Section 4.3

Space of Integrable Functions

[29/100%]
Show
LaTeX
Problem 4.3.1

Let XX be a positive measure space of total measure 11. Show that for any [0,∞)[0,\infty )-valued measurable function ff on XX,

I(p):=(∫Xfp)1/p I(p) := \left(\int _X f^p\right)^{1/p}

is a non-decreasing function of p∈(1,∞)p \in (1,\infty ). Under what circumstances is I(p)I(p) strictly increasing as a function of pp?

?
Problem 4.3.2

Let (X,M,μ)(X, \mathcal{M}, \mu ) be a fixed measure space, let {pi}i=1n\{ p_i\}_{i=1}^{n} be positive numbers such that pi>1p_i > 1, i=1,…,ni = 1, \ldots , n, p>1p > 1 and

∑i=1n1pi=1p. \sum _{i=1}^{n} \frac{1}{p_i} = \frac{1}{p}.

If, for i∈{1,…,n}i \in \{ 1, \ldots , n\}, fi∈Lpi(μ)f_i \in L^{p_i}(\mu ), must it be the case that

f1f2⋯fn∈Lp(μ)? f_1 f_2 \cdots f_n \in L^p(\mu )?

Justify.

?
Problem 4.3.3

Let X={a,b,c}X = \{ a,b,c\}, let M\mathcal{M} be the σ\sigma-algebra of all subsets of XX, and let μ\mu be the measure determined by μ({a})=0\mu (\{ a\} ) = 0, μ({b})=1\mu (\{ b\} ) = 1 and μ({c})=∞\mu (\{ c\} ) = \infty. What are the dimensions of the vector spaces L1(μ)L^1(\mu ), L2(μ)L^2(\mu ), and L∞(μ)L^{\infty }(\mu )? Justify your answers.

?
Problem 4.3.4

Let f,g∈L2([0,1])f, g \in L^2([0,1]),

∫01f dm=0. \int _0^1 f \, dm = 0.

Show that

(∫01fg dm)2≤(∫01f2 dm)(∫01g2 dm−(∫01g dm)2). \left(\int _0^1 fg \, dm\right)^2 \leq \left(\int _0^1 f^2 \, dm\right) \left(\int _0^1 g^2 \, dm - \left(\int _0^1 g \, dm\right)^2\right).
?
Problem 4.3.5

Let ff be a non-negative function in Lp(R)L^p(\mathbb {R}) for some 1≤p<∞1 \leq p < \infty and let r+s=pr+s = p, r>0r > 0, s>0s > 0. Also let fh(x)=f(x+h)f_h(x) = f(x+h).

?
(a)

Show that fhrfs∈L1(R)f_h^r f^s \in L^1(\mathbb {R}).

(b)

Investigate what happens to ∥fhrfs∥1\left\| f_h^r f^s\right\|_{1} as ∣h∣→∞\left|h\right| \to \infty.

Problem 4.3.6

Let f:R→R+f : \mathbb {R} \to \mathbb {R}_+ be measurable, and let 0<r<∞0 < r < \infty. Show that for any interval II,

11∣I∣∫If≤(1∣I∣∫I1fr)1/r. \frac{1}{\frac{1}{\left|I\right|} \int _I f} \leq \left(\frac{1}{\left|I\right|} \int _I \frac{1}{f^r}\right)^{1/r}.
?
Problem 4.3.7

Let ff be a bounded measurable function on (0,1)(0,1). Prove that

lim⁡p→∞(∫01∣f(x)∣p dx)1/p=∥f∥∞, \lim _{p \to \infty } \left(\int _0^1 \left|f(x)\right|^p \, dx\right)^{1/p} = \left\| f\right\| _{\infty },

where

∥f∥∞=inf⁡{t∣m({∣f∣>t})=0}. \left\| f\right\| _{\infty } = \inf \{ t \mid m(\{ \left|f\right| > t\} ) = 0\} .
?
Problem 4.3.8

Let ff be a non-negative measurable function on [0,1][0,1] satisfying

m({x∣f(x)>t})≤11+t2,t>0. m(\{ x \mid f(x) > t\} ) \leq \frac{1}{1+t^2}, \qquad t > 0.

Determine those values of pp, 1≤p<∞1 \leq p < \infty, for which f∈Lpf \in L^p and find the minimum value of pp for which ff may fail to be in LpL^p.

?
Problem 4.3.9

With Lebesgue measure on [0,1][0,1], prove that

S={f∈C[0,1]∣∥f∥∞≤1} S = \{ f \in C[0,1] \mid \left\| f\right\| _{\infty } \leq 1\}

is not compact in L1[0,1]L^1[0,1].

