叠加原理和齐次化原理
从通解求特解的方法对多数偏微分方程定解问题不适用。求解线性偏微分方程定解问题较多采用以下做法:先求方程的一族特殊解,再将这族特殊解作适当线性叠加求得定解问题的特解。接下将介绍这种做法的理论根据——线性叠加原理,以及由叠加原理导出的处理非齐次方程初值问题的齐次化原理。
1. 线性叠加原理
n个自变量的二阶线性偏微分方程
∑
i
,
j
=
1
n
a
i
j
∂
2
u
∂
x
i
∂
x
j
+
∑
j
=
1
n
b
j
∂
u
∂
x
j
+
c
u
=
f
\sum_{i,j=1}^na_{ij}\frac{\partial^2u}{\partial x_i\partial x_j}+\sum_{j=1}^nb_j\frac{\partial u}{\partial x_j}+cu=f
i,j=1∑naij∂xi∂xj∂2u+j=1∑nbj∂xj∂u+cu=f
引入偏微分算子
L
=
∑
i
,
j
=
1
n
a
i
j
∂
2
∂
x
i
∂
x
j
+
∑
j
=
1
n
b
j
∂
∂
x
j
+
c
(1)
L=\sum_{i,j=1}^na_{ij}\frac{\partial^2}{\partial x_i\partial x_j}+\sum_{j=1}^nb_j\frac{\partial}{\partial x_j}+c \tag{1}
L=i,j=1∑naij∂xi∂xj∂2+j=1∑nbj∂xj∂+c(1)
则可简单表示为
L
u
(
x
)
=
f
(
x
)
,
x
=
(
x
1
,
x
2
,
⋅
⋅
⋅
,
x
n
)
Lu(\bold x)=f(\bold x),\quad \bold x=(x_1,x_2,···,x_n)
Lu(x)=f(x),x=(x1,x2,⋅⋅⋅,xn)
一般地,称从一个函数类(定义域)到另一个函数类(值域)的映射T为一个算子。如果定义域与值域都是数域
Λ
\Lambda
Λ上的线性空间,对定义域中任意两个函数
u
1
,
u
2
u_1,u_2
u1,u2和
Λ
\Lambda
Λ中任意两个数
λ
1
,
λ
2
\lambda_1,\lambda_2
λ1,λ2有
T
(
λ
1
u
1
+
λ
2
u
2
)
=
λ
1
T
u
1
+
λ
2
T
u
2
T(\lambda_1u_1+\lambda_2u_2)=\lambda_1Tu_1+\lambda_2Tu_2
T(λ1u1+λ2u2)=λ1Tu1+λ2Tu2,则称T是线性算子。
显然,上述二阶偏微分算子(1)式是函数空间
C
2
C^2
C2到
C
C
C的线性算子,拉普拉斯算子
Δ
3
=
∂
2
∂
x
2
+
∂
2
∂
y
2
+
∂
2
∂
z
2
\Delta_3=\frac{\partial^2}{\partial x^2}+\frac{\partial^2}{\partial y^2}+\frac{\partial^2}{\partial z^2}
Δ3=∂x2∂2+∂y2∂2+∂z2∂2是其特殊情况。Fourier变换
F
[
f
(
x
)
]
=
∫
−
∞
+
∞
f
(
x
)
e
−
i
λ
x
d
x
F[f(x)]=\int_{-\infty}^{+\infty}f(x)e^{-i\lambda x}dx
F[f(x)]=∫−∞+∞f(x)e−iλxdx,Laplace变换
L
[
f
(
t
)
]
=
∫
0
+
∞
f
(
t
)
e
−
p
t
d
t
L[f(t)]=\int^{+\infty}_0f(t)e^{-pt}dt
L[f(t)]=∫0+∞f(t)e−ptdt都是线性积分算子。若记
L
1
=
l
i
m
t
→
0
,
L
2
=
l
i
m
t
→
0
∂
∂
t
,
L
3
=
l
i
m
x
→
x
0
(
α
+
β
∂
∂
n
)
L_1=lim_{t\to 0},L_2=lim_{t\to 0}\frac{\partial}{\partial t},L_3=lim_{x\to x_0}(\alpha+\beta\frac{\partial}{\partial n})
