=4$, so由于 $y(0)=3\in(2,4)$ 且 $f$ 在该区间上 $f>0$,$y$ 单调递增且不能越过平衡点 $y^*=4$(由唯一性保证)。(M1) 所有从 $(2,4)$ 出发的解均被稳定平衡点 $y^*=4$ 吸引,故

$$ \lim_{t\to\infty}y(t)=4. $$

(A1)(A1)

Insight.解题要点。 The phase line encodes global qualitative behavior without solving the ODE. The sign of $f$ between equilibria is the only information needed; computing $f'(y^*)$ then makes the local stability precise. The unstable equilibrium $y^*=2$ acts as a threshold: any $y(0)>2$ is eventually drawn toward $4$, while any $y(0)\in(0,2)$ is drawn toward $0$. This is precisely the basin-of-attraction structure the logistic model lacks (it has only one stable equilibrium), making this cubic richer dynamically.相线在无需求解微分方程的情况下,编码了全局定性行为。平衡点之间 $f$ 的符号是所需的唯一信息;计算 $f'(y^*)$ 则使局部稳定性得到精确刻画。不稳定平衡点 $y^*=2$ 起到阈值的作用:任何 $y(0)>2$ 最终都趋向 $4$,而任何 $y(0)\in(0,2)$ 则趋向 $0$。这正是逻辑斯谛模型所缺乏的吸引盆结构(逻辑斯谛模型仅有一个稳定平衡点),使得本题的三次方程在动力学上更为丰富。
PART III  ·  APPLICATIONS AND SYNTHESISExtended problems · 28 marks综合应用题 · 28分

Worked Solutions详细解答

Q7HARDAPPLIEDlogistic IVP: closed-form solution and long-term behavior逻辑斯谛初值问题:闭合形式解与长期行为[8 marks]

$P'=0.4P(1-P/500)$, $P(0)=50$: (a) state $K$, $k$, equilibria and classify them; (b) separate variables and derive $P/(1-P/K)=Ae^{kt}$; (c) solve for $P(t)$; (d) find $\lim_{t\to\infty}P(t)$ and confirm it equals the stable equilibrium.$P'=0.4P(1-P/500)$,$P(0)=50$:(a) 写出 $K$、$k$、平衡点并分类;(b) 分离变量并推导 $P/(1-P/K)=Ae^{kt}$;(c) 求 $P(t)$;(d) 求 $\lim_{t\to\infty}P(t)$ 并确认其等于稳定平衡点。

Answers:答案:  (a) $K=500$, $k=0.4$, equilibria $P=0$ (unstable) and $P=500$ (stable)平衡点 $P=0$(不稳定)和 $P=500$(稳定)  ·  (b) derived below见下方推导  ·  (c) $P(t)=\dfrac{500}{1+9e^{-0.4t}}$  ·  (d) $\lim_{t\to\infty}P=500$

(a) Parameters and equilibria(a) 参数与平衡点 A1

Reading off the equation: carrying capacity $K=500$ and intrinsic growth rate $k=0.4$. Setting $f(P)=0.4P(1-P/500)=0$ gives equilibria $P=0$ (unstable, since $f'(0)=0.4>0$) and $P=500$ (stable, since $f'(500)=-0.4<0$). (A1)从方程中读出:环境容量 $K=500$,固有增长率 $k=0.4$。令 $f(P)=0.4P(1-P/500)=0$,得平衡点 $P=0$(不稳定,因为 $f'(0)=0.4>0$)和 $P=500$(稳定,因为 $f'(500)=-0.4<0$)。(A1)

