Beruflich Dokumente
Kultur Dokumente
j=0
a
j
y
nj
+ h
p
j=1
b
j
f(x
nj
, y
nj
), n p
In this, p 0 is a given integer, and
_
a
j
_
and
_
b
j
_
are given coecients.
This is called a (p + 1)-step method.
EXAMPLES
(1) The midpoint rule:
y
n+1
= y
n1
+ 2hf (x
n
, y
n
) , n = 1, 2, 3, ...
This is a two-step method. Note that y
0
= Y
0
in gen-
eral; but y
1
must be obtained by some other means.
y
2
is obtained from y
0
and y
1
; and similarly for y
3
,
etc.
In this case, p = 1.
(2) The trapezoidal rule:
y
n+1
= y
n
+
h
2
[f (x
n
, y
n
) + f (x
n+1
, y
n+1
)] , n 0
This is a single step method. It is also our rst ex-
ample of an implicit method; and many of the most
desirable formulas are of this type. In this case, p = 0.
PROBLEMS
How do we derive multistep methods?
What is the truncation error?
Does the method converge? If so, how fast?
Is the method stable?
Is there an asymptotic error formula?
TRUNCATION ERROR
For any dierentiable function Y (x), dene the trun-
cation error in going from x
n
to x
n+1
by
T
n
(Y ) = Y (x
n+1
)
_
_
p
j=0
a
j
Y (x
nj
)
+h
p
j=1
b
j
Y
(x
nj
)
_
_
, n p
By how much, does the numerical formula fail to pre-
dict Y (x
n+1
) from the preceding values?
For the trapezoidal rule, we can show
T
n
(Y ) =
h
3
12
Y
(
n
), x
n
n
x
n+1
For the midpoint rule, we can show
T
n
(Y ) =
h
3
3
Y
(
n
), x
n1
n
x
n+1
CONSISTENCY
Introduce
n
(Y ) =
1
h
T
n
(Y ), n p
(h; Y ) = max
pnN1
|
n
(Y )|
where the node points in [x
0
, b] are
x
0
< x
1
< < x
N
b
We say the multistep method is consistent if for every
function Y (x) that is continuously dierentiable on
[x
0
, b],
(h; Y ) 0 as h 0
The method has order m if
(h; Y ) = O(h
m
)
for all such functions Y (x).
When is a method consistent? When is it of order m?
EXAMPLE
Consider the midpoint method
y
n+1
= y
n1
+ 2hf (x
n
, y
n
) , n = 1, 2, 3, ...
In this case,
T
n
(Y ) = Y (x
n+1
) Y (x
n1
) 2hY
(x
n
)
n
(Y ) = 2
_
Y (x
n+1
) Y (x
n1
)
2h
Y
(x
n
)
_
Thus consistency in this case says that
Y (x
n+1
) Y (x
n1
)
2h
Y
(x
n
)
which seems a reasonable request in solving a dier-
ential equation.
THEOREM
(a) In order that the method be consistent, it is nec-
essary and sucient that
p
j=0
a
j
= 1
p
j=1
b
j
j=0
ja
j
= 1
(b) In order that (h; Y ) = O(h
m
) for some m
1, it is necessary and sucient that the method be
consistent and that
p
j=0
(j)
i
a
j
+ i
p
j=1
(j)
i1
b
j
= 1
for i = 2, ..., m.
In this case that (h; Y ) = O(h
m
),
T
n
(Y ) =
c
m+1
(m + 1)!
h
m+1
Y
(m+1)
(x
n
) + O
_
h
m+2
_
with
c
m+1
= 1
_
_
p
j=0
(j)
m+1
a
j
+ (m + 1)
p
j=1
(j)
m
b
j
_
_
In addition,
(h; Y )
c
m+1
(m + 1)!
h
m
_
_
_Y
(m+1)
_
_
_
+ O
_
h
m+1
_
DERIVATION OF TRUNCATION ERROR
Recall the formula
T
n
(Y ) = Y (x
n+1
)
_
_
p
j=0
a
j
Y (x
nj
)
+h
p
j=1
b
j
Y
(x
nj
)
_
_
, n p
We expand Y (x) and Y
(x) about x
n
with x = x
n+1
,
x
n1
, ..., x
np
.
Y (x) = Y (x
n
) + (x x
n
)Y
(x
n
) +
+
(x x
n
)
m
m!
Y
(m)
(x
n
)
+
(x x
n
)
m+1
(m + 1)!
Y
(m+1)
(x
n
) +
Y
(x) = Y
(x
n
) + (x x
n
)Y
(x
n
) +
+
(x x
n
)
m1
(m1)!
Y
(m)
(x
n
)
+
(x x
n
)
m
m!
Y
(m+1)
(x
n
) +
We substitute these into the formula for the trunca-
tion error, rearrange the terms, and obtain the formula
T
n
(Y ) =
m+1
i=0
c
i
i!
h
i
Y
(i)
(x
n
) + O(h
m+2
)
c
0
= 1
p
j=0
a
j
c
1
= 1
_
_
j=0
ja
j
+
p
j=1
b
j
_
_
c
i
= 1
_
_
p
j=0
(j)
i
a
j
+ i
p
j=1
(j)
i1
b
j
_
_
for i 2.
