Scribd
Upload
66 views

∞ s−1 −x ε→0 ε>0 1/ε s−1 −x

This document contains solutions to problems involving properties of the gamma function Γ(s), Gronwall's inequality, and an integral identity involving partial derivatives. Specifically: 1) It shows that the integral defining Γ(s) converges when s > 0, and that Γ(s+1) = sΓ(s) and Γ(1) = 1. 2) It proves Gronwall's inequality, which bounds a function u(x) in terms of an integral involving another function β(x). 3) It proves an identity relating the partial derivative Ft(x,t) to an integral involving F(x,t).

Uploaded by

JJ
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
66 views

∞ s−1 −x ε→0 ε>0 1/ε s−1 −x

This document contains solutions to problems involving properties of the gamma function Γ(s), Gronwall's inequality, and an integral identity involving partial derivatives. Specifically: 1) It shows that the integral defining Γ(s) converges when s > 0, and that Γ(s+1) = sΓ(s) and Γ(1) = 1. 2) It proves Gronwall's inequality, which bounds a function u(x) in terms of an integral involving another function β(x). 3) It proves an identity relating the partial derivative Ft(x,t) to an integral involving F(x,t).

Uploaded by

JJ
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

M361K (56225) Problem Set 4 Solutions 1.

The gamma function is dened for s > 0 by the improper integral


1/

(s) =
0

xs1 ex dx = lim

0 >0

xs1 ex dx.

(a) Show that the integral in the denition (s) converges when s > 0. Solution: The integrand xs1 ex is nonnegative for x 0. Hence the inte1/ gral xs1 ex increases as 0 (once < 1). Thus it is sucient to show that these integrals are bounded above. First lets consider the case where s > 1. The function log x is concave (i.e., log x is convex) with derivative 1/x, so we have log x x + log(2s 2) 1. 2s 2

The upper bound is simply the tangent line to the graph of log(x) at the point (2s 2, log(2s 2)). Thus (s 1) log x x x + (s 1)(log(2s 2) 1), 2

so xs1 ex Cex/2 where C is the positive constant (2(s 1)e1 )s1 . Thus
1/ 1/

xs1 ex dx

Cex/2 dx
0

Cex/2 dx = 2C.

Now lets consider the case where s 1. If 0 x 1, then xs1 ex xs1 , while if x > 1, then xs1 1, so xs1 ex ex . Hence
1/ 1

s1 x

dx
0

s1

dx +
1

ex dx = s1 + e1 .

(b) Show that (s + 1) = s(s) and (1) = 1. Solution: Using integration by parts, we see that
1/

(s + 1) = lim

xs ex dx
1/ xs ex 1/

= lim

+s

xs1 ex dx

= s(s).

Problem Set 4 Solutions

When s = 1, we have (1) =


0

ex dx = 1.

(c) Show that the function (s) = log (s) is convex. [Hint: use H olders inequality.] Solution: Suppose 1/p + 1/q = 1 where p, q > 1. Let s, t > 0. Then

(s/p + t/q ) =
0

xs/p+t/q1 ex dx (xs/p1/p ex/p )(et/q1/q ex/q ) dx


0 1/p 1/q

xs1 ex
0 0

xt1 ex

by H olders inequality. Applying log, we see that (s/p + t/q ) (s)/p + (t)/q . 2. Let a < b. Suppose u : [a, b) R and : [a, b) R are continuous functions and that u is dierentiable on (a, b). Prove Gr onwalls inequality: show that if u satises the inequality u (x) (x)u(x) for all x (a, b), then
x

u(x) u(a) exp


a

(t) dt

for all x [a, b). [Hint: dierentiate the function u(t)/g (t) where g is a solution of the dierential equation g (t) = (t)g (t).] Solution: Let g (x) = exp
a

(t) dt .

Observe that g (x) > 0 for all x [a, b) and g (a) = 1. By the chain rule and the fundamental theorem of calculus, we have g (x) = (x)g (x). Let f (x) = u(x)/g (x). Then u (x)g (x) u(x)g (x) u (x) u(x) (x) f (x) = = 0 2 (g (x)) g (x) for all x (a, b). By the mean value theorem, if x > a, then f (x) f (a), so u(x) u(a)g (x). 3. Let F : [a, b] [a , b ] R be a continuous function continuously dierentiable in the second variable. Write F F (x, t + h) F (x, t) Ft (x, t) = = lim . t (x,t) h0 h

Problem Set 4 Solutions

Suppose that Ft : [a, b] (a , b ) R extends to a continuous function [a, b] [a , b ] R. Prove that b b d F (x, t) dx = Ft (x, s) dx dt t=s a a for s [a , b ]. [Hint: use the Mean Value Theorem and the uniform continuity of Ft .] Solution: Let > 0. There exists > 0 such that |t0 t1 | < implies (b a) Ft (x, t1 ) Ft (x, t0 ) < by the uniform continuity of Ft (recall that [a, b] [a , b ] is compact and continuous functions on compact sets are uniformly compact). Suppose t0 [a , b ] and |h| < so that t0 + h [a , b ] as well. If x [a, b], then F (x, t0 + h) F (x, t0 ) = Ft (x, s)h for some s between t0 and t0 + h by the mean value theorem, so ( b a) This implies that
b a

F (x, t0 + h) F (x, t0 ) Ft (x, t0 ) < . h

F (x, t0 + h) F (x, t0 ) Ft (x, t0 ) dx h b F (x, t0 + h) F (x, t0 ) Ft (x, t0 ) dx h a


b

Hence
h0

lim

F (x, t0 + h) F (x, t0 ) dx = h

Ft (x, t0 ).
a

You might also like