(b) Separate variables and partial fractions(b) 分离变量与部分分式 M1·M1·A1·A1

Rewrite: $\dfrac{dP}{P(1-P/500)}=0.4\,dt$. (M1) Decompose the left integrand by partial fractions, writing $\dfrac{1}{P(1-P/500)}=\dfrac{1}{P}+\dfrac{1/500}{1-P/500}$ (verified by common denominator): (M1)改写为:$\dfrac{dP}{P(1-P/500)}=0.4\,dt$。(M1) 用部分分式分解左侧被积函数,写成 $\dfrac{1}{P(1-P/500)}=\dfrac{1}{P}+\dfrac{1/500}{1-P/500}$(通分可验证):(M1)

$$ \int\left(\frac{1}{P}+\frac{1/500}{1-P/500}\right)dP=\int0.4\,dt. $$ $$ \ln|P|-\ln\!\left|1-\frac{P}{500}\right|=0.4t+C_1. $$

(A1) Exponentiate: $\dfrac{P}{1-P/500}=e^{C_1}e^{0.4t}=A\,e^{0.4t}$. (A1)(A1) 取指数:$\dfrac{P}{1-P/500}=e^{C_1}e^{0.4t}=A\,e^{0.4t}$。(A1)

(c) Solve for $P(t)$ and apply $P(0)=50$(c) 求 $P(t)$ 并代入 $P(0)=50$ M1·A1

At $t=0$: $A=\dfrac{P_0}{1-P_0/K}=\dfrac{50}{1-50/500}=\dfrac{50}{0.9}=\dfrac{500}{9}$. (M1) Solve $\dfrac{P}{1-P/500}=\dfrac{500}{9}e^{0.4t}$ for $P$: cross-multiply and rearrange,当 $t=0$ 时:$A=\dfrac{P_0}{1-P_0/K}=\dfrac{50}{1-50/500}=\dfrac{50}{0.9}=\dfrac{500}{9}$。(M1) 从 $\dfrac{P}{1-P/500}=\dfrac{500}{9}e^{0.4t}$ 中解出 $P$:交叉相乘并整理,

$$ P=\frac{K P_0}{P_0+(K-P_0)e^{-kt}}=\frac{500\cdot50}{50+450\,e^{-0.4t}}=\frac{500}{1+9e^{-0.4t}}. $$

(A1) Verification at $t=0$: $P=500/(1+9)=50$. Correct.(A1) 验证 $t=0$ 时:$P=500/(1+9)=50$,正确。

(d) Long-term limit(d) 长期极限 A1

As $t\to\infty$, $e^{-0.4t}\to0$, so $P(t)\to500=K$. (A1) This confirms $P=500$ is the stable equilibrium: any positive initial population is attracted to the carrying capacity, consistent with the linearised stability finding $f'(500)=-0.4<0$.当 $t\to\infty$ 时,$e^{-0.4t}\to0$,故 $P(t)\to500=K$。(A1) 这证实 $P=500$ 是稳定平衡点:任何正的初始种群都被环境容量吸引,与线性化稳定性分析结果 $f'(500)=-0.4<0$ 一致。

Insight.解题要点。 Test $M_y=N_x$ before assuming exactness; in a logistic problem the equation is autonomous, so the relevant test is whether $P=K$ is a stable equilibrium, read off from $f'(K)<0$. The partial-fraction step is mechanical once you recognise $1/(P(1-P/K))$; the key is to write the decomposition with the $1/K$ scaling in the second term so integration gives clean logarithms. The logistic formula $P(t)=K/(1+Ae^{-kt})$ is worth memorising in this form: $A=(K-P_0)/P_0$ measures how far the initial population is from carrying capacity.在假定恰当性之前先验证 $M_y=N_x$;在逻辑斯谛问题中,方程是自治的,相关验证是判断 $P=K$ 是否为稳定平衡点,由 $f'(K)<0$ 读出。一旦识别出 $1/(P(1-P/K))$ 的形式,部分分式分解步骤是机械性的;关键在于第二项中写入 $1/K$ 的缩放因子,从而积分后得到整洁的对数。逻辑斯谛公式 $P(t)=K/(1+Ae^{-kt})$ 值得以这种形式记忆:$A=(K-P_0)/P_0$ 衡量初始种群偏离环境容量的程度。
Q8HARDAPPLIEDmixing tank: linear ODE with constant-volume balance混合槽:恒容守恒下的线性常微分方程[6 marks]

