First-Order Differential Equation
形如
F
(
t
,
y
,
y
′
)
=
0
F(t,y,y')=0
F(t,y,y′)=0 或
d
y
d
t
=
ϕ
(
t
,
y
)
\frac{dy}{dt}=\phi(t,y)
dtdy=ϕ(t,y)的方程式,即为一阶微分方程。例如:
y
′
−
t
2
−
1
=
0
y'-t^2-1=0
y′−t2−1=0
一阶微分方程的解
f
(
t
)
f(t)
f(t) 需满足使得
F
(
t
,
f
(
t
)
,
f
′
(
t
)
)
=
0
F(t,f(t),f'(t))=0
F(t,f(t),f′(t))=0 对于任意
t
t
t 均成立
First-Order Initial Value Problem
形如
F
(
t
,
y
,
y
′
)
=
0
,
y
(
t
0
)
=
y
0
F(t,y,y')=0,\quad y(t_0)=y_0
F(t,y,y′)=0,y(t0)=y0 的一系列方程式,即为一阶储值问题。例如
y
′
=
t
2
+
1
,
y
(
1
)
=
4
y'=t^2+1,\quad y(1)=4
y′=t2+1,y(1)=4 的解为
f
(
t
)
=
t
3
/
3
+
t
+
8
/
3
f(t)=t^3/3+t+8/3
f(t)=t3/3+t+8/3
可以尝试用多个大小不同的 h 来试试,然后列张表格啥的
像这样
(1)
d
y
d
t
=
f
(
t
,
y
)
\frac{dy}{dt}=f(t,y)
dtdy=f(t,y)
(2) y ( t 0 ) = y 0 y(t_0)=y_0 y(t0)=y0
(3) d ϕ d t ( t n ) = f ( t n , ϕ ( t n ) ) \frac{d\phi}{dt}(t_n)=f(t_n,\phi(t_n)) dtdϕ(tn)=f(tn,ϕ(tn))
(4) ϕ ( t n + 1 ) − ϕ ( t n ) t n + 1 − t n ≈ f ( t n , ϕ ( t n ) ) \frac{\phi(t_{n+1})-\phi(t_n)}{t_{n+1}-t_n}\approx f(t_n,\phi(t_n)) tn+1−tnϕ(tn+1)−ϕ(tn)≈f(tn,ϕ(tn))
Euler
(5)
y
n
+
1
=
y
n
+
h
f
(
t
n
,
y
n
)
y_{n+1}=y_n+hf(t_n,y_n)
yn+1=yn+hf(tn,yn)
Backward Euler
(6)
y
n
+
1
=
y
n
+
h
f
(
t
n
+
1
,
y
n
+
1
)
y_{n+1}=y_n+hf(t_{n+1},y_{n+1})
yn+1=yn+hf(tn+1,yn+1)
Global Truncation Error
无限精度的情况下
difference between solution
y
=
ϕ
(
t
)
y=\phi(t)
y=ϕ(t) and the numerical approximation
y
n
y_n
yn at
t
n
t_n
tn
(7)
E
n
=
ϕ
(
t
n
)
−
y
n
E_n=\phi(t_n)-y_n
En=ϕ(tn)−yn
Round-off Error
但是现实的精度肯定是有限的
difference between numerical approximation ideally
y
n
y_n
yn and that practicingly
Y
n
Y_n
Yn (即实际通过数值法计算出来的结果)
(8)
R
n
=
y
n
−
Y
n
R_n=y_n-Y_n
Rn=yn−Yn
(9) ∣ ϕ ( t n ) − Y n ∣ = ∣ ϕ ( t n ) − y n + y n − Y n ∣ ≤ ∣ ϕ ( t n ) − y n ∣ + ∣ y n − Y n ∣ = ∣ E n ∣ + ∣ R n ∣ |\phi(t_n)-Y_n|=|\phi(t_n)-y_n+y_n-Y_n| \\ \leq |\phi(t_n)-y_n|+|y_n-Y_n| = |E_n| + |R_n| ∣ϕ(tn)−Yn∣=∣ϕ(tn)−yn+yn−Yn∣≤∣ϕ(tn)−yn∣+∣yn−Yn∣=∣En∣+∣Rn∣
Improved Euler
y
n
+
1
=
y
n
+
f
(
t
n
,
y
n
)
+
f
(
t
n
+
1
,
y
n
+
1
)
2
h
y_{n+1}=y_n+\frac{f(t_n,y_n)+f(t_{n+1},y_{n+1})}{2}h
yn+1=yn+2f(tn,yn)+f(tn+1,yn+1)h