?
Problem 4.3.10

Let H\mathcal{H} be Hilbert space L2(0,2π)L^2(0,2\pi ), with inner product defined by

(u,v)=∫02πu(x)v(x) dx,u,v∈H. (u,v) = \int _0^{2\pi } u(x)v(x) \, dx, \qquad u, v \in \mathcal{H}.

Consider the elements un∈Hu_n \in \mathcal{H}, n=1,2,…n = 1, 2, \ldots, defined by un(x)=sin⁡(nx)u_n(x) = \sin (nx) for x∈(0,2π)x \in (0,2\pi ). Show that

?
(a)

the set (un)n=1∞(u_n)_{n=1}^{\infty } is closed and bounded, but not compact, in the strong (i.e., norm) topology of H\mathcal{H}.

(b)

un→0u_n \to 0 as n→∞n \to \infty in the weak topology of H\mathcal{H}, i.e., for every v∈Hv \in \mathcal{H},

lim⁡n→∞(un,v)=0. \lim _{n \to \infty } (u_n, v) = 0.
Problem 4.3.11

Let H1H^1 be the Sobolev space on the unit interval [0,1][0,1], i.e., the Hilbert space consisting of functions f∈L2[0,1]f \in L^2[0,1] such that

∥f∥H12:=∑n=−∞∞(1+n2)∣f^(n)∣2<∞, \left\| f\right\| _{H^1}^2 := \sum _{n=-\infty }^{\infty } (1+n^2) \left|\hat{f}(n)\right|^2 < \infty ,

where

f^(n)=12π∫01f(x)exp⁡(−2πinx) dx \hat{f}(n) = \frac{1}{2\pi } \int _0^1 f(x) \exp (-2\pi i n x) \, dx

are Fourier coefficients of ff. Show that there exists constant C>0C > 0 such that ∥f∥∞≤C∥f∥H1\left\| f\right\|_{\infty } \leq C\left\| f\right\|_{H^1}.

?
Problem 4.3.12

Let ff be a periodic function on R\mathbb {R} with period 2π2\pi such that f∣[0,2π]f|_{[0,2\pi ]} belongs to L2(0,2π)L^2(0,2\pi ). Suppose

f(x)=∑n=−∞+∞aneinx. f(x) = \sum _{n=-\infty }^{+\infty } a_n e^{inx}.

For each h∈Rh \in \mathbb {R} define the function fhf_h by fh(x)=f(x−h)f_h(x) = f(x-h).

?
(a)

Give the Fourier expansion of fhf_h.

(b)

Find the L2L^2-norm ∥fh−f∥2\left\| f_h - f\right\|_{2} in terms of the ana_n and hh.

(c)

Prove that

lim⁡h→0∥fh−f∥2h=∞ \lim _{h \to 0} \frac{\left\| f_h - f\right\| _{2}}{h} = \infty

unless ff is constant almost everywhere.

Problem 4.3.13

Let fn(x)f_n(x) be an orthonormal family of functions in the Hilbert space L2(0,1)L^2(0,1). Prove that

∑n∣∫0xfn(t) dt∣2≤x \sum _n \left| \int _0^x f_n(t) \, dt \right|^2 \leq x

for all x∈[0,1]x \in [0,1], and that this inequality is sharp (equality) if and only if {fn∣n=1,…}\{ f_n \mid n = 1, \ldots \} span a dense subspace of L2(0,1)L^2(0,1).

?
Problem 4.3.14

If ff is a function on R\mathbb {R}, let ftf_t be the translate ft(x)=f(t+x)f_t(x) = f(t+x). Prove that if ff is square integrable with respect to Lebesgue measure, then

lim⁡t→0∥ft−f∥L2(R)=0. \lim _{t \to 0} \left\| f_t - f\right\| _{L^2(\mathbb {R})} = 0.
?
Problem 4.3.15

A trigonometric polynomial is a function pp on R\mathbb {R} of the form

p(θ)=∑n=−NNcneinθ, p(\theta ) = \sum _{n=-N}^{N} c_n e^{in\theta },

for some cn∈Cc_n \in \mathbb {C}, N≥0N \geq 0. Suppose ff is continuous and 2π2\pi-periodic on R\mathbb {R}, and let

an=∫−ππf(θ)e−inθ dθ. a_n = \int _{-\pi }^{\pi } f(\theta ) e^{-in\theta } \, d\theta .

Show that for every ε>0\varepsilon > 0, there exists a trigonometric polynomial

p(θ)=∑n=−NNcneinθ p(\theta ) = \sum _{n=-N}^{N} c_n e^{in\theta }

such that

∥f−p∥∞<ε \left\| f-p\right\| _{\infty } < \varepsilon

and

∣cn∣≤∣an∣for all n. \left|c_n\right| \leq \left|a_n\right| \qquad \text{for all } n.
?
Problem 4.3.16

Prove or disprove the following statements:

?
(a)

the set of continuous functions on the interval [0,1][0,1] is dense in L∞([0,1])L^{\infty }([0,1]).