Tank: 200 L pure water, inflow 0.05 kg/L at 4 L/min, outflow 4 L/min. $A(t)$ = salt mass. (a) set up $A'+pA=q$; (b) solve with $A(0)=0$; (c) find $A(30)$ and $\lim_{t\to\infty}A(t)$.水槽:200 L 纯水,0.05 kg/L 盐水以 4 L/min 流入,4 L/min 流出。$A(t)$ 为盐的质量。(a) 建立 $A'+pA=q$;(b) 以 $A(0)=0$ 求解;(c) 求 $A(30)$ 和 $\lim_{t\to\infty}A(t)$。

Answers:答案:  (a) $A'+\tfrac{1}{50}A=0.2$  ·  (b) $A(t)=10(1-e^{-t/50})$  ·  (c) $A(30)\approx4.51$ kg; $\lim_{t\to\infty}A=10$ kg

(a) Set up the mass-balance ODE(a) 建立质量守恒微分方程 M1·A1

Apply the balance law: $A'=(\text{rate in})-(\text{rate out})$. Rate in: $0.05\times4=0.2\,\text{kg/min}$. Rate out: $(A/200)\times4=A/50\,\text{kg/min}$. (M1) Therefore应用守恒律:$A'=(\text{流入速率})-(\text{流出速率})$。流入速率:$0.05\times4=0.2\,\text{kg/min}$。流出速率:$(A/200)\times4=A/50\,\text{kg/min}$。(M1) 因此

$$ A'+\frac{1}{50}A=0.2, \quad p=\frac{1}{50},\quad q=0.2. \quad\text{(A1)}$$

(b) Solve by integrating factor(b) 用积分因子法求解 M1·A1

$\mu=e^{t/50}$. Multiply: $(e^{t/50}A)'=0.2e^{t/50}$. (M1) Integrate: $e^{t/50}A=10e^{t/50}+C$, so $A=10+Ce^{-t/50}$. Apply $A(0)=0$: $C=-10$. (A1)$\mu=e^{t/50}$。两边乘以 $\mu$:$(e^{t/50}A)'=0.2e^{t/50}$。(M1) 积分:$e^{t/50}A=10e^{t/50}+C$,故 $A=10+Ce^{-t/50}$。代入 $A(0)=0$:$C=-10$。(A1)

$$ A(t)=10\!\left(1-e^{-t/50}\right). $$

Verification: $A'=0.2e^{-t/50}$ and $0.2-A/50=0.2e^{-t/50}$. Matches.验证:$A'=0.2e^{-t/50}$,$0.2-A/50=0.2e^{-t/50}$,两者吻合。

(c) Evaluate and interpret the limit(c) 计算并解释极限 A1·A1

$A(30)=10(1-e^{-30/50})=10(1-e^{-0.6})\approx10(1-0.5488)=10\times0.4512\approx4.51\,\text{kg}$. (A1)$A(30)=10(1-e^{-30/50})=10(1-e^{-0.6})\approx10(1-0.5488)=10\times0.4512\approx4.51\,\text{kg}$。(A1)

$\displaystyle\lim_{t\to\infty}A(t)=10\,\text{kg}$. (A1) Physical interpretation: at steady state the salt concentration in the tank equals $10/200=0.05\,\text{kg/L}$, which is exactly the inflow concentration. The tank equilibrates to the incoming brine strength, as the balance law requires (rate in equals rate out at steady state).$\displaystyle\lim_{t\to\infty}A(t)=10\,\text{kg}$。(A1) 物理解释:稳态时槽内盐的浓度为 $10/200=0.05\,\text{kg/L}$,恰好等于流入浓度。槽内浓度趋近于流入盐水的浓度,这与守恒律的要求一致(稳态时流入速率等于流出速率)。

