Section VIII.3

Summable Functions

[47/87%]
Show
LaTeX
Problem VIII.3.1

Let the union of any three sets of the family E1,…,ENE_{1}, \ldots , E_{N} coincide with the interval [0,1]\left[0,1\right]. Set

S1=∑1⩽k⩽Nλ(Ek),S2=∑1⩽j<k⩽Nλ(Ej∩Ek). S_{1}=\sum _{1 \leqslant k \leqslant N} \lambda \left(E_{k}\right), \quad S_{2}=\sum _{1 \leqslant j<k \leqslant N} \lambda \left(E_{j} \cap E_{k}\right) .

Express in terms of S1S_{1} and S2S_{2} the integral Y=∫01(∑k=1NχEk(x))3dx\mathcal{Y}=\int_{0}^{1}\left(\sum_{k=1}^{N} \chi_{E_{k}}(x)\right)^{3} d x.

?
Problem VIII.3.2

Let each point of the interval [0,1]\left[0,1\right] belong to at least kk of the sets E1,…,ENE_{1}, \ldots , E_{N}. Prove that for some nn the inequality λ(En)⩾kN\lambda \left(E_{n}\right) \geqslant \frac{k}{N} holds.

?
Problem VIII.3.3

Let E⊂Rm,λm(E)<∞,f∈L0(E)E \subset \mathbb {R}^{m}, \lambda_{m}(E)<\infty , f \in \mathscr {L}^{0}(E). Prove the equivalence of the following conditions:

  1. f∈L(E)f \in \mathscr {L}(E);

  2. ∑k⩾1λm(E(∣f∣⩾k))<∞\sum_{k \geqslant 1} \lambda_{m}(E(\left|f\right| \geqslant k))<\infty;

  3. ∑k⩾1kλm(E(k⩽∣f∣<k+1))<∞\sum_{k \geqslant 1} k \lambda_{m}(E(k \leqslant \left|f\right|<k+1))<\infty.

?
Problem VIII.3.4

Let the sets Ek⊂Rm(k∈N)E_{k} \subset \mathbb {R}^{m}(k \in \mathbb {N}) satisfy the condition ∑k⩾1λm(Ek)<∞\sum_{k \geqslant 1} \lambda_{m}\left(E_{k}\right)<\infty, and let

An={x∈Rm∣x∈Ek for exactly n values of k},Bn={x∈Rm∣x∈Ek for at least n values of k}. \begin{aligned} & A_{n}=\left\{ x \in \mathbb {R}^{m} \mid x \in E_{k} \text{ for exactly } n \text{ values of } k\right\} , \\ & B_{n}=\left\{ x \in \mathbb {R}^{m} \mid x \in E_{k} \text{ for at least } n \text{ values of } k\right\} . \end{aligned}

Prove that the sets An,BnA_{n}, B_{n} are measurable and

λm(Bn)=∑k⩾nλm(Ak),∑n⩾1nλm(An)=∑n⩾1λm(Bn)=∑n⩾1λm(En). \begin{gathered} \lambda _{m}\left(B_{n}\right)=\sum _{k \geqslant n} \lambda _{m}\left(A_{k}\right), \\ \sum _{n \geqslant 1} n \lambda _{m}\left(A_{n}\right)=\sum _{n \geqslant 1} \lambda _{m}\left(B_{n}\right)=\sum _{n \geqslant 1} \lambda _{m}\left(E_{n}\right) . \end{gathered}
?
Problem VIII.3.5

Let xn>0,xn→0,f(t)=card⁡{n∈N∣xn>t}x_{n}>0, x_{n} \rightarrow 0, f(t)=\operatorname {card}\left\{ n \in \mathbb {N} \mid x_{n}>t\right\}. Prove that ∫0+∞f(t)dt=∑xn\int_{0}^{+\infty } f(t) d t=\sum x_{n}.

?
Problem VIII.3.6
?
(a)

Prove that the series ∑2−n∣x−rn∣−1/2\sum 2^{-n}\left|x-r_{n}\right|^{-1 / 2}, where {rn}\left\{ r_{n}\right\} is an arbitrary numerical sequence, converges almost everywhere on R\mathbb {R}.

(b)

Construct a function from L0(R)\mathscr {L}^{0}(\mathbb {R}) that is not summable on any (nonempty) interval.

Problem VIII.3.7

Can an everywhere differentiable function on [0,1]\left[0,1\right] have a derivative that is not summable on [0,1]\left[0,1\right]?