L1=limt→0,L2=limt→0∂t∂,L3=limx→x0(α+β∂n∂),则初始条件、边界条件均可表示为线性算子形式
L
j
u
(
x
)
=
φ
(
x
)
L_ju(x)=\varphi(x)
Lju(x)=φ(x)
由线性算子的定义,我们有
定理1:(叠加原理)设L是关于 x = ( x 1 , x 2 , ⋅ ⋅ ⋅ , x n ) \bold x=(x_1,x_2,···,x_n) x=(x1,x2,⋅⋅⋅,xn)的任意阶线性微分算子(常或偏),则有
(1)有限叠加原理:若 u j ( x ) u_j(\bold x) uj(x)满足 L u j ( x ) = f j ( x ) ( j = 1 , 2 , ⋅ ⋅ ⋅ , m ) Lu_j(\bold x)=f_j(\bold x)(j=1,2,···,m) Luj(x)=fj(x)(j=1,2,⋅⋅⋅,m),则当 u ( x ) = ∑ j = 1 m λ j u j ( x ) u(\bold x)=\sum_{j=1}^m\lambda_ju_j(x) u(x)=∑j=1mλjuj(x)时,有 L u ( x ) = ∑ j = 1 m λ j f j ( x ) , λ j ∈ Λ , j = 1 , 2 , ⋅ ⋅ ⋅ , m Lu(\bold x)=\sum_{j=1}^m\lambda_jf_j(x),\quad\lambda_j\in \Lambda,\quad j=1,2,···,m Lu(x)=∑j=1mλjfj(x),λj∈Λ,j=1,2,⋅⋅⋅,m。
(2)级数叠加原理:若 u j ( x ) u_j(x) uj(x)满足 L u j ( x ) = f j ( x ) ( j = 1 , 2 , ⋅ ⋅ ⋅ ) , Lu_j(\bold x)=f_j(\bold x)(j=1,2,···), Luj(x)=fj(x)(j=1,2,⋅⋅⋅),则当 u ( x ) = ∑ j = 1 ∞ λ j u j ( x ) u(\bold x)=\sum_{j=1}^\infty \lambda_ju_j(x) u(x)=∑j=1∞λjuj(x)时,有 L u ( x ) = ∑ j = 1 ∞ λ j f j ( x ) , λ j ∈ Λ , j = 1 , 2 , ⋅ ⋅ ⋅ Lu(x)=\sum_{j=1}^\infty \lambda_jf_j(\bold x), \space \lambda_j\in \Lambda,j=1,2,··· Lu(x)=∑j=1∞λjfj(x), λj∈Λ,j=1,2,⋅⋅⋅
(3)积分叠加原理:若 u ( x ; ξ ) u(\bold x;\xi) u(x;ξ)满足 L u ( x ; ξ ) = f ( x ; ξ ) , ξ ∈ V Lu(\bold x;\bold \xi)=f(\bold x;\bold \xi),\quad \xi\in V Lu(x;ξ)=f(x;ξ),ξ∈V,则当 U ( x ) = ∫ V λ ( ξ ) u ( x ; ξ ) d ξ U(\bold x)=\int_V\lambda(\bold \xi)u(\bold x;\bold \xi)d\xi U(x)=∫Vλ(ξ)u(x;ξ)dξ时,有 L U ( x ) = ∫ V λ ( ξ ) f ( x ; ξ ) d ξ , λ ( ξ ) ∈ Λ , ξ ∈ V LU(\bold x)=\int_V \lambda(\xi)f(\bold x;\bold \xi)d\xi, \lambda(\xi)\in \Lambda, \xi\in V LU(x)=∫Vλ(ξ)f(x;ξ)dξ,λ(ξ)∈Λ,ξ∈V
有限叠加原理是线性算子定义的直接推广,而级数叠加原理及积分叠加原理分别要求有关函数项级数、含参量积分收敛,算子L与求和号、积分号可交换次序。在经典意义下这些条件要求很高,实际问题不一定能满足,但是在推广意义下,这种交换总是可进行的。今后,我们将不受限制地使用这些叠加原理。
在物理问题中,方程或定解条件中的非齐次项通常反映引起物理过程的源,线性叠加原理则说明多个源共同作用的结果等于各个源单独作用的总和,这在物理上称为独立作用原理,这是线性问题的基本特征。利用叠加原理我们总可以把一个较复杂的线性问题分解成若干简单线性问题来求解。