Insight.解题要点。 The mixing ODE is always $A'+(\text{outflow rate}/V)A=c_{\text{in}}\cdot r_{\text{in}}$; the constant-volume assumption makes both $p$ and $q$ constant, so the integrating factor is a simple exponential. The steady-state value $A_\infty=q/p=0.2/(1/50)=10$ can be read from the ODE directly (set $A'=0$) without solving it, and gives a useful sanity check on the formula.混合槽的微分方程始终为 $A'+(\text{流出速率}/V)A=c_{\text{流入}}\cdot r_{\text{流入}}$;恒容假设使 $p$ 和 $q$ 均为常数,故积分因子为简单指数函数。稳态值 $A_\infty=q/p=0.2/(1/50)=10$ 可直接从方程读出(令 $A'=0$),无需求解,可作为对公式的有用验证。
Q9HARDAPPLIEDNewton's law of cooling: parameter estimation and inversion牛顿冷却定律:参数估计与反解[8 marks]

Coffee cools from 90°C, room at 20°C; after 10 min it is 70°C. (a) solve for $T(t)$; (b) find $k$; (c) find time to reach 50°C; (d) state $\lim_{t\to\infty}T(t)$ and link to autonomous stability.咖啡从 90°C 冷却,室温 20°C;10 分钟后降至 70°C。(a) 求 $T(t)$;(b) 求 $k$;(c) 求降至 50°C 所需时间;(d) 写出 $\lim_{t\to\infty}T(t)$ 并联系自治方程稳定性。

Answers:答案:  (a) $T=20+70e^{-kt}$  ·  (b) $k=\tfrac{1}{10}\ln\tfrac{7}{5}$  ·  (c) $t^*=10\ln(7/3)/\ln(7/5)\approx25.2\,\text{min}$  ·  (d) $20=T_a$, stable equilibrium稳定平衡点

(a) Solve the separable linear ODE(a) 求解可分离线性常微分方程 M1·A1

Let $u=T-20$; then $u'=-ku$, giving $u=70e^{-kt}$. (M1)令 $u=T-20$,则 $u'=-ku$,得 $u=70e^{-kt}$。(M1)

$$ T(t)=20+70\,e^{-kt}. $$

Verification: $T(0)=20+70=90$. (A1)验证:$T(0)=20+70=90$。(A1)

(b) Determine $k$ from $T(10)=70$(b) 由 $T(10)=70$ 确定 $k$ M1·A1

$70=20+70e^{-10k}$, so $e^{-10k}=5/7$ and (M1)$70=20+70e^{-10k}$,故 $e^{-10k}=5/7$,(M1)

$$ k=\frac{1}{10}\ln\frac{7}{5}. $$

Numerical value: $k\approx0.03365\,\text{min}^{-1}$. (A1)数值:$k\approx0.03365\,\text{min}^{-1}$。(A1)

(c) Find the time to reach 50°C(c) 求降至 50°C 的时间 M1·A1

Set $T(t^*)=50$: $50=20+70e^{-kt^*}$, so $30=70e^{-kt^*}$ and $e^{-kt^*}=\dfrac{3}{7}$. (M1) Taking the natural log:令 $T(t^*)=50$:$50=20+70e^{-kt^*}$,故 $30=70e^{-kt^*}$,$e^{-kt^*}=\dfrac{3}{7}$。(M1) 取自然对数:

$$ kt^*=\ln\frac{7}{3} \implies t^*=\frac{\ln(7/3)}{k}=\frac{10\ln(7/3)}{\ln(7/5)}. $$

Numerically: $\ln(7/3)\approx0.8473$, $\ln(7/5)\approx0.3365$, so $t^*\approx8.473/0.3365\approx25.2\,\text{min}$. (A1)数值计算:$\ln(7/3)\approx0.8473$,$\ln(7/5)\approx0.3365$,故 $t^*\approx8.473/0.3365\approx25.2\,\text{min}$。(A1)