?
Problem VIII.3.8

For which p∈Rp \in \mathbb {R} are the following functions summable on the half-line (0,+∞0,+\infty)

?
(a)

sin⁡xxp\frac{\sin x}{x^{p}};

(b)

e−xpsin⁡2xe^{-x^{p} \sin^{2} x};

(c)

xp1+x6sin⁡2x?\frac{x^{p}}{1+x^{6} \sin^{2} x} ?

Problem VIII.3.9

Are the functions ff indicated below summable on the square [−1,1]×[−1,1]\left[-1,1\right] \times [-1,1]? Find the iterated integrals

∫−11(∫−11f(x,y)dy)dx and ∫−11(∫−11f(x,y)dx)dy \int _{-1}^{1}\left(\int _{-1}^{1} f(x, y) d y\right) d x \quad \text{ and } \quad \int _{-1}^{1}\left(\int _{-1}^{1} f(x, y) d x\right) d y

if:

?
(a)

f(x,y)=xyx2+y2f(x, y)=\frac{x y}{x^{2}+y^{2}};

(b)

f(x,y)=x2−y2x2+y2f(x, y)=\frac{x^{2}-y^{2}}{x^{2}+y^{2}};

(c)

f(x,y)=xy(x2+y2)2f(x, y)=\frac{x y}{\left(x^{2}+y^{2}\right)^{2}};

(d)

f(x,y)=x2−y2(x2+y2)2f(x, y)=\frac{x^{2}-y^{2}}{\left(x^{2}+y^{2}\right)^{2}} (in all cases the value f(0,0)f(0,0) may be defined arbitrarily).

Problem VIII.3.10

For which p∈Rp \in \mathbb {R} is the function f(x,y)=∣1−xy∣−pf(x, y)=\left|1-x y\right|^{-p} summable on the set EE in the following cases:

?
(a)

E=[0,1]2E=\left[0,1\right]^{2};

(b)

E=[0,2]2E=\left[0,2\right]^{2};

(c)

E={(x,y)∈[0,1]2∣(x−2)2+y2⩾2,x2+(y−2)2⩾2}E=\left\{ (x, y) \in [0,1]^{2} \mid (x-2)^{2}+y^{2} \geqslant 2, x^{2}+(y-2)^{2} \geqslant 2\right\} ?

Problem VIII.3.11
?
(a)

Let p>0,E⊂Rmp>0, E \subset \mathbb {R}^{m} and λm(E)=λm(B(0,r))\lambda_{m}(E)=\lambda_{m}(B(0, r)). Prove that

∫Edy∥x−y∥p⩽∫B(0,r)dz∥z∥p for any x∈Rm.  \int _{E} \frac{d y}{\left\| x-y\right\| ^{p}} \leqslant \int _{B(0, r)} \frac{d z}{\left\| z\right\| ^{p}} \quad \text{ for any } x \in \mathbb {R}^{m} \text{. }
(b)

Prove that

∣∫Eeixdx∣⩽2sin⁡λ(E)2 \left|\int _{E} e^{i x} d x\right| \leqslant 2 \sin \frac{\lambda (E)}{2}

for any set E⊂[0,2π]E \subset [0,2 \pi ].

(c)

Prove that

∣∬Edxdyx+iy∣⩽πλ2(E) \left|\iint _{E} \frac{d x d y}{x+i y}\right| \leqslant \sqrt{\pi \lambda _{2}(E)}

for any set EE of finite measure (E⊂R2E \subset \mathbb {R}^{2}).

Problem VIII.3.12

Let p,t>0,E⊂Rm,f∈L0(E)p, t>0, E \subset \mathbb {R}^{m}, f \in \mathscr {L}^{0}(E). Prove that

λm(E(∣f∣⩾t))⩽1tp∫E∣f(x)∣pdx \lambda _{m}(E(\left|f\right| \geqslant t)) \leqslant \frac{1}{t^{p}} \int _{E}\left|f(x)\right|^{p} d x

In particular, for p=2p=2 we obtain Chebyshev's inequality:

λm(E(∣f∣⩾t))⩽1t2∫E∣f(x)∣2dx \lambda _{m}(E(\left|f\right| \geqslant t)) \leqslant \frac{1}{t^{2}} \int _{E}\left|f(x)\right|^{2} d x
?
Problem VIII.3.13

Prove that if ∣f∣p∈L(E)(p>0,E⊂Rm)\left|f\right|^{p} \in \mathscr {L}(E)\left(p>0, E \subset \mathbb {R}^{m}\right), then

λm(E(∣f∣⩾t))=o(t−p) as t→+∞. \lambda _{m}(E(\left|f\right| \geqslant t))=o\left(t^{-p}\right) \quad \text{ as } t \rightarrow +\infty .

