高数笔记 P04:微分方程

  • 通解中独立常数的个数等于方程的阶数。
  • 求解过程中不确定正负的因子要加绝对值。
  • 可能出现丢解的情况,这种解称为奇解,全部解包含通解和奇解,只有在线性的微分方程中,通解才等同于全部解。

1 变量可分离的微分方程

形如\(\cfrac {dy}{dx} = h(x)g(y)\)

\(\implies \cfrac {dy}{g(y)} = h(x)dx\)

\(\implies \displaystyle \int \cfrac {dy}{g(y)} = \int h(x)dx + C\)

2 齐次微分方程

形如\(\cfrac {dy}{dx} = f(\cfrac yx)\)

\(\implies\)\(u = \cfrac yx \implies y = ux \implies \cfrac{dy}{dx} = \cfrac {du}{dx}x + u\)

\(\implies \cfrac {du}{dx}x + u = f(u) \implies \cfrac {du}{f(u) - u} = \cfrac {dx}x\)

\(\implies \displaystyle \int \cfrac {du}{f(u) - u} = \int \cfrac {dx}x + C\)

可化作齐次的微分方程

形如\(\cfrac {dy}{dx} = \cfrac {ax + by + c}{a_1 x + b_1 y + c_1}\),当\(c = c_1 = 0\)时,方程时齐次的,否则就不是齐次的。

\(x = X +h, y = Y + k \implies dx = dX, dy = dY\)

\(\implies \cfrac {dY}{dX} = \cfrac {aX + bY + ah + bk + c}{a_1 X + b_1 Y + a_1 h + b_1 k + c_1}\)

若方程组\(\begin{cases} ah + bk + c = 0 \\ a_1h + b_1 k + c_1 = 0 \\ \end{cases}\)的系数行列式\(\left| \begin{matrix} a & b \\ c & d\end{matrix} \right| \neq 0\),即\(\cfrac {a_1}a \neq \cfrac {b_1}b\),可以找出满足这个方程组的\(h, k\),将上式化简成\(\cfrac {dY}{dX} = \cfrac {aX + bY}{a_1X + b_1Y}\),以齐次微分方程的解法求解;

\(\cfrac {a_1}a = \cfrac {b_1}b = \lambda\),原式可以化简为\(\cfrac {dy}{dx} = \cfrac {ax + by + c}{\lambda (ax + by) + c_1}\),取\(v = ax + by\)

\(\implies \cfrac {dv}{dx} = a + b \cfrac {dy}{dx} \implies \cfrac {dy}{dx} = \cfrac 1b (\cfrac {dv}{dx} - a)\),即

\(\implies \cfrac 1b (\cfrac {dv}{dx} - a) = \cfrac {v + c}{\lambda v + c_1}\),为可分离变量的微分方程。

3 一阶线性微分方程

形如\(y'+ P(x)y = Q(x)\)

\(\implies\)同乘\(e^{\int (Px)dx} \implies e^{\int P(x)dx} \cdot y' + P(x) \cdot e^{\int P(x)dx} \cdot y = Q(x) \cdot e^{\int P(x)dx}\)

\(\implies\)\((uv' + u'v = (uv)') \implies (e^{\int P(x)dx} \cdot y)' = Q(x) \cdot e^{\int P(x)dx}\)

\(\implies \displaystyle e^{\int P(x)dx} \cdot y = \int Q(x) e^{\int P(x)dx} dx + C\)

\(\implies \displaystyle y = e^{-\int P(x)dx} \left[ \int Q(x) e^{\int P(x)dx} dx + C \right]\)

伯努利方程

形如\(\cfrac {dy}{dx} + P(x)y = Q(x)y^n, (n \neq 0,1)\)

\(\implies\)同乘\(y^{-n} \implies y^{-n} \cfrac {dy}{dx} + P(x) y^{1-n} = Q(x)\),令\(z = y^{1 - n}\)

\(\implies \cfrac {dz}{dx} = (1-n) y^{-n} \cfrac {dy}{dx} \implies \cfrac 1{1-n} \cfrac {dz}{dx} + P(x)z = Q(x)\),同乘\((1-n)\)

\(\implies \cfrac {dz}{dx} = [(1-n)P(x)] \cdot z = [(1-n)Q(x)]\),利用一阶线性微分方程的解法解左式,然后求得原方程的解。

全微分方程

若存在二元函数\(u(x, y)\)使得\(du = P(x, y)dx + Q(x, y)dy\),则称微分方程\(P(x, y)dx = Q(x, y)dy = 0\)为全微分方程,它的通解为\(u(x, y) = C\)

4 可降阶的高阶微分方程

  • 形如\(y^{(n)} = f(x)\),逐次积分即可

    \(\implies \displaystyle y^{(n - 1)} = \int f(x)dx + C_1\)