(d) Long-term limit and autonomous stability(d) 长期极限与自治方程稳定性 A1·A1

Since $k>0$, $e^{-kt}\to0$ as $t\to\infty$, so $\displaystyle\lim_{t\to\infty}T(t)=20=T_a$. (A1)由于 $k>0$,当 $t\to\infty$ 时 $e^{-kt}\to0$,故 $\displaystyle\lim_{t\to\infty}T(t)=20=T_a$。(A1)

This matches the autonomous stability analysis: rewrite $T'=-k(T-20)=f(T)$. The equilibrium is $T^*=20$, and $f'(T^*)=-k<0$, so $T^*$ is a stable equilibrium (a sink). Any initial temperature is attracted to room temperature, which is exactly what the formula $T=20+70e^{-kt}$ confirms. (A1)这与自治方程的稳定性分析一致:改写为 $T'=-k(T-20)=f(T)$。平衡点为 $T^*=20$,$f'(T^*)=-k<0$,故 $T^*$ 是稳定平衡点(汇)。任何初始温度都被室温吸引,这正是公式 $T=20+70e^{-kt}$ 所证实的。(A1)

Insight.解题要点。 Newton's law of cooling is a linear equation in disguise: the substitution $u=T-T_a$ converts it to exponential decay $u'=-ku$. Parameter estimation proceeds by plugging in the second data point ($T(10)=70$) after the general solution is known. The inversion in (c) shows a key applied skill: given a target temperature, solve for $t$ by dividing through, then taking a logarithm. Always check the limiting temperature against the ambient value $T_a$ as a built-in sanity check.牛顿冷却定律本质上是一个线性方程:换元 $u=T-T_a$ 将其转化为指数衰减 $u'=-ku$。参数估计的步骤是:在已知通解后,代入第二个数据点($T(10)=70$)。(c) 中的反解展示了一项重要的应用技能:给定目标温度,通过两边除法再取对数解出 $t$。务必将极限温度与环境温度 $T_a$ 对照,作为内置的合理性检验。
Q10HARDAPPLIEDhomogeneous substitution and exact method compared齐次换元法与恰当方程法的比较[6 marks]

$2xy\,dx+(y^2-x^2)\,dy=0$: (a) find $\mu(y)$ via $(N_x-M_y)/M$; (b) multiply, verify exactness, solve; (c) confirm the same family via the homogeneous substitution $w=x/y$.$2xy\,dx+(y^2-x^2)\,dy=0$:(a) 通过 $(N_x-M_y)/M$ 求 $\mu(y)$;(b) 乘以积分因子,验证恰当性,求解;(c) 通过齐次换元 $w=x/y$ 确认相同的曲线族。

Answers:答案:  (a) $\mu=y^{-2}$  ·  (b) $x^2+y^2=Cy$  ·  (c) same family $x^2+y^2=Cy$ confirmed同一曲线族 $x^2+y^2=Cy$ 已验证

(a) Identify the $y$-only integrating factor(a) 确定仅关于 $y$ 的积分因子 M1·A1

$M=2xy$, $M_y=2x$; $N=y^2-x^2$, $N_x=-2x$. Since $M_y\ne N_x$, not exact. Test the $y$-only formula: (M1)$M=2xy$,$M_y=2x$;$N=y^2-x^2$,$N_x=-2x$。由于 $M_y\ne N_x$,非恰当方程。检验仅关于 $y$ 的公式:(M1)

$$ \frac{N_x-M_y}{M}=\frac{-4x}{2xy}=\frac{-2}{y}. $$

Depends only on $y$, so $\mu=e^{\int(-2/y)\,dy}=y^{-2}$. (A1)仅依赖于 $y$,故 $\mu=e^{\int(-2/y)\,dy}=y^{-2}$。(A1)