Give an example showing that the converse statement is false.

?
Problem VIII.3.14

Let p>0,E⊂Rm,f∈L0(E)p>0, E \subset \mathbb {R}^{m}, f \in \mathscr {L}^{0}(E) and λm(E(∣f∣⩾t))=O(t−p)\lambda_{m}(E(\left|f\right| \geqslant t))=O\left(t^{-p}\right) as t→+∞t \rightarrow +\infty. Prove that if λm(E)<∞\lambda_{m}(E)<\infty, then ∣f∣p−ε∈L(E)\left|f\right|^{p-\varepsilon } \in \mathscr {L}(E) for any ε∈(0,p)\varepsilon \in (0, p).

?
Problem VIII.3.15

Let E⊂Rm,f∈L(E)E \subset \mathbb {R}^{m}, f \in \mathscr {L}(E). Prove that

∫E∣f(x)∣pdx⟶p→+0λm(E(f≠0)). \int _{E}\left|f(x)\right|^{p} d x \underset {p \rightarrow +0}{\longrightarrow } \lambda _{m}(E(f \neq 0)) .
?
Problem VIII.3.16

Let f,g∈L0(E),f,g⩾0f, g \in \mathscr {L}^{0}(E), f, g \geqslant 0. Prove that if ∫Eg(x)dx=1\int_{E} g(x) d x=1 and ∫Eg(x)f(x)dx<∞\int_{E} g(x) f(x) d x<\infty, then

(∫Efp(x)g(x)dx)1/p⟶p→+0exp⁡(∫Eg(x)ln⁡f(x)dx) \left(\int _{E} f^{p}(x) g(x) d x\right)^{1 / p} \underset {p \rightarrow +0}{\longrightarrow } \exp \left(\int _{E} g(x) \ln f(x) d x\right)

Can the condition ∫Eg(x)f(x)dx<∞\int_{E} g(x) f(x) d x<\infty be dropped?

?
Problem VIII.3.17

Let ff be a measurable nonnegative function on R\mathbb {R}. Prove that for every ε>0\varepsilon >0 the following inequalities hold:

?
(a)

∫12(∑k∈Z,k≠0f(kεx))dx⩽1ε∫Rf(x)dx\int_{1}^{2}\left(\sum_{k \in \mathbb {Z}, k \neq 0} f(k \varepsilon x)\right) d x \leqslant \frac{1}{\varepsilon } \int_{\mathbb {R}} f(x) d x;

(b)

∫12(∑k∈Z,k≠0f(kεx))dx⩽4ε∫Rf(x)dx\int_{1}^{2}\left(\sum_{k \in \mathbb {Z}, k \neq 0} f\left(\frac{k}{\varepsilon x}\right)\right) d x \leqslant 4 \varepsilon \int_{\mathbb {R}} f(x) d x.

Problem VIII.3.18

Let ff be a summable nonnegative function on R\mathbb {R}, {(an,bn)}n∈Z\left\{ \left(a_{n}, b_{n}\right)\right\}_{n \in \mathbb {Z}} a two-sided sequence of intervals. Suppose there exists δ>0\delta >0 such that an+1⩾bn+δa_{n+1} \geqslant b_{n}+\delta for all n∈Zn \in \mathbb {Z}. Prove that then

∑n∈Z∫an+1+∞f(x)(x−bn)3/2dx<+∞ \sum _{n \in \mathbb {Z}} \int _{a_{n+1}}^{+\infty } \frac{f(x)}{\left(x-b_{n}\right)^{3 / 2}} d x<+\infty
?
Problem VIII.3.19

Let x∈R,x=[x]+∑εk2−kx \in \mathbb {R}, x=\left[x\right]+\sum \varepsilon_{k} 2^{-k} be the binary representation of the number xx.

?
(a)

Find the integrals ∫01εk(x)dx,∫01εk(x)εm(x)dx(k,m∈N,k≠m)\int_{0}^{1} \varepsilon_{k}(x) d x, \int_{0}^{1} \varepsilon_{k}(x) \varepsilon_{m}(x) d x(k, m \in \mathbb {N}, k \neq m).

