Find Docs

Astronomical Applications Department, U.S. Naval Observatory - lyapcalc (Page 2)

Wodocs >> Science : Astronomy >> Astronomical Applications Department, U.S. Naval Observatory lyapcalc Page 2
Astronomical Applications Department, U.S. Naval Observatory - lyapcalc
(6)
k
2
=
1
t
2
-
t
0
ln
d
2
$ a
1
d
0
=
1
t
2
-
t
0
ln(a
1
$ a
2
)
k
3
=
1
t
3
-
t
0
ln
d
3
$ a
2
$ a
1
d
0
=
1
t
3
-
t
0
ln(a
1
$ a
2
$ a
3
)
®
and so on. The multiplicative factors
are derived in section 3, in
a
1
, a
1
$ a
2
,
¬
case it is not intuitively obvious. We there-
fore conclude that the instantaneous Lyapu-
nov exponent is
(7)
k
n
=
1
t
n
-
t
0
S
i
=
1
n
ln a
i
where we have defined
(8)
a
i
h
d
(
t
i
)
d
(
t
0
)
As long as the rescalings take place in the
linear regime, this construction is valid.
Notice that, in a computer, only the accumu-
lating sum of the natural log of the
i
need
be stored. In addition, the time intervals
need not be evenly spaced.
2. Renormalization of the Test Orbit.
The rescaling of the test particle orbit is per-
formed on the test - reference phase space
distance vector. Whenever the distance d(t)
becomes greater than or equal to the thresh-
old D, we scale the test particle distance
from the reference particle by the factor 1/
i
,
maintaining the current relative orientation
between the two particles in phase space.
Write the reference and test particle phase
space vectors as
(9)
R h

x
y
z
v
x
v
y
v
z

ref
and r h

x
y
z
v
x
v
y
v
z

test
Define
. Then the adjustment to
q
q h r
-
R
the test particle phase space coordinates at
time t
i
is
(10)
r
i
b R
i
+
q
q
i
a
i
Alternatively, one could write the equivalent
expression
(11)
r
i
b r
i
-
a
i
-
1
a
i
$ qq
i
Eq. (10) is slightly less expensive to calcu-
late than eq. (11). All we are doing is res-
caling the distance d(t),
(12)
d
(
t
i
)
b
d
(
t
i
)
a
i
in the appropriate direction in phase space.
I have found that
and
d
0
=
10
-
6
D
=
10
-
4
work well in practice. The figure below
shows the instantaneous Lyapunov exponent
for a chaotic restricted three-body orbit,
with several values of d
0
ranging from 10
-5
to
10
-15
, with a rescaling threshold of 10
-4
. One
can see that values of d
0
in the range 10
-5
to
10
-8
are adequate. Smaller than this invites
numerical trouble due to the finite word size
of the machine.
I have also found that any difference
between using the full phase space distance
f:\dynamics\text\lyapcalc.lwp
Maximum Lyapunov Exponent Calculation
7:07pm September 30, 1995
2

<< Previous       1 2 3 4       Next >>



Other Documents:
wedges, Scifull, Precession Memo, Precession Maple, Curves 3 D, NOVAS 2004 Overview, NOVAS 2006 Overview, thesis, Ranson,


WODocs | |
All rights reserved. wodocs.com© 2009
WODocs


New Docs

Documents Category:
Arts (Design, Movies, Music, Radio, Television)
Automotive (Cars, Marine, Motorcycle, ATV, Snowmobiles)
Business (Biotechnology and Pharmaceuticals, Chemicals, Construction and Maintenance, Materials, Real Estate, Services)
Electronics (Computers, Motion Control, Power Supply)
Games (Board, Family, Party, Card, Construction, RC Toys)
Health (Animalm, Beauty, Healthcare, Medicine, Pharmacy, Surgery, Weight Loss)
Home (Accessories, Cooking, Decor and Design, Electrical, Family, Pets)
News (Newspapers, Sports, Television)
Recreation (Collecting, Hiking, Scouting, Survival, Travel)
Reference (Education, Libraries and Archives, Museums)
Science (Agriculture, Astronomy, Biology, Chemistry)
Shopping (Antiques and Collectibles, Clothing, Flowers, Food, Home and Garden)
Sports (Bicycle, Snowboard, Skiing, Other)