(b) Multiply, verify exactness, and solve(b) 乘以积分因子、验证恰当性并求解 M1·A1

Multiply by $y^{-2}$: $\tilde{M}=2x/y$, $\tilde{N}=1-x^2/y^2$. Check: $\tilde{M}_y=-2x/y^2=\tilde{N}_x$. Exact. (M1) Integrate $\tilde{M}$ in $x$: $F=x^2/y+g(y)$. Match $F_y=1-x^2/y^2$ gives $g'(y)=1$, $g=y$. (A1)两边乘以 $y^{-2}$:$\tilde{M}=2x/y$,$\tilde{N}=1-x^2/y^2$。验证:$\tilde{M}_y=-2x/y^2=\tilde{N}_x$,恰当方程。(M1) 对 $\tilde{M}$ 关于 $x$ 积分:$F=x^2/y+g(y)$。匹配 $F_y=1-x^2/y^2$ 得 $g'(y)=1$,$g=y$。(A1)

$$ x^2+y^2=Cy. $$

(c) Confirm via the homogeneous substitution $w=x/y$(c) 通过齐次换元 $w=x/y$ 验证 M1·A1

Rewrite the equation treating $x$ as a function of $y$: $dx/dy=-(y^2-x^2)/(2xy)=(x^2-y^2)/(2xy)$. This is a homogeneous equation with the ratio $w=x/y$. Set $x=wy$, so $dx/dy=w+y\,dw/dy$: (M1)将方程改写为以 $x$ 作为 $y$ 的函数:$dx/dy=-(y^2-x^2)/(2xy)=(x^2-y^2)/(2xy)$。这是含比值 $w=x/y$ 的齐次方程。令 $x=wy$,则 $dx/dy=w+y\,dw/dy$:(M1)

$$ w+y\frac{dw}{dy}=\frac{w^2y^2-y^2}{2wy^2}=\frac{w^2-1}{2w} \implies y\frac{dw}{dy}=\frac{w^2-1}{2w}-w=\frac{-w^2-1}{2w}. $$

Separate: $\dfrac{2w}{w^2+1}\,dw=-\dfrac{dy}{y}$. Integrate: $\ln(w^2+1)=-\ln y+C_1$. Substitute $w=x/y$:分离变量:$\dfrac{2w}{w^2+1}\,dw=-\dfrac{dy}{y}$。积分:$\ln(w^2+1)=-\ln y+C_1$。代入 $w=x/y$:

$$ \ln\!\left(\frac{x^2}{y^2}+1\right)+\ln y=C_1 \implies \ln\!\left(\frac{x^2+y^2}{y}\right)=C_1 \implies \frac{x^2+y^2}{y}=C. $$

This is $x^2+y^2=Cy$, identical to the exact-method result. (A1)即 $x^2+y^2=Cy$,与恰当方程法的结果完全一致。(A1)

Insight.解题要点。 When $(N_x-M_y)/M$ is a function of $y$ alone, the integrating factor formula gives $\mu(y)=e^{\int[(N_x-M_y)/M]\,dy}$; note the sign convention: the $y$-formula uses $N_x-M_y$ over $M$, not $M_y-N_x$ over $N$. The agreement between the exact method and the homogeneous substitution is a consistency check, not a coincidence: both routes find the same one-parameter family of curves because they are solving the same ODE. The homogeneous route via $x=wy$ is slightly non-standard (usually $v=y/x$), but is cleaner here because the equation is more naturally viewed as $dx/dy = f(x/y)$.当 $(N_x-M_y)/M$ 仅为 $y$ 的函数时,积分因子公式给出 $\mu(y)=e^{\int[(N_x-M_y)/M]\,dy}$;注意符号约定:关于 $y$ 的公式使用 $N_x-M_y$ 除以 $M$,而非 $M_y-N_x$ 除以 $N$。恰当方程法与齐次换元法的结果吻合是一致性检验,而非巧合:两条路径求出相同的单参数曲线族,因为它们求解的是同一个常微分方程。通过 $x=wy$ 进行的齐次换元略显非标准(通常用 $v=y/x$),但在此处更简洁,因为方程更自然地被视为 $dx/dy = f(x/y)$。