(b)

Find the integral ∫01∣sn(x)−12∣2dx\int_{0}^{1}\left|s_{n}(x)-\frac{1}{2}\right|^{2} d x, where sn(x)=1n∑1⩽k⩽nεk(x)s_{n}(x)=\frac{1}{n} \sum_{1 \leqslant k \leqslant n} \varepsilon_{k}(x).

(c)

Prove that ∫01∣sn(x)−12∣4dx=O(n−2)\int_{0}^{1}\left|s_{n}(x)-\frac{1}{2}\right|^{4} d x=O\left(n^{-2}\right).

Problem VIII.3.20

Let x∈R,x=[x]+∑εk⋅10−kx \in \mathbb {R}, x=\left[x\right]+\sum \varepsilon_{k} \cdot 10^{-k} be the decimal representation of the number xx. Let jj be one of the digits 0,1,…,90,1, \ldots , 9, and let

aj(x)={1, if ε1(x)=j,0, if ε1(x)≠j. a_{j}(x) = \begin{cases} 1, & \text{ if } \varepsilon _{1}(x)=j, \\ 0, & \text{ if } \varepsilon _{1}(x) \neq j . \end{cases}

Find the integrals (k,m∈N,k≠mk, m \in \mathbb {N}, k \neq m)

?
(a)

∫01aj(10k−1x)dx\int_{0}^{1} a_{j}\left(10^{k-1} x\right) d x;

(b)

∫01aj(10k−1x)aj(10m−1x)dx\int_{0}^{1} a_{j}\left(10^{k-1} x\right) a_{j}\left(10^{m-1} x\right) d x.

Problem VIII.3.21

Let a∈R,a≠0,a≠1,x=∑k⩾1εk(x)⋅10−ka \in \mathbb {R}, a \neq 0, a \neq 1, x=\sum_{k \geqslant 1} \varepsilon_{k}(x) \cdot 10^{-k} be the decimal representation of the number x,x∈(0,1)x, x \in (0,1). Set

f(x)={a, if εn(x)≠0 for every n;1, if the first digit εn(x)≠0 has an even index; 0 otherwise.  f(x) = \begin{cases} a, & \text{ if } \varepsilon _{n}(x) \neq 0 \text{ for every } n ; \\ 1, & \text{ if the first digit } \varepsilon _{n}(x) \neq 0 \text{ has an even index; } \\ 0 & \text{ otherwise. } \end{cases}

Prove that the function ff is measurable, and find ∫01f(x)dx\int_{0}^{1} f(x) d x.

?
Problem VIII.3.22

In constructing the Cantor set from the interval [0,1]\left[0,1\right], intervals of length 13,19\frac{1}{3}, \frac{1}{9}, and so on were successively removed (see the text before problem I.1.29). Let f(x)=nf(x)=n on intervals of length 3−n3^{-n}, f(x)=0f(x)=0 on the Cantor set.

?
(a)

Prove that the function ff is measurable, and find ∫01f(x)dx\int_{0}^{1} f(x) d x.

(b)

Find the increasing distribution function FF of the function ff and the nonincreasing rearrangement f∗f^{*} of the function ff (see the definitions before problems 2.8,2.132.8, 2.13).

Problem VIII.3.23

Let KK be a closed subset of the interval [a,b]\left[a, b\right],

ρ(y)=inf⁡{∣y−t∣∣t∈K} \rho (y)=\inf \left\{ \left|y-t\right| \mid t \in K\right\}

the distance from the point y∈Ry \in \mathbb {R} to the set KK. Prove that for any s>0s>0 the function

Is(x)=∫abρs(y)dy∣x−y∣s+1(x∈K) \mathcal{I}_{s}(x)=\int _{a}^{b} \frac{\rho ^{s}(y) d y}{\left|x-y\right|^{s+1}} \quad (x \in K)

is summable on KK.

?
Problem VIII.3.24

Prove that for a nonnegative function ff the equalities hold (0<a⩽∞)(0<a \leqslant \infty )

?
(a)

∫[−a,a]mf(max⁡1⩽k⩽m∣xk∣)dx=m2m∫0atm−1f(t)dt\int_{\left[-a, a\right]^{m}} f\left(\max_{1 \leqslant k \leqslant m}\left|x_{k}\right|\right) d x=m 2^{m} \int_{0}^{a} t^{m-1} f(t) d t;

(b)

∫[−a,a]mf(x)dx=∫0atm−1(∫max⁡∣yk∣=1f(ty)dλm−1(y))dt\int_{\left[-a, a\right]^{m}} f(x) d x=\int_{0}^{a} t^{m-1}\left(\int_{\max \left|y_{k}\right|=1} f(t y) d \lambda_{m-1}(y)\right) d t, where λm−1\lambda_{m-1} is the (m−1)(m-1)-dimensional Lebesgue measure on the boundary of the cube [−1,1]m\left[-1,1\right]^{m}.