(b)

L∞([0,1])L^{\infty }([0,1]) is a separable metric space.

Problem 4.3.17

Let {fn}n=1∞\{ f_n\}_{n=1}^{\infty } be a sequence in Lp([0,1])L^p([0,1]) (with Lebesgue measure) and {gn}n=1∞\{ g_n\}_{n=1}^{\infty } a sequence in Lq([0,1])L^q([0,1]), where 1p+1q=1\frac{1}{p}+\frac{1}{q}=1. If lim⁡n→∞fn=f\lim_{n \to \infty } f_n = f in LpL^p and lim⁡n→∞gn=g\lim_{n \to \infty } g_n = g in LqL^q, is it true that fngn→fgf_n g_n \to fg in measure? Justify.

?
Problem 4.3.18

Let XX be a measure space with measure μ\mu and suppose that μ(X)<∞\mu (X) < \infty. Let

S={(equivalence class) of measurable complex functions on X}. S = \{ (\text{equivalence class}) \text{ of measurable complex functions on } X\} .

(Here, as usual, two measurable complex functions are equivalent if they agree a.e.) For f∈Sf \in S, define

ρ(f)=∫X∣f∣1+∣f∣ dμ. \rho (f) = \int _X \frac{\left|f\right|}{1+\left|f\right|} \, d\mu .

Show that d(f,g)=ρ(f−g)d(f,g) = \rho (f-g) is a metric on SS, and that fn→ff_n \to f in this metric if and only if fn→ff_n \to f in measure.

?
Problem 4.3.19

Let gg be a measurable function such that ∫R∣fg∣<∞\int_{\mathbb {R}} \left|fg\right| < \infty for every f∈Lp(R)f \in L^p(\mathbb {R}) (fixed p≥1p \geq 1). Prove that there is a constant MM such that

∫R∣fg∣≤M∥f∥p \int _{\mathbb {R}} \left|fg\right| \leq M \left\| f\right\| _{p}

for all f∈Lp(R)f \in L^p(\mathbb {R}).

?
Problem 4.3.20

Let S1S^1 denote the unit circle (the set of complex numbers with modulus one, or the real numbers modulo 2π2\pi). The convolution of two functions on S1S^1 is

f∗g(θ)=12π∫02πf(ϑ)g(θ−ϑ) dϑ. f * g(\theta ) = \frac{1}{2\pi } \int _0^{2\pi } f(\vartheta ) g(\theta - \vartheta ) \, d\vartheta .

Suppose that ff is an element of L2(S1)L^2(S^1) with the property that its Fourier coefficient

f^(n)=∫02πf(θ)e−inθ dθ \hat{f}(n) = \int _0^{2\pi } f(\theta ) e^{-in\theta } \, d\theta

is non-zero for all n∈Zn \in \mathbb {Z}. Show that the linear space {f∗k∣k∈L2(S1)}\{ f * k \mid k \in L^2(S^1)\} is dense in L2(S1)L^2(S^1).

?
Problem 4.3.21

For functions f,g∈L2(R)f, g \in L^2(\mathbb {R}), define the convolution f∗gf*g by

f∗g(x)=∫−∞+∞f(y)g(x−y) dy, f*g(x) = \int _{-\infty }^{+\infty } f(y) g(x-y) \, dy,

where the integral is with respect to Lebesgue measure.

?
(a)

Show that f∗g∈L∞(R)f*g \in L^{\infty }(\mathbb {R}). Do not neglect to check the measurability of f∗gf*g.

(b)

Suppose that ff has the property that for all g∈L2(R)g \in L^2(\mathbb {R}) the convolution f∗gf*g is also an element of L2(R)L^2(\mathbb {R}). Define Tf:L2(R)→L2(R)T_f : L^2(\mathbb {R}) \to L^2(\mathbb {R}) by Tf(g)=f∗gT_f(g) = f*g. It is evident that TfT_f is a linear operator; you need not check this. Show that TfT_f is a bounded linear operator.

Problem 4.3.22

If f∈L1(R)f \in L^1(\mathbb {R}), define

f^(ξ)=∫−∞∞f(x)eiξx dx. \hat{f}(\xi ) = \int _{-\infty }^{\infty } f(x) e^{i\xi x} \, dx.

Prove that, for any f∈L1(R)f \in L^1(\mathbb {R}), f^(ξ)→0\hat{f}(\xi ) \to 0 as ∣ξ∣→∞\left|\xi \right| \to \infty.