例1:求解狄利克雷问题
{
∂
2
u
∂
x
2
+
∂
2
u
∂
y
2
=
1
,
x
2
+
y
2
=
1
u
∣
x
2
+
y
2
=
1
=
x
2
\begin{cases} \frac{\partial^2u}{\partial x^2}+\frac{\partial^2u}{\partial y^2}=1,\quad x^2+y^2=1 \\ u|_{x^2+y^2=1}=x^2 \end{cases}
{∂x2∂2u+∂y2∂2u=1,x2+y2=1u∣x2+y2=1=x2
解:易见
u
0
=
1
4
(
x
2
+
y
2
)
u_0=\frac{1}{4}(x^2+y^2)
u0=41(x2+y2)是非齐次泛定方程的一个特解。令
v
=
u
−
u
0
v=u-u_0
v=u−u0,由叠加原理知,v满足边值问题
{
∂
2
v
∂
x
2
+
∂
2
v
∂
y
2
=
0
v
∣
x
2
+
y
2
=
1
=
x
2
−
1
4
\begin{cases} \frac{\partial^2v}{\partial x^2}+\frac{\partial^2v}{\partial y^2}=0 \\ v|_{x^2+y^2=1}=x^2-\frac{1}{4} \end{cases}
{∂x2∂2v+∂y2∂2v=0v∣x2+y2=1=x2−41
解析函数
z
n
(
z
=
x
+
i
y
)
z^n(z=x+iy)
zn(z=x+iy)的实、虚部均为调和函数,故
1
,
x
,
y
,
x
2
−
y
2
,
x
y
1,x,y,x^2-y^2,xy
1,x,y,x2−y2,xy为次数不超过2的调和多项式,故可设
v
=
v
0
+
a
1
x
+
a
2
y
+
a
3
(
x
2
−
y
2
)
+
a
4
x
y
v=v_0+a_1x+a_2y+a_3(x^2-y^2)+a_4xy
v=v0+a1x+a2y+a3(x2−y2)+a4xy
由叠加原理知,对任意一组
{
a
j
,
j
=
0
,
1
,
⋅
⋅
⋅
,
4
}
\{a_j,j=0,1,···,4\}
{aj,j=0,1,⋅⋅⋅,4},v都是拉普拉斯方程的解。代入边界条件
v
∣
x
2
+
y
2
=
1
=
a
0
+
a
1
x
+
a
2
1
−
x
2
+
a
3
(
2
x
2
−
1
)
+
a
4
x
1
−
x
2
=
x
2
−
1
4
v|_{x^2+y^2=1}=a_0+a_1x+a_2\sqrt{1-x^2}+a_3(2x^2-1)+a_4x\sqrt{1-x^2}=x^2-\frac{1}{4}
v∣x2+y2=1=a0+a1x+a21−x2+a3(2x2−1)+a4x1−x2=x2−41
比较系数,可得
v
=
1
4
+
1
2
(
x
2
−
y
2
)
v=\frac{1}{4}+\frac{1}{2}(x^2-y^2)
v=41+21(x2−y2)
从而
u
=
3
4
x
2
−
1
4
y
2
+
1
4
u=\frac{3}{4}x^2-\frac{1}{4}y^2+\frac{1}{4}
u=43x2−41y2+41
对于非线性方程,自然不存在线性叠加原理,因此求解要困难得多。对某些特殊的非线性方程,如果能发现某种形式的叠加原理,则对此非线性问题的研究有重大意义。
2. 齐次化原理(冲量原理)
例1中通过求非齐次方程的一个特解将方程齐次化。齐次化原理将解决一般非齐次发展方程齐次化的问题。
仍以理想弦的横振动为例,考虑在外力作用下弦的纯受迫振动
{
∂
2
u
∂
t
2
=
a
2
∂
2
u
∂
x
2
+
f
(
t
,
x
)
,
t
>
0
,
−
∞
<
x
<
+
∞
u
∣
t
=
0
=
0
,
∂
u
∂
t
∣
t
=
0
=
0
(2)
\begin{cases} \frac{\partial^2u}{\partial t^2}=a^2\frac{\partial^2u}{\partial x^2}+f(t,x),\quad t>0,\quad -\infty<x<+\infty \\ u|_{t=0}=0,\quad \frac{\partial u}{\partial t}|_{t=0}=0 \end{cases} \tag{2}