Problem VIII.3.25

Compute the integral (p>−1p>-1)

∫[0,1]m(min⁡xkmax⁡xk)pdx \int _{\left[0,1\right]^{m}}\left(\frac{\min x_{k}}{\max x_{k}}\right)^{p} d x
?
Problem VIII.3.26

Prove by induction the validity of the equalities (m≥2m \geq 2)

?
(a)

∫Sm−1f(x1,…,xm−1,xm)dμm−1(x)==∫0πsin⁡m−2θ{∫Sm−2f(u1sin⁡θ,…,um−1sin⁡θ,cos⁡θ)dμm−2(u)}dθ;\int_{S^{m-1}} f\left(x_{1}, \ldots , x_{m-1}, x_{m}\right) d \mu_{m-1}(x)= =\int_{0}^{\pi } \sin^{m-2} \theta \left\{ \int_{S^{m-2}} f\left(u_{1} \sin \theta , \ldots , u_{m-1} \sin \theta , \cos \theta \right) d \mu_{m-2}(u)\right\} d \theta ;

(b)

∫Rmf(x)dx=∫0∞tm−1{∫Sm−1f(tω)dμm−1(ω)}dt\int_{\mathbb {R}^{m}} f(x) d x=\int_{0}^{\infty } t^{m-1}\left\{ \int_{S^{m-1}} f(t \omega ) d \mu_{m-1}(\omega )\right\} d t,

where the function ff is assumed to be summable on Sm−1S^{m-1} or Rm\mathbb {R}^{m} respectively. For m=2m=2 in formula a), S0S^{0} should be understood as the boundary of the segment [−1,1]\left[-1,1\right], i.e. the two-point set {−1,1}\left\{ -1,1\right\}, and the measure μ0\mu_{0} should be taken to consist of unit loads placed at the points +1 and -1.

Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}.

Problem VIII.3.27
?
(a)

Let rr be a nonnegative continuous function defined on Sm−1,T={x∈Rm∣ 0<∥x∥⩽r(x∥x∥)}S^{m-1}, T=\left\{ x \in \mathbb {R}^{m} \left\lvert \, 0<\left\| x\right\| \leqslant r\left(\frac{x}{\left\| x\right\| }\right)\right.\right\}. Prove that

λm(T)=1m∫Sm−1rm(ω)dμm−1(ω) \lambda _{m}(T)=\frac{1}{m} \int _{S^{m-1}} r^{m}(\omega ) d \mu _{m-1}(\omega )
(b)

Let a function ff continuous on Rm\mathbb {R}^{m} satisfy the condition f(tx)=tf(x)>0f(t x)=t f(x)>0 for x≠0,t>0x \neq 0, t>0. Prove that

∫Sm−1dμm−1(x)fm(x)=mλm(V), where V={x∈Rm∣f(x)⩽1} \int _{S^{m-1}} \frac{d \mu _{m-1}(x)}{f^{m}(x)}=m \lambda _{m}(V), \quad \text{ where } \quad V=\left\{ x \in \mathbb {R}^{m} \mid f(x) \leqslant 1\right\}
Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}.

Problem VIII.3.28

Let the function ff, continuous on Rm\mathbb {R}^{m}, satisfy the condition f(tx)=tf(x)>0f(t x)=t f(x)>0 for x≠0,t>0x \neq 0, t>0, and let AA be a nondegenerate linear transformation on Rm\mathbb {R}^{m}. Prove that

∫Sm−1dμm−1(x)fm(Ax)=1∣det⁡(A)∣∫Sm−1dμm−1(x)fm(x) \int _{S^{m-1}} \frac{d \mu _{m-1}(x)}{f^{m}(A x)}=\frac{1}{\left|\operatorname {det}(A)\right|} \int _{S^{m-1}} \frac{d \mu _{m-1}(x)}{f^{m}(x)}
?
Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}.

Problem VIII.3.29

Let EE be a measurable subset of the half-line (0,+∞)(0,+\infty ), AE={x∈Rm∣∥x∥∈E}A_{E}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| \in E\right\}. Prove that

λm(AE)=mαm∫Etm−1dt \lambda _{m}\left(A_{E}\right)=m \alpha _{m} \int _{E} t^{m-1} d t
?
Remark.
?

