Autonomization refers to the rewriting of a non-autonomous differential equation system into an autonomous differential equation system in numerical mathematics .
description
Let the initial value problem be with
u
′
(
t
)
=
f
(
t
,
u
(
t
)
)
With
u
(
t
0
)
=
u
0
{\ displaystyle u '(t) = f (t, u (t)) \ quad {\ text {with}} \ quad u (t_ {0}) = u_ {0}}
given.
Then the goal is to convert this non-autonomous DGL into an autonomous, i.e. H. Transferring DGL that is independent of time (= invariant against time translation) (" autonomizing ").
t
{\ displaystyle t}
We can restrict ourselves to the beginning , since there must be a solution at any given time if the function can be autonomous .
t
0
=
0
{\ displaystyle t_ {0} = 0}
u
(
t
)
{\ displaystyle u (t)}
In the numerical method, the points in time are correspondingly omitted. That is, it applies to the explicit Euler method
u
n
+
1
=
u
n
+
H
n
f
(
u
(
t
n
)
)
{\ displaystyle u_ {n + 1} = u_ {n} + h_ {n} f (u (t_ {n}))}
for an autonomous differential equation.
Autonomy
If we now rewrite the non-autonomous DGL
in an autonomous DGL
, then we get an additional dimension.
u
′
(
t
)
=
f
(
t
,
u
(
t
)
)
,
u
∈
R.
d
,
t
∈
[
t
0
,
T
]
{\ displaystyle u '(t) = f (t, u (t)), u \ in \ mathbb {R} ^ {d}, t \ in [t_ {0}, T]}
U
′
(
t
)
=
F.
(
U
(
t
)
)
,
U
∈
R.
d
+
1
{\ displaystyle U '(t) = F (U (t)), U \ in \ mathbb {R} ^ {d + 1}}
Define
τ
=
t
-
t
0
,
U
=
(
u
τ
)
,
F.
(
U
)
=
(
f
(
τ
+
t
0
,
u
)
1
)
{\ displaystyle \ tau = t-t_ {0}, U = {\ begin {pmatrix} u \\\ tau \ end {pmatrix}}, F (U) = {\ begin {pmatrix} f (\ tau + t_ {0}, u) \\ 1 \ end {pmatrix}}}
.
Then the relationship applies:
d
U
d
τ
=
d
d
τ
(
u
τ
)
=
(
d
u
d
τ
1
)
=
(
d
u
d
t
d
t
d
τ
1
)
=
(
f
(
t
,
u
)
∗
1
1
)
=
t
=
τ
+
t
0
(
f
(
τ
+
t
0
,
u
)
1
)
=
F.
(
U
)
{\ displaystyle {\ frac {dU} {d \ tau}} = {\ frac {d} {d \ tau}} {\ begin {pmatrix} u \\\ tau \ end {pmatrix}} = {\ begin { pmatrix} {\ frac {du} {d \ tau}} \\ 1 \ end {pmatrix}} = {\ begin {pmatrix} {\ frac {du} {dt}} {\ frac {dt} {d \ tau }} \\ 1 \ end {pmatrix}} = {\ begin {pmatrix} f (t, u) * 1 \\ 1 \ end {pmatrix}} {\ underset {t = \ tau + t_ {0}} { =}} {\ begin {pmatrix} f (\ tau + t_ {0}, u) \\ 1 \ end {pmatrix}} = F (U)}
.
Examples
1. The Euler methods (explicit, implicit), the Heun method and the Trapez method are all invariant to autonomization.
2. The "crooked" Euler method (note time step in )
is not invariant to autonomization.
f
{\ displaystyle f}
u
n
+
1
=
u
n
+
H
n
f
(
t
n
+
1
,
u
n
)
{\ displaystyle u_ {n + 1} = u_ {n} + h_ {n} f (t_ {n + 1}, u_ {n})}
Individual evidence
↑ a b Kloeden, PE: Script for the lecture: "Numerical Methods for Differential Equations". (No longer available online.) February 20, 2012, p. 36 ff. , Archived from the original on January 27, 2018 ; accessed on January 30, 2018 . Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. @1 @ 2 Template: Webachiv / IABot / www.math.uni-frankfurt.de
<img src="https://de.wikipedia.org//de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">