Sunteți pe pagina 1din 3

Asymptotic Methods: Example Sheet 1

1
Send corrections to David Stuart dmas2@cam.ac.uk
Copyright © 2017 University of Cambridge. Not to be quoted or reproduced without permission.

Part IID
Lent 2017
David Stuart dmas2@cam.ac.uk

1. Suppose that the functions f and g have the asymptotic expansions



X ∞
X
−n
f (z) ∼ an z , and g(z) ∼ bn z −n
n=0 n=0

as z −→ ∞ . Show that

X
f (z) g(z) ∼ cn z −n ,
n=0
Pn
as z −→ ∞ , where cn = k=0 an−k bk .
an xn as
P∞
2. (a) Show that if a function admits as asymptotic expansion f (x) ∼ n=0
x → 0+ , then the an are determined uniquely by f .
(b) Consider the function 
e(x) = exp − 1/x
for x > 0 . Show that, in an asymptotic expansion of the form

e(x) ∼ β0 + β1 x + β2 x2 + . . . ,

valid as x −→ 0+ , all the coefficients β0 , β1 , β2 , . . . are zero. Deduce that in


(a) the coefficients {an } do not determine f uniquely.
3. (a) Taking δ to be a positive constant, show that as z −→ ∞ in the complex plane
(not necessarily along a ray)
1 z
cosh(z) ∼ e
2
in the sector − π2 + δ < arg z < π2 − δ and
 

1 −z
cosh(z) ∼ e
2
π 3π
 
in the sector 2
+ δ < arg z < 2
− δ . Is this still true if δ = 0?
(b) Find asymptotic expansions for tanh z as z → ∞ in the complex plane, stating
in which sectors they hold and specifying the Stokes lines.
Copyright © 2017 University of Cambridge. Not to be quoted or reproduced without permission.

4. (a) Show that the Stieltjes integral



ρ(t)
Z
F (x) = dt
0 1 + xt
n n +
P
R n the asymptotic expansion F (x) ∼ (−1) an x , (x → 0 ), where an =
admits
t ρ(t) dt, under the assumption that the continuous function ρ satisfies ρ(t) ≤
R ∞ xe−t
Ce−ǫt for some positive C, ǫ and all t ≥ 0 . Deduce that F (x) = 0 1+xt dt admits
n n+1 +
P∞
the expansion F (x) ∼ n=0 (−1) n! x as x → 0 . Show similarly that
Z ∞ ∞
e−t X
G(x) = 2 dt ∼ (−1)n (n + 1)! xn , (x → 0+ ) .
0 1 + xt n=0

(b) Differentiating through the integral show that F ′ = G and comment on the re-
lation between the two asymptotic series you just obtained. Give an example of a
smooth functionP H : (0, ∞) → (0, ∞) with the property that H admits an asymp-
totic expansion αn xn as x → 0+ , but term-by-term differentation does not give
an asymptotic expansion for H ′ . Show however, that Pif in this situation H ′ is continu-
ous on [0, ∞) and admits an asymptotic expansion βn x as x → 0+ , then neces-
n

sarily this expansion is given by term-by-term differentiation, i.e. βn = (n + 1)αn+1 .


(c) For a given small positive value of x , find the value(s) of n giving the term(s)
of smallest magnitude in the asymptotic expansion for G . Hence, use optimal trun-
cation to obtain an estimate of the ‘exact’ value G(0.1) = 0.84366660602... [By
convention optimal truncation of an asymptotic expansion means keeping all terms
in the expansion up to the one BEFORE the smallest.]
(d) For the case ρ(t) = e−t recall from lectures that F (x) = N n n
P
n=0 (−1) n!x + ErrN
with error bound |ErrN | ≤ (N + 1)!xN +1 . Using this to define optimal truncation
by N + 1 = [x−1 ], the integer part of x−1 , use Stirling’s formula to show that the
1
resulting “optimal error bound” is O [x−1 ] 2 exp(−[x−1 ]) = o(xM ) , as x → 0+ for
every positive integer M .
5. (a) Use integration by parts to find an asymptotic expansion, valid as x −→ ∞ , for
the exponential integral
Z ∞ −t
e 
E1 (x) = dt ∼ e−x b1 x−1 + b2 x−2 + b3 x−3 + . . . ,
x t

2

for suitable constants b1 , b2 , b3 , . . . . Show that the remainder is O e−x x−N −1
as x −→ ∞ , for suitable N .
(b) Check your answer by making the substitution t = x (1 + s) in the integral and
applying Watson’s Lemma.
d
(c) Obtain an asymptotic expansion of E1 (x) as x → 0+ by considering dx
(E1 (x) +
ln x) and integrating.
6. Find asymptotic expansions as x → ∞ of
Z 1 Z ∞
− xt(1 − t)2 2
I1 (x) = e dt and I2 (x) = e− xt(1 − t) dt ,
0 0
Copyright © 2017 University of Cambridge. Not to be quoted or reproduced without permission.


giving all terms up to and including O x −1
.
7. By means of Laplace’s method, show that the first two terms in an asymptotic expan-
sion as x −→ ∞ of
Z π
2
I(x) = exp(− x t3 cos t) dt
0

are given by
   
1 1 1 8 1
I(x) ∼ Γ + + 3 + ... .
3 x1/3 3 6 π x

*Find the next term in the expansion.


8. Show that
π2
√ i
 
2 2
Z
4
h
x
exp x cos t dt ∼ e + + ...
0 x 3x2

as x → ∞ and obtain the corresponding asymptotic expansion when the upper


limit is replaced by 4π 2 .