?
Problem 4.3.23

Let f:[0,1]→[0,∞)f : [0,1] \to [0,\infty ) be an essentially bounded function, ∥f∥∞>0\left\| f\right\|_{\infty } > 0. Show that

lim⁡n→∞∫01f(x)n+1 dx∫01f(x)n dx=∥f∥∞. \lim _{n \to \infty } \frac{\int _0^1 f(x)^{n+1} \, dx}{\int _0^1 f(x)^n \, dx} = \left\| f\right\| _{\infty }.
?
Problem 4.3.24

Let (X,M,μ)(X, \mathcal{M}, \mu ) be a measure space for which μ(X)<∞\mu (X) < \infty. Let 1<p<∞1 < p < \infty. Suppose that {fk}\{ f_k\} is a sequence in Lp(X)L^p(X) such that sup⁡k∥fk∥p<∞\sup_k \left\| f_k\right\|_{p} < \infty and lim⁡n→∞fk(x)\lim_{n \to \infty } f_k(x) exists for μ\mu-a.e.xx. Prove that

lim⁡k→∞∥fk−f∥1=0. \lim _{k \to \infty } \left\| f_k - f\right\| _{1} = 0.
?
Problem 4.3.25

Give an example of a derivative on [0,1][0,1] which is not Lebesgue integrable.

?
Problem 4.3.26

Let (X,F,μ)(X, \mathcal{F}, \mu ) be a finite measure space. Let fnf_n be a sequence of measurable functions with f1∈L1(μ)f_1 \in L^1(\mu ) and with the property that

μ({x∈X:∣fn(x)∣>λ})≤μ({x∈X:∣f1(x)∣>λ}) \mu (\{ x \in X : \left|f_n(x)\right| > \lambda \} ) \leq \mu (\{ x \in X : \left|f_1(x)\right| > \lambda \} )

for all nn and all λ>0\lambda > 0. Prove that

lim⁡n→∞1n∫X(max⁡j≤n∣fj∣) dμ=0. \lim _{n \to \infty } \frac{1}{n} \int _X \Bigl(\max _{j \leq n} \left|f_j\right|\Bigr) \, d\mu = 0.
?
Problem 4.3.27

Let g:Rn→Rg : \mathbb {R}^n \to \mathbb {R} be non-negative with ∫Rng(y) dy=1\int_{\mathbb {R}^n} g(y) \, dy = 1, let gε(x)=ε−ng(x/ε)g_{\varepsilon }(x) = \varepsilon^{-n} g(x/\varepsilon ) for ε>0\varepsilon > 0, and define

(gε∗f)(x)=∫Rngε(y)f(x−y) dy. (g_{\varepsilon } * f)(x) = \int _{\mathbb {R}^n} g_{\varepsilon }(y) f(x-y) \, dy.

If f:Rn→Rf : \mathbb {R}^n \to \mathbb {R} is continuous and bounded, then show that fε→ff_{\varepsilon } \to f as ε→0\varepsilon \to 0 uniformly on compact subsets of Rn\mathbb {R}^n.

?
Problem 4.3.28

Fix p≥1p \geq 1. Assume without proof that the set of continuous functions with compact support is dense in Lp(Rn)L^p(\mathbb {R}^n). Let f∈Lp(Rn)f \in L^p(\mathbb {R}^n) and, for t∈Rnt \in \mathbb {R}^n, let ft(x)=f(x+t)f_t(x) = f(x+t) be the translate of ff by tt. Show

?
(a)

the map t↦ftt \mapsto f_t is continuous from Rn\mathbb {R}^n to Lp(Rn)L^p(\mathbb {R}^n), and

(b)

gε∗f→fg_{\varepsilon } * f \to f in LpL^p as ε→0\varepsilon \to 0.

Problem 4.3.29

This problem concerns Lebesgue integration of functions from R\mathbb {R} to R\mathbb {R}. For such functions, a trivial form of Holder's inequality says that if f∈L∞(R)f \in L^{\infty }(\mathbb {R}) and g∈L1(R)g \in L^1(\mathbb {R}) then

∫Rfg≤∥f∥∞∥g∥1.(*) \int _{\mathbb {R}} fg \leq \left\| f\right\| _{\infty } \left\| g\right\| _{1}. \tag {*}
?
(a)

Prove that for each g∈L1(R)g \in L^1(\mathbb {R}) there is an f∈L∞(R)f \in L^{\infty }(\mathbb {R}) such that ∥f∥∞=1\left\| f\right\|_{\infty } = 1 such that equality holds in (*).

(b)

Prove that for each f∈L∞(R)f \in L^{\infty }(\mathbb {R}) and each ε>0\varepsilon > 0 there is a g∈L1(R)g \in L^1(\mathbb {R}) with ∥g∥1=1\left\| g\right\|_{1} = 1 such that

∫Rfg>∥f∥∞−ε. \int _{\mathbb {R}} fg > \left\| f\right\| _{\infty } - \varepsilon .