{∂t2∂2u=a2∂x2∂2u+f(t,x),t>0,−∞<x<+∞u∣t=0=0,∂t∂u∣t=0=0(2)
这里,
f
(
t
,
x
)
f(t,x)
f(t,x)是作用在弦身单位质量上的外力,
u
(
t
,
x
)
u(t,x)
u(t,x)是由此外力从初始时刻
t
=
0
t=0
t=0持续到t时刻作用引起的位移。由独立作用原理,
u
(
t
,
x
)
u(t,x)
u(t,x)可看成前后相继的瞬间外力作用冲量
f
(
τ
,
x
)
d
τ
(
0
≤
τ
≤
t
)
f(\tau,x)d\tau(0\leq \tau \leq t)
f(τ,x)dτ(0≤τ≤t)在t时刻引起的位移
ω
(
t
,
x
;
τ
)
d
r
\omega(t,x;\tau)dr
ω(t,x;τ)dr关于
d
r
dr
dr的叠加,即
u
(
t
,
x
)
=
∫
0
t
ω
(
t
,
x
;
τ
)
d
τ
u(t,x)=\int_0^t\omega(t,x;\tau)d\tau
u(t,x)=∫0tω(t,x;τ)dτ。x显然,当
t
<
τ
t<\tau
t<τ时
ω
(
t
,
x
;
τ
)
≡
0
\omega(t,x;\tau)\equiv 0
ω(t,x;τ)≡0,当
t
>
τ
t>\tau
t>τ时,
τ
\tau
τ时刻瞬间冲量的作用已转化为从
t
=
τ
−
0
t=\tau-0
t=τ−0到
t
=
τ
+
0
t=\tau+0
t=τ+0动量的增加,故
w
(
t
,
x
;
τ
)
w(t,x;\tau)
w(t,x;τ)应满足
{
∂
2
ω
∂
t
2
=
a
2
∂
2
ω
∂
x
2
,
t
>
τ
,
−
∞
<
x
<
+
∞
w
∣
t
=
0
=
0
,
∂
w
∂
t
∣
t
=
τ
=
f
(
τ
,
x
)
(3)
\begin{cases} \frac{\partial^2 \omega}{\partial t^2}=a^2\frac{\partial^2 \omega}{\partial x^2},\quad t>\tau, \space -\infty<x<+\infty \\ w|_{t=0}=0, \quad \frac{\partial w}{\partial t}|_{t=\tau}=f(\tau,x) \end{cases} \tag{3}
{∂t2∂2ω=a2∂x2∂2ω,t>τ, −∞<x<+∞w∣t=0=0,∂t∂w∣t=τ=f(τ,x)(3)
而(2)的解则为
u
(
t
,
x
)
=
∫
0
t
w
(
t
,
x
;
τ
)
d
τ
(4)
u(t,x)=\int_0^tw(t,x;\tau)d\tau \tag{4}
u(t,x)=∫0tw(t,x;τ)dτ(4)
实际上,当
w
(
t
,
x
;
τ
)
w(t,x;\tau)
w(t,x;τ)是问题(3)的解时,由(4)可验证
u
∣
t
=
0
=
0
∂
u
∂
t
=
w
(
t
,
x
;
τ
)
∣
τ
=
t
+
∫
0
t
∂
ω
(
t
,
x
;
τ
)
∂
t
d
τ
=
∫
0
t
∂
ω
(
t
,
x
;
r
)
∂
t
d
τ
∂
u
∂
t
∣
t
=
0
=
0
∂
2
u
∂
t
2
=
∂
ω
(
t
,
x
;
r
)
∂
t
∣
τ
=
t
+
∫
0
t
∂
2
ω
(
t
,
x
;
τ
)
∂
t
2
d
r
=
f
(
t
,
x
)
+
a
2
∫
0
t
∂
2
w
(
t
,
x
;
r
)
∂
x
2
d
r
=
f
(
t
,
x
)
+
a
2
∂
2
u
∂
x
2