For the consistency, we want to look at
n
(Y ) =
1
h
T
n
(Y )
n
(Y ) =
c
0
h
Y (x
n
) + c
1
Y
(x
n
) +
c
2
2
hY
(x
n
) +
and we want it to go to zero as h goes to zero. This
requires c
0
= 0 and c
1
= 0. These are exactly the
equations asserted earlier. Consistency makes the er-
ror exactly zero whenever Y (x) is a linear function.
To have
n
(Y ) = O(h
m
), we must also have
c
2
= ... = c
m
= 0
Then
T
n
(Y ) =
c
m+1
(m + 1)!
h
m+1
Y
(m+1)
(x
n
) + O
_
h
m+2
_
CONVERGENCE
The general multistep method for solving
y
= f(x, y), x x
0
, y(x
0
) = Y
0
is given by
y
n+1
=
p
j=0
a
j
y
nj
+ h
p
j=1
b
j
f(x
nj
, y
nj
), n p
(1)
with y
0
.
= Y
0
. The values y
1
, ..., y
p
must be calcu-
lated by other means, sometimes by using a single step
method. To talk about the accuracy of the method
(1), we must also consider the accuracy in the initial
values y
1
, ..., y
p
; and these will depend on the stepsize
h, since
y
i
= y
h
(x
i
) Y (x
0
+ ih), i = 0, 1, ..., p
Introduce the maximum of the initial errors,
(h; Y ) = max
j=0,1,...,p
|Y (x
0
+ ih) y
h
(x
0
+ ih)|
Also recall the quantity
(h; Y ) =
1
h
max
pnN1
|T
n
(Y )|
where x
0
, ..., x
N
are the node points on [x
0
, b].
THEOREM
Assume: (a) The multistep method is consistent;
(b) All a
j
0.
Then for all suciently small values of h, say 0 < h
h
0
for some choice of h
0
, the multistep method has
a solution y
n+1
y
h
(x
n+1
) at each node point. In
addition, we have
max
x
0
x
j
b
Y (x
j
) y
h
(x
j
)
c
1
(h; Y ) + c
2
(h; Y )
with suitably chosen constants c
1
, c
2
which depend on
the function f(x, y), but not on the stepsize h.
EXAMPLE. Recall the midpoint rule
y
n+1
= y
n1
+ 2hf (x
n
, y
n
) , n = 1, 2, 3, ...
in which p = 1, a
0
= 0, a
1
= 1. In this case,
(h; Y ) = max {|Y (x
0
) y
h
(x
0
)| , |Y (x
1
) y
h
(x
1
)|}
From earlier, I said
T
n
(Y ) =
h
3
3
Y
(
n
), x
n1
n
x
n+1
Then
(h; Y )
h
2
3
_
_
_Y
_
_
_
Y (x
j
) y
h
(x
j
)
c
1
(h; Y ) + c
2
h
2
3
_
_
_Y
_
_
_
We usually have y
0
= Y
0
; and therefore, we should
choose y
1
with an accuracy of
Y (x
1
) y
h
(x
1
) = O(h
2
)
THE PROOF
Recall that
T
n
(Y ) = Y (x
n+1
)
_
_
p
j=0
a
j
Y (x
nj
)
+h
p
j=1
b
j
Y
(x
nj
)
_
_
, n p
n
(Y ) =
1
h
T
n
(Y ), n p
Y
j=0
a
j
Y (x
nj
)
+h
p
j=1
b
j
f(x
nj
, Y (x
nj
))
+h
n
(Y ), n p
The multistep method is
y
n+1
=
p
j=0
a
j
y
nj
+ h
p
j=1
b
j
f(x
nj
, y
nj
), n p
We subtract these to get
e
n+1
=
p
j=0
a
j
e
nj
+h
p
j=1
b
j
_
f(x
nj
, Y (x
nj
)) f(x
nj
, y
nj
)
_
+h
n
(Y ), n p
with e
j
Y (x
j
) y
h
(x
j
).
Recall the Lipschitz condition
|f(x, y) f(x, z)| K|y z|
Then
|e
n+1
|
p
j=0
a
j
e
nj
+hK
p
j=1
b
j
e
nj
+h
n
(Y ), n p
Introduce
f
j
= max
i=0,1,...j
|e
i
|
the maximum of the errors up thru x
j
. Then
|e
n+1
| f
n
p
j=0
a
j
+ hKf
n+1
p
j=1
b
j
+h(h; Y ), n p
Using consistency,
p
j=0
a
j
= 1
Introduce
c = K
p
j=1
b
j
j=0
a
j
y
nj
+ h
p
j=1
b
j
f(x
nj
, y
nj
), n p
Suppose we perturb the initial values y
0
, ..., y
p
to z
0
, ..., z
p
,
with
|y
i
z
i
| , i = 0, 1, ..., p
Consider the numerical solution
z
n+1
=
p
j=0
a
j
z
nj
+ h
p
j=1
b
j
f(x
nj
, z
nj
), n p
Then by methods similar to the above, we can show
(with the same assumptions) that
|y
n
z
n
| e
2c(x
n
x
0
)
OTHER ANALYSES
We can also do asymptotic error analyses, rounding
error analyses, and analyses involving systems of rst
order ODEs. Rather than doing a general theory, we
do special cases in later sections.