When solving the problem it is useful to use the formula (see problem 3.26 b))

∫a⩽∥x∥⩽bf(∥x∥)dλm(x)=mαm∫abtm−1f(t)dt, \int _{a \leqslant \left\| x\right\| \leqslant b} f(\left\| x\right\| ) d \lambda _{m}(x)=m \alpha _{m} \int _{a}^{b} t^{m-1} f(t) d t,

where ff is a measurable nonnegative function on the interval ⟨a,b⟩\langle a, b\rangle (0⩽a<b⩽∞0 \leqslant a<b \leqslant \infty), αm=λm(Bm)=πm/2/Γ(m2+1)\alpha_{m}=\lambda_{m}\left(B^{m}\right)=\pi^{m / 2} / \Gamma \left(\frac{m}{2}+1\right).

Problem VIII.3.30

What is the "Gaussian volume" of the ball B(0,r)B(0, r)?

?
Remark.
?

The symbol γm\gamma_{m} denotes the standard Gaussian measure on Rm\mathbb {R}^{m}, i.e., the measure with density (2π)−m/2e−∥x∥2/2(2 \pi )^{-m / 2} e^{-\left\| x\right\|^{2} / 2}.

When solving the problem it is useful to use the formula (see problem 3.26 b))

∫a⩽∥x∥⩽bf(∥x∥)dλm(x)=mαm∫abtm−1f(t)dt, \int _{a \leqslant \left\| x\right\| \leqslant b} f(\left\| x\right\| ) d \lambda _{m}(x)=m \alpha _{m} \int _{a}^{b} t^{m-1} f(t) d t,

where ff is a measurable nonnegative function on the interval ⟨a,b⟩\langle a, b\rangle (0⩽a<b⩽∞0 \leqslant a<b \leqslant \infty), αm=λm(Bm)=πm/2/Γ(m2+1)\alpha_{m}=\lambda_{m}\left(B^{m}\right)=\pi^{m / 2} / \Gamma \left(\frac{m}{2}+1\right).

Problem VIII.3.31

Let a>0a>0, and let KaK_{a} be the cone in the space R4\mathbb {R}^{4}:

Ka={(x1,x2,x3,x4)∈R4∣x22+x32+x42⩽a2x12}. K_{a}=\left\{ \left(x_{1}, x_{2}, x_{3}, x_{4}\right) \in \mathbb {R}^{4} \mid x_{2}^{2}+x_{3}^{2}+x_{4}^{2} \leqslant a^{2} x_{1}^{2}\right\} .

Find γ4(Ka)\gamma_{4}\left(K_{a}\right).

?
Remark.
?

The symbol γm\gamma_{m} denotes the standard Gaussian measure on Rm\mathbb {R}^{m}, i.e., the measure with density (2π)−m/2e−∥x∥2/2(2 \pi )^{-m / 2} e^{-\left\| x\right\|^{2} / 2}.

When solving the problem it is useful to use the formula (see problem 3.26 b))

∫a⩽∥x∥⩽bf(∥x∥)dλm(x)=mαm∫abtm−1f(t)dt, \int _{a \leqslant \left\| x\right\| \leqslant b} f(\left\| x\right\| ) d \lambda _{m}(x)=m \alpha _{m} \int _{a}^{b} t^{m-1} f(t) d t,

where ff is a measurable nonnegative function on the interval ⟨a,b⟩\langle a, b\rangle (0⩽a<b⩽∞0 \leqslant a<b \leqslant \infty), αm=λm(Bm)=πm/2/Γ(m2+1)\alpha_{m}=\lambda_{m}\left(B^{m}\right)=\pi^{m / 2} / \Gamma \left(\frac{m}{2}+1\right).

Problem VIII.3.32

Compute the volume of the generalized torus T⊂R4T \subset \mathbb {R}^{4}

T={(x1,x2,x3,x4)∈R4∣(x1,x22+x32+x42)∈B⊂R2}, T=\left\{ \left(x_{1}, x_{2}, x_{3}, x_{4}\right) \in \mathbb {R}^{4} \mid \left(x_{1}, \sqrt{x_{2}^{2}+x_{3}^{2}+x_{4}^{2}}\right) \in B \subset \mathbb {R}^{2}\right\} ,

where BB is the disk with center at the point (0,a)(0, a) and radius R⩽aR \leqslant a.