    \(\implies \displaystyle y^{(n - 2)} = \int (\int f(x)dx + C_1)dx + C_2\)

    \(\implies \cdots \cdots\)

  • 形如\(y'' = f(x, y')\),令\(p = y' \implies y'' = \cfrac{dp}{dx} = p'\)

    \(\implies\)方程化为\(p' = f(x, p)\)的一阶微分方程,解得\(p = \varphi(x, C_1)\)带入\(p = y'\),再次获得一个一阶微分方程,求解即可。通解为\(\displaystyle y = \int \varphi(x, C_1) dx + C_2\)

  • 形如\(y'' = f(y, y')\),令\(p = y' \implies y'' = \cfrac {dp}{dx} = \cfrac {dp}{dy} \cdot \cfrac {dy}{dx} = p \cfrac {dp} {dy}\)

    \(\implies\)方程化为\(p \cfrac {dp}{dy} = f(y, p)\)的一阶微分方程,解得\(p = \varphi (y, C_1)\)带入\(p = y'\),再次获得一个一阶微分方程,求解即可。通解为\(\displaystyle \int \cfrac {dy}{\varphi (y, C_1)} = \int dx + C_2 = x + C_2\)

5 高阶线性微分方程解的结构

  • 对于二阶齐次线性方程\(y'' + P(x)y' + Q(x)y = 0\):
    • 如果\(y_1(x), y_2(x)\)是其两个解,则\(y = C_1 y_1(x) + C_2 y_2(x)\)也是它的解;
    • 如果\(y_1(x), y_2(x)\)是其两个线性无关的特解,则\(y = C_1 y_1(x) + C_2 y_2(x)\)是它的通解。
  • 对于\(n\)阶齐次线性方程\(y^{(n)} + a_1(x) y^{(n - 1)} + \cdots + a_{n-1}(x) y' + a_n(x) y = 0\),如果\(y_1(x), y_2(x), \cdots , y_n(x)\)是其\(n\)个线性无关的特解,则此方程的通解为\(y = C_1 y_1(x) + C_2 y_2(x) + \cdots + C_n y_n(x)\)
  • 对于二阶非齐次线性方程\(y'' + P(x) y' + Q(x) y = f(x)\),如果\(y^*(x)\)是其一个特解,\(Y(x)\)是其对应的齐次方程的通解,则\(y = Y(x) + y^*(x)\)是它的通解。
  • 线性微分方程的解的叠加原理:设\(y_1^*(x), y_2^*(x)\)分别是方程\(y'' +P(x) y' + Q(x)y = f_1(x), \ and \ \ y'' + P(x) y' + Q(x)y = f_2(x)\)的特解,则\(y = y_1^*(x) + y_2^*(x)\)是方程\(y'' + p(x) y' + Q(x)y = f_1(x) + f_2(x)\)的特解。

6 常系数齐次线性微分方程

  • 对于二阶常系数齐次线性微分方程\(y'' + py' + qy = 0\),其中\(p, q\)为常数,特征方程为\(r^2 + pr + q = 0\),解得特征根\(r\)。则其通解为

    \(\begin{array}{c|c} r_1 \neq r_2 & y = C_1 e^{r_1 x} + C_2 e^{r_2 x} \\ r_1 = r_2 & y = (C_1 + C_2 x) e^{r_1x} \\ r_{1,2} = \alpha \pm \beta i (\beta \neq 0) & y = e^{\alpha x} (C_1 \cos \beta x + C_2 \sin \beta x) \\ \end{array}\)

  • 对于\(n​\)阶常系数齐次线性微分方程\(y^{(n)} + p_1 y^{(n-1)} + p_2 y^{(n-2)} + \cdots + p_{n-1}y' + p_n y = 0​\),解其特征方程,通解为

    \(\begin{array}{c|l} \text{single real root } r & C e^{rx} \\ \text{pair of single complex root } r_{1, 2} = \alpha + \beta i & e^{\alpha x}(C_1 \cos \beta x + C_2 \sin \beta x) \\ k \text{ repeated real root } r & e^{rx}(C_1 + C_2 x + \cdots + C_k x^{k - 1}) \\ \text{pair of } k \text{ repeated complex root } r_{1, 2} = \alpha + \beta i & e^{\alpha x} [(C_1 + C_2 x + \cdots + C_k x^{k - 1}) \cos \beta x + (D_1 + D_2 x + \cdots + D_k x^{k - 1}) \sin \beta x] \end{array}\)

7 常系数非齐次线性微分方程

  • 对于二阶常系数齐次线性微分方程\(y'' + py' + qy = f(x)​\),其中\(p, q​\)为常数,可以将求解问题归结为求其对于齐次方程的通解和求该方程的一个特解特征方程为。两种常见的\(f(x)​\)形式

