Week1+Week2 Notes
Week1+Week2 Notes
Ivp
y Hl Hy T oates
yeol I
since this is a linear
ODE
y 7
ency f eat 1C
f Cet
ye
I 1
substituting the initial
condition
yHl s
to
By chopping
1 101 0.3333
4
to
c
2 Ct I Ig'ext
error
is not
significant
If X o
then the
error can be come
t
significant as
grows
The al of well
g
posedness is to make
sure that the round off
error is
neglible
is well
Ivp posed
4.2.21
Friday
The of introducing
goal
well is to male
posedless
sun that the round off
is
error
insignificant
conditions
Lipschitz
f Ct satisfies a
Lipschitz
y
condition in set
y on a
A sufficient condition
for a function f
to
satisfy
condition
a
Lipschitz
is if a constant L o exists
such that
i I let fit b
p
Theorem 5 6
suppose
that D Let.pl If's
If f is continuous and
satisfies a
Lipschitz condition
ftp.flt.y d
a Eb
yeah
is well posed
Example
dy
It thy 11 octet
11101 1
show that the
Ivp
problem above is well posed
Solution
The function is continuous
of
t
Ty
of
IT l t'ed
So f satisfies a
Lipschitz
condition
B y theorem 5.6
well
the
Ivp is
posed
continuous
Lipschitz
For
example
Fey ysincty
I I
sinfy f
1 1 cos
let then
ye z h
Encer
The
simplest numerical
method for solving IVP
It is based on a
first order
Taylor approximation
he be
N
In Matlab a b Ntl
Derivation of the
Euler method for
the
solving Ivp problem
Flt y
y Ctt L
Mat
the
We Taylor series
to approximate
expansion
the solution to the next
grid point
Yeltiti yctiltlti i.fi y'Hi
ti n fi
2 y al
ti tis
E
I E
ylti.it yftifthy'HiHhIyY
yftilthflti ylti thIyYl
first order
In Euler our
approximation
removes the reminder in
Wit With f Hi wi
i o.li N
Example
Use the Euler method
to
approximate yal
dy
Tt HY with
ylo s t h o s
Wo I
w I Ifl I
f
wz f 1 l
Il
2
I
8
1 117 if
The exact solution
is 0.36
41,1
Error Analysis
We can use Taylor's
theorem to
guantify the
error in our
approximation
For the
step
error
single
a
in our
approximation is governed
h
by Y al If M is
2
an
upper bound for
y t
I
over a b then the magnitude
of the error at each
step is bounded by h
of the one
step errors
e
thin but h'm
hlb a In
The
global error is
to h
proportional
Geometric Interpertation
f Hi wi i
Y'Hi
fl ti
4ft
Yr yet
f
ath
L
f aid
f ft wi
y
wi the approximation
for
yet
Theorem Error
analysis
of Euler's method
et f be a continuous
function that satisfies a
L on D c a 57 112
f I EM for all
and
ly
te a b and is
yet
a
the Euler
generated by
method satisfies
a
Ect
lytti w
ezh
for
y
It trivial
proof is
Denote yet
yi
fHi.wil
y
titi Wit
Yi withhlflti.fi
E y ll
lylti.li WitilE
lyi wilthIfHi yi ffti.Ui
thly hill
E l Yi Willi 1hL 112M
We need the following
Lemma
and a b satisfying
a I
s l
for i o K 1
The
ai e Ca 1 SI
proof
Ait Ects Its ai 1T
t E Its a
t t Its thts I t
fits 14
geometric series
9
es Hst e ZHS
E N
I Es
I
Yi ti Wit l E
111hL
di f y W
h 214
27L
htt
2h22
III
ekiti att
Weakness is that the
bound of the second
derivative must be
known If
Itis fct yet Tf
y
Y HI II It yet
t
Efft.yHD fit
yet
yfti.ir ylti
hflti
yilthIy II
neglected
we define the truncation
error at each step
Titi h
yfti.it thftiyHi
h
YHiti y Iti
fftiiylt.tl
h
y ll where
2 7 ecti.ti.ie
If is bounded
M
y q
by
Ii 1,911 E had
och
with bigger
k
01h1
method of order
Taylor
in och
In order
higher Taylor
method we use more terns
than just the linear term
Given Ivp problem
Itis fit
y L
yet
aeteb
yea
with Euler II f
ti
yfti.it yCti thy
t
Ey ntl
t tilt
thy Iti
y ll
y.lt fct.ylt1
y f ft 4111
f ft Nfl
y
I
f It
Yffiti y Hilth yall
Example
t
fit t t
yet y
Find It to
Corresponding
method of order
Taylor
2
Solution
f Ct Ntl it
y
Lt
y t 11
Then
E Wi t.hl fffi.lv
hz f Iti Wi
Wi t 11 th wi th 24
wi t'tl It E HE let
The local truncation
error
kit h Yi 11
Yi hollti.fi
2
I ol Iti Yi
n
01h
Example
If
th
y
4101 1
with Euler and h 0.5
will 2.25
of degree
with
Taylor
2 we
2.64
get Wal 2.69
4111
with of
Taylor degree 2
better
we performed much
Definition
A numerical method is
Lion maxtichllio
Definition A numerical
method is convergent if the
error gets smaller as h to
iwi Tv lekly
i
Il
0,1 n
Futions
Iviul
tu
toil
Eiti
whenever heh the 1B al
D Ht.mn aeotEIeY.e
Then the method is
stable
Lik
stability is
which
hell pose dress is
to the numerical
applied
method we use
Example
dy on
Euler method
Using
wit With f 100W 2
lo h
Will ooh Wi I
i
N
was fl Cooh
For h Too 3
Was Co S
If h to then
was f S
he fo then
Then if
the method is unstable
and he tooo is stable