?
Problem VIII.3.33

Let A⊂R2A \subset \mathbb {R}^{2} be a compact set lying in the upper half-plane. If T0T_{0} is the solid in R3\mathbb {R}^{3} obtained by rotating the set AA about the axis x1x_{1}, then by Guldin's theorem λ3(T0)=λ2(A)2πyc\lambda_{3}\left(T_{0}\right)=\lambda_{2}(A) 2 \pi y_{c}, where ycy_{c} is the ordinate of the centroid of the set AA. Regarding R2\mathbb {R}^{2} as canonically embedded in R4\mathbb {R}^{4}, consider the solid TT in R4\mathbb {R}^{4} "obtained by rotating AA about the axis x1x_{1}":

T={(x1,x2,x3,x4)∈R4∣(x1,x22+x32+x42)∈A}. T=\left\{ \left(x_{1}, x_{2}, x_{3}, x_{4}\right) \in \mathbb {R}^{4} \mid \left(x_{1}, \sqrt{x_{2}^{2}+x_{3}^{2}+x_{4}^{2}}\right) \in A\right\} .

Prove that the following modification of Guldin's theorem holds:

λ4(T)=λ2(A)⋅4πρ2, \lambda _{4}(T)=\lambda _{2}(A) \cdot 4 \pi \rho ^{2},

where ρ\rho is the "radius of gyration" of the set AA with respect to the axis x1x_{1}, i.e., the positive number defined by the equality

ρ2=1λ2(A)∬Ay2dxdy \rho ^{2}=\frac{1}{\lambda _{2}(A)} \iint _{A} y^{2} d x d y
?
Problem VIII.3.34

Let A⊂R2A \subset \mathbb {R}^{2} be a compact set lying in the upper half-plane. Regarding R2\mathbb {R}^{2} as canonically embedded in Rm\mathbb {R}^{m}, consider in Rm\mathbb {R}^{m} the set TT, the "solid obtained by rotating AA about the axis x1x_{1}":

T={(x1,x2,…,xm)∈Rm∣(x1,x22+…+xm2)∈A}. T=\left\{ \left(x_{1}, x_{2}, \ldots , x_{m}\right) \in \mathbb {R}^{m} \mid \left(x_{1}, \sqrt{x_{2}^{2}+\ldots +x_{m}^{2}}\right) \in A\right\} .

Prove that

?
(a)

λm(T)=σm−2∬Ax2m−2dx1dx2\lambda_{m}(T)=\sigma_{m-2} \iint_{A} x_{2}^{m-2} d x_{1} d x_{2};

(b)

γm(T)=σm−2(2π)m/2∬Ae−(x12+x22)/2x2m−2dx1dx2\gamma_{m}(T)=\frac{\sigma_{m-2}}{(2 \pi )^{m / 2}} \iint_{A} e^{-\left(x_{1}^{2}+x_{2}^{2}\right) / 2} x_{2}^{m-2} d x_{1} d x_{2},

where σm−2\sigma_{m-2} is the surface area of the sphere Sm−2S^{m-2}, and γm(T)\gamma_{m}(T) is the "Gaussian volume" of the set TT.

Remark.
?

The symbol γm\gamma_{m} denotes the standard Gaussian measure on Rm\mathbb {R}^{m}, i.e., the measure with density (2π)−m/2e−∥x∥2/2(2 \pi )^{-m / 2} e^{-\left\| x\right\|^{2} / 2}.

Problem VIII.3.35

Compute the integral

∫Rmexp⁡(−∑1⩽j⩽k⩽m∑jxjxk)dx. \int _{\mathbb {R}^{m}} \exp \left(-\sum _{1 \leqslant j \leqslant k \leqslant m} \sum _{j} x_{j} x_{k}\right) d x .
?
Problem VIII.3.36

Find the Lebesgue measure of the set

{(x1,…,xm)∈Rm∣∑1⩽k⩽m∣xk∣p⩽1}(p>0). \left\{ \left(x_{1}, \ldots , x_{m}\right) \in \mathbb {R}^{m} \mid \sum _{1 \leqslant k \leqslant m}\left|x_{k}\right|^{p} \leqslant 1\right\} \quad (p>0) .
?
Problem VIII.3.37
?
(a)

Compute the integral

Y=1σm−1∫Sm−1∣(x,a)∣pdμm−1(x) \mathcal{Y}=\frac{1}{\sigma _{m-1}} \int _{S^{m-1}}\left|(x, a)\right|^{p} d \mu _{m-1}(x)
(b)