    • \(f(x) = e^{\lambda x} P_m(x)\),其中\(P_m\)\(m\)次多项式。

      \(y^*(x) = R(x) e^{\lambda x} \implies {y^*}' = e^{\lambda x} [\lambda R(x) + R'(x)], {y^*}'' = e^{\lambda x} [\lambda^2 R(x) + 2\lambda R'(x) + R''(x)]\)

      带入方程并消去\(e^{\lambda x}\)得,\(R''(x) + (2\lambda +p)R'(x) + (\lambda^2 + p\lambda + q)R(x) = P_m(x)\)

      1. \(\lambda\)不是特征方程的根,\(y^* = R_m(x) e^{\lambda x}\)
      2. \(\lambda\)是特征方程的单根,\(y^* = x R_m e^{\lambda x}\)
      3. \(\lambda\)是特征方程的重根,\(y^* = x^2 R_m(x) e^{\lambda x}\)

      其中\(x^k R_m(x)\)的系数通过与\(P_m(x)\)的同次幂系数恒等来建立方程组求得。

    • \(f(x) = e^{\lambda x} [P_l(x) \cos \omega x + Q_n(x) \sin \omega x]\),使用欧拉公式\(\begin{cases} \cos \theta = \frac 12 (e^{i\theta} + e^{-i\theta}) \\ \sin \theta = \frac 12 (e^{i\theta} - e^{-i\theta}) \end{cases}\),把\(f(x)\)变成复变指数函数的形式

      \(\begin{aligned} f(x) & = e^{\lambda x} [P_l \cfrac {e^{\omega x i} + e^{-\omega x i}}2 + Q_n \cfrac {e^{\omega x i} - e^{-\omega x i}}2] \\ & = (\cfrac {P_l}2 + \cfrac {Q_n}{2i}) e^{(\lambda + \omega i)x} + (\cfrac {P_l}2 - \cfrac {Q_m}{2i}) e^{(\lambda - \omega i)x} \\ & =P(x) e^{(\lambda + \omega i)x} + \overline P(x) e^{(\lambda - \omega i)x} \\ \end{aligned}\)

      其中\(P(x) = \cfrac {P_l}2 - \cfrac {Q_n}2 i, \ \overline P(x) = \cfrac {P_l}2 + \cfrac {Q_n}2 i, m = \max\{l, n\}\)

      使用叠加原理,将\(f(x)\)分解成两项,方程变为两个方程,按照前面的进行求解。

      \(y_1^* = x^k R_m(x) e^{(\lambda + \omega i) x}\)为方程\(y'' + py' + qy = P(x) e^{(\lambda + \omega i)x}\)的一个特解,则其共轭\(y_2^* = x^k \overline R_m(x) e^{(\lambda - \omega i) x}\)为方程\(y'' + py' + qy = \overline P(x) e^{(\lambda - \omega i)x}\)的一个特解,其中\(\overline R_m(x)\)\(R_m(x)\)的共轭\(m\)次多项式,\(m = \max \{ l, n \}\)

      叠加,\(y^* = x^k e^{\lambda x} [R_m e^{\omega x i} + \overline R_m e^{-\omega x i}] = x^k e^{\lambda x} [R_m (\cos \omega x + i \sin \omega x) + \overline R_m (\cos \omega x - i \sin \omega x)]\)

      由于括号内的两项是互成共轭的,相加之后无虚部,因此

      \(y^* = x^k e^{\lambda x} [R_m^{(1)}(x) \cos \omega x + R_m^{(2)}(x) \sin \omega x]\),其中\(R_m^{(1)}(x), R_m^{(2)}(x)\)\(m\)次多项式,系数按照前面的方法求得。

8 欧拉方程

形如\(x^n y^{(n)} + p_1 x^{n - 1} y^{(n - 1)} + \cdots + p_{n - 1}xy' + p_n y = f(x)\),作变换\(x = e^t\),取记号\(D\)\(\cfrac d{dt}\),则

\(xy' = Dy, \quad x^2 y'' = D(D-1)y,\quad \cdots, \quad x^k y^{(k)} = D(D-1)(D-2) \cdots (D-k+1)y \quad \cdots\)

将上述带入原方程,方程转化为一个以\(t\)为自变量的常系数线性微分方程,求解即可。

如二阶的方程\(x^2 y'' + pxy' + qy = f(x)\),可以化简为\(y''(t) + (p-1)y'(t) + qy(t) = f(e^t)\)

posted @ 2020-07-27 09:50  ixtwuko  阅读(2506)  评论(0编辑  收藏  举报