Analytic approximation for nonlinear problems

The Case of Blasius Similarity Solution

Saleh Tanveer
(Ohio State University)
Collaborator: O. Costin
Background
Analytic solution for nonlinear problems, including ones arising
in vortex dynamics are very desirable.
Unfortunately, this can be only achieved for only a special class
of problems
When small asymptotic parameters are available; we can make
some headway if the limiting problem is solvable, even rigorously.
N[u; ] = 0 , suppose N[u
0
; 0] = 0
Dene u = u
0
+ E, then L[E] = N
1
[E], where
= N[u
0
; ], L :=
N
u
[u
0
; ], N
1
[E] := N[u
0
+ E] L[E]
Inversion of L and contraction leads to E = O() estimates
New non-perturbative methods
Recently, in some problems with no perturbation parameter,
Costin, Huang, Schlag [2012] and Costin, Huang & T [2012]
determined u
0
for which N[u
0
] is small. Proof of error estimates
E = u u
0
found by inversion of L in L[E] = N
1
[E]
The idea is simple and natural. Use empirical data to construct
orthogonal polynomial approximation in a nite part of domain;
for the rest use far-eld asymptotics; matching gives a composite
u
0
. The method can in principle be applied to any nonlinear
problem that allows convenient theoretical estimates of L
1
We illustrate this for the classical Blasius Similarity solution:
f

+ ff

= 0 , with f(0) = 0 = f

(0) = 0 , lim
x
f

(x) = 1 ,
Blasius equation: Background
Blasius (1912) introduced this as similarity solution for Prandtl
Boundary layer equation for ow past a at plate
Much work since then. Topfer (1912) introduced transformation
f(x) = a
1/2
F(xa
1/2
) that makes this equivalent to
F

+ FF

= 0 F(0) = 0 = F

(0) , F

(0) = 1 ,
provided lim
x
F

(x) = a > 0 exists. Weyl (1942) proved

existence and uniqueness. Series representation using
hodograph transformations (Calleghari-Friedman, 1968) have
been applied, but representation not very convenient for nding
f. Liao (99) obtained an emprically accurate solution through a
formal process. No rigorous control on the error available for
accurate and efcient approximation.
Denitions
Dene P(y) =
12

j=0
2
5(j + 2)(j + 3)(j + 4)
p
j
y
j
(1)
where [p
0
, ..., p
12
] are given by

510
10445149
,
18523
5934
,
42998
441819
,
113448
81151
,
65173
22093
,
390101
6016
,
2326169
9858
,
4134879
7249
,
1928001
1960
,
20880183
19117
,
1572554
2161
,
1546782
5833
,
1315241
32239

(2)
Dene t(x) =
a
2
(x + b/a)
2
, I
0
(t) = 1

te
t
erfc(

t)
J
0
(t) = 1

2te
2t
erfc(

2t),
q
0
(t) = 2c

te
t
I
0
+ c
2
e
2t
_
2J
0
I
0
I
2
0
_
, (3)
Main Theorem
Theorem: Let F
0
be dened by
F
0
(x) =
_

_
x
2
2
+ x
4
P
_
2
5
x
_
for x [0,
5
2
]
ax + b +
_
a
2t(x)
q
0
(t(x)) for x >
5
2
(4)
Then, there is a unique triple (a, b, c) in S close to
(a
0
, b
0
, c
0
) =
_
3221
1946
,
2763
1765
,
377
1613
_
S =
_
(a, b, c) :
_
(a a
0
)
2
+
1
4
(b b
0
)
2
+
1
4
(c c
0
)
2
5 10
5
_
(5)
with the property that F F
0
in the sense
F(x) = F
0
(x) + E(x) , (6)
Rigorous Error bounds for approximation
E

3.510
6
, E

4.510
6
, E

410
6
on [0,
5
2
]
(7)
and for x >
5
2
,

1.69 10
5
t
2
e
3t
,

d
dx
E

9.20 10
5
t
3/2
e
3t

d
2
dx
2
E

5.02 10
4
t
1
e
3t
(8)
Remark: These rigorous bounds are likely overestimates since
comparison with numerical solutions give at least ten times
smaller errors.
F
N
F
0
, F

N
F

0
, F

N
F

0
for x >
5
2
O O
x
3 4 5 6 7 8 9 10
0
2. #10
- 7
4. #10
- 7
6. #10
- 7
8. #10
- 7
F
N
F
0
, F

N
F

0
, F

N
F

0
for x [0,
5
2
]
O O
x
0.5 1 1.5 2 2.5
K5. #10
- 7
K4. #10
- 7
K3. #10
- 7
K2. #10
- 7
K1. #10
- 7
0
1. #10
- 7
Guess F
0
Guess F
0
in
_
0,
5
2

found by using empirical data for F

= FF

on [0,
5
2
] and projecting it to a low order Chebyshev polynomial.
Guess for F
0
in
_
5
2
,
_
found from asymptotics for large x,
rigorously controlled, that includes upto O
_
exp
_
a
_
x +
b
a
_
2
__
contribution exactly, for each a > 0, b and c.
Parameters (a, b, c) determined from continuity of solution
representation of F at x =
5
2
, casting it as a contraction map in R
3
Once guess F
0
is found so that residual R = F

0
+ F
0
F

0
is
uniformly small, rest is a mathematical analysis of E, where
F = F
0
+ E, helped by the smallness of R, which allows a
contraction mapping argument.
Conclusion
We have shown how an accurate and efcient analytical
approximation of Blasius solution with rigorous error bounds is
possible.
The method is general and simply relies on nding a candidate F
0
so that N[F
0
] is small.
Candidate F
0
on nite domain I = [0,
5
2
] obtained through
orthogonal projection of empirical solution using a small number
of Chebyshev polynomials. More accuracy requires more modes.
Outside I rigorous exponential asymptotics ideas was used to
come up with highly accurate F F
0
approximation
The ideas equally applicable to problems with parameters or to
class of nonlinear PDEs
Full analysis available in
http://www.math.ohio-state.edu/tanveer