u|_{t=0}=0 \\ \frac{\partial u}{\partial t}=w(t,x;\tau)|_{\tau=t}+\int_0^t\frac{\partial \omega(t,x;\tau)}{\partial t}d\tau=\int_0^t\frac{\partial \omega(t,x;r)}{\partial t}d\tau \\ \frac{\partial u}{\partial t}|_{t=0}=0 \\ \frac{\partial^2u}{\partial t^2}=\frac{\partial \omega(t,x;r)}{\partial t}|_{\tau=t}+\int_0^t\frac{\partial^2\omega(t,x;\tau)}{\partial t^2}dr=f(t,x)+a^2\int_0^t\frac{\partial^2w(t,x;r)}{\partial x^2}dr=f(t,x)+a^2\frac{\partial^2 u}{\partial x^2}
u∣t=0=0∂t∂u=w(t,x;τ)∣τ=t+∫0t∂t∂ω(t,x;τ)dτ=∫0t∂t∂ω(t,x;r)dτ∂t∂u∣t=0=0∂t2∂2u=∂t∂ω(t,x;r)∣τ=t+∫0t∂t2∂2ω(t,x;τ)dr=f(t,x)+a2∫0t∂x2∂2w(t,x;r)dr=f(t,x)+a2∂x2∂2u
故式(4)给出了纯受迫振动(2)的解。在(3)中令
t
′
=
t
−
τ
,
ω
(
t
,
x
;
τ
)
t'=t-\tau,\omega(t,x;\tau)
t′=t−τ,ω(t,x;τ)可由达朗贝尔公式得到,故纯受迫振动
u
(
t
,
x
)
=
1
2
a
∫
0
t
d
τ
∫
x
−
a
(
t
−
τ
)
x
+
a
(
t
−
τ
)
f
(
τ
,
ξ
)
d
ξ
u(t,x)=\frac{1}{2a}\int_0^td\tau \int_{x-a(t-\tau)}^{x+a(t-\tau)}f(\tau,\xi)d\xi
u(t,x)=2a1∫0tdτ∫x−a(t−τ)x+a(t−τ)f(τ,ξ)dξ
若作变量代换
r
=
a
(
t
−
τ
)
r=a(t-\tau)
r=a(t−τ),则
u
(
t
,
x
)
=
1
2
a
2
∫
0
a
t
d
r
∫
x
−
r
x
+
r
f
(
t
−
r
a
,
ξ
)
d
ξ
u(t,x)=\frac{1}{2a^2}\int_0^{at}dr\int_{x-r}^{x+r}f(t-\frac{r}{a},\xi)d\xi
u(t,x)=2a21∫0atdr∫x−rx+rf(t−ar,ξ)dξ
由此可见纯外力引起的x点t时刻的位移受
[
x
−
a
t
,
x
+
a
t
]
[x-at,x+at]
[x−at,x+at]内各点所受外力的影响,而在
[
x
−
r
,
x
+
r
]
(
0
≤
r
<
a
t
)
[x-r,x+r](0\leq r<at)
[x−r,x+r](0≤r<at)上的外力是
t
−
r
a
t-\frac{r}{a}
t−ar时刻的
由叠加原理可写出一般弦振动方程初值问题
{
∂
2
u
∂
t
2
=
a
2
∂
2
u
∂
x
2
+
f
(
t
,
x
)
,
t
>
0
,
−
∞
<
x
<
+
∞
u
∣
t
=
0
=
φ
(
x
)
,
∂
u
∂
t
∣
t
=
0
=
ψ
(
x
)
\begin{cases} \frac{\partial^2u}{\partial t^2}=a^2\frac{\partial^2u}{\partial x^2}+f(t,x), \quad t>0,-\infty<x<+\infty \\ u|_{t=0}=\varphi(x),\quad \frac{\partial u}{\partial t}|_{t=0}=\psi(x) \end{cases}
{∂t2∂2u=a2∂x2∂2u+f(t,x),t>0,−∞<x<+∞u∣t=0=φ(x),∂t∂u∣t=0=ψ(x)
的解
u
(
t
,
x
)
=
1
2
[
φ
(
x
−
a
t
)
+
φ
(
x
+
a
t
)
]
+
1
2
a
∫
x
−
a
t
x
+
a
t
ψ
(
ξ
)
d
ξ
+
1
2
a
∫
0
t
d
r
∫
x
−
a
(
t
−
τ
)
x
+
a
(
t
−
τ
)
f
(
τ
,
ξ
)
d
ξ