Prove that for m,n∈N,1⩽n⩽m,n+q>0m, n \in \mathbb {N}, 1 \leqslant n \leqslant m, n+q>0, the following equality holds

1σm−1∫Sm−1(x12+…+xn2)q/2dμm−1(x)=Γ(n+q2)Γ(m+q2)−Γ(m2)Γ(n2)(∼n→∞(nm)q/2) \frac{1}{\sigma _{m-1}} \int _{S^{m-1}}\left(x_{1}^{2}+\ldots +x_{n}^{2}\right)^{q / 2} d \mu _{m-1}(x)=\frac{\Gamma \left(\frac{n+q}{2}\right)}{\Gamma \left(\frac{m+q}{2}\right)}-\frac{\Gamma \left(\frac{m}{2}\right)}{\Gamma \left(\frac{n}{2}\right)} \quad \left(\underset {n \rightarrow \infty }{\sim }\left(\frac{n}{m}\right)^{q / 2}\right)
Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}.

Problem VIII.3.38

Let F∈C(R),a∈RmF \in C(\mathbb {R}), a \in \mathbb {R}^{m}. Reduce the integrals of F((x,a))F((x, a)) over the unit ball and over the unit sphere in Rm\mathbb {R}^{m} to integrals over the segment [−1,1]\left[-1,1\right].

?
Problem VIII.3.39

Let a∈Rm,A:Rm→Rma \in \mathbb {R}^{m}, A: \mathbb {R}^{m} \rightarrow \mathbb {R}^{m} be a nondegenerate linear transformation, F∈C(R)F \in C(\mathbb {R}). Prove that

∫Sm−1F((x,a)∥A(x)∥)dμm−1(x)∥A(x)∥m=σm−1∣det⁡(A)∣∫−11F(ct)(1−t2)m−3/2dt \int _{S^{m-1}} F\left(\frac{(x, a)}{\left\| A(x)\right\| }\right) \frac{d \mu _{m-1}(x)}{\left\| A(x)\right\| ^{m}}=\frac{\sigma _{m-1}}{\left|\operatorname {det}(A)\right|} \int _{-1}^{1} F(c t)\left(1-t^{2}\right)^{m-3 / 2} d t

where c=∥(A−1)∗(a)∥,σm−1=μm−1(Sm−1)c=\left\| \left(A^{-1}\right)^{*}(a)\right\| , \sigma_{m-1}=\mu_{m-1}\left(S^{m-1}\right).

?
Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}.

Problem VIII.3.40

Let θ\theta be the angle between the nonzero vectors a,b∈Rm(0⩽θ⩽π)a, b \in \mathbb {R}^{m} (0 \leqslant \theta \leqslant \pi ). Prove that

∫Sm−1sign⁡(a,x)sign⁡(b,x)dμm−1(x)=(1−2πθ)σm−1 \int _{S^{m-1}} \operatorname {sign}(a, x) \operatorname {sign}(b, x) d \mu _{m-1}(x)=\left(1-\frac{2}{\pi } \theta \right) \sigma _{m-1}
?
Remark.
?

The symbol μm−1\mu_{m-1} denotes the Lebesgue measure (surface area) on the sphere Sm−1={x∈Rm∣∥x∥=1}S^{m-1}=\left\{ x \in \mathbb {R}^{m} \mid \left\| x\right\| =1\right\}, σm−1=μm−1(Sm−1)\sigma_{m-1}=\mu_{m-1}\left(S^{m-1}\right).

Problem VIII.3.41

Let Im=∫0∞γm(B(0,r))(1−γm(B(0,r)))drI_{m}=\int_{0}^{\infty } \gamma_{m}(B(0, r))\left(1-\gamma_{m}(B(0, r))\right) d r. Prove that

?
(a)

Im=12∫Rm∫Rm∣∥x∥−∥y∥∣dγm(x)dγm(y)I_{m}=\frac{1}{2} \int_{\mathbb {R}^{m}} \int_{\mathbb {R}^{m}}|\left\| x\right\| -\left\| y\right\| | d \gamma_{m}(x) d \gamma_{m}(y);

(b)

Im→12πI_{m} \rightarrow \frac{1}{\sqrt{2 \pi }}.

Remark.
?

The symbol γm\gamma_{m} denotes the standard Gaussian measure on Rm\mathbb {R}^{m}, i