u(t,x)=\frac{1}{2}[\varphi(x-at)+\varphi(x+at)]+\frac{1}{2a}\int_{x-at}^{x+at}\psi(\xi)d\xi+\frac{1}{2a}\int_0^tdr\int_{x-a(t-\tau)}^{x+a(t-\tau)}f(\tau,\xi)d\xi
u(t,x)=21[φ(x−at)+φ(x+at)]+2a1∫x−atx+atψ(ξ)dξ+2a1∫0tdr∫x−a(t−τ)x+a(t−τ)f(τ,ξ)dξ
将此齐次化方法推广到一般发展方程的初值问题,有如下原理。
定理:(齐次化原理)设L是关于t与
x
=
(
x
1
,
x
2
,
⋅
⋅
⋅
,
x
n
)
x=(x_1,x_2,···,x_n)
x=(x1,x2,⋅⋅⋅,xn)中各分量的线性偏微分算子,其中关于t的最高阶导数不超过
m
−
1
m-1
m−1阶。若
ω
(
t
,
x
;
τ
)
\omega(t,x;\tau)
ω(t,x;τ)满足齐次方程初值问题
{
∂
m
w
∂
t
m
=
L
w
,
t
>
τ
>
0
,
x
∈
R
n
w
∣
t
=
τ
=
∂
w
∂
t
∣
t
=
τ
=
⋅
⋅
⋅
=
∂
m
−
2
ω
∂
t
m
−
2
∣
t
=
r
=
0
,
∂
m
−
1
w
∂
t
m
−
1
∣
t
=
τ
=
f
(
τ
,
x
)
(5)
\begin{cases} \frac{\partial^mw}{\partial t^m}=Lw,\quad t>\tau>0,x\in R^n \\ w|_{t=\tau}=\frac{\partial w}{\partial t}|_{t=\tau}=···=\frac{\partial^{m-2}\omega}{\partial t^{m-2}}|_{t=r}=0 ,\\ \frac{\partial^{m-1}w}{\partial t^{m-1}}|_{t=\tau}=f(\tau,\bold x) \end{cases} \tag{5}
⎩⎪⎨⎪⎧∂tm∂mw=Lw,t>τ>0,x∈Rnw∣t=τ=∂t∂w∣t=τ=⋅⋅⋅=∂tm−2∂m−2ω∣t=r=0,∂tm−1∂m−1w∣t=τ=f(τ,x)(5)
则
u
(
t
,
x
)
=
∫
0
t
w
(
t
,
x
;
τ
)
d
τ
(6)
u(t,\bold x)=\int_0^tw(t,\bold x;\tau)d\tau \tag{6}
u(t,x)=∫0tw(t,x;τ)dτ(6)
是非齐次方程初值问题
{
∂
m
u
∂
t
m
=
L
u
+
f
(
t
,
w
)
,
t
>
0
,
x
∈
R
n
u
∣
t
=
0
=
∂
u
∂
t
∣
t
=
0
=
⋅
⋅
⋅
=
∂
m
−
1
u
∂
t
m
−
1
∣
t
=
0
=
0
(7)
\begin{cases} \frac{\partial^mu}{\partial t^m}=Lu+f(t,w),\quad t>0,\bold x\in R^n \\ u|_{t=0}=\frac{\partial u}{\partial t}|_{t=0}=···=\frac{\partial^{m-1}u}{\partial t^{m-1}}|_{t=0}=0 \end{cases} \tag{7}
{∂tm∂mu=Lu+f(t,w),t>0,x∈Rnu∣t=0=∂t∂u∣t=0=⋅⋅⋅=∂tm−1∂m−1u∣t=0=0(7)
的解。
定理不难直接验证。因为初值问题是适定的,式(6)给出了初值问题(7)的唯一解。当 x ∈ V ⊂ R n x\in V \subset R^n x∈V⊂Rn,齐次化原理仍然成立,只需在式(5)和式(7)中分别增加齐次边界条件 L 1 w ∣ ∂ V = 0 L_1w|_{\partial V=0} L1w∣∂V=0和 L 1 u ∣ ∂ V = 0 L_1u|_{\partial V}=0 L1u∣∂V=0, L 1 L_1 L1是关于x的线性偏微分算子。这就是混合问题的齐次化原理。由齐次化原理知,今后对线性发展方程定解问题的讨论主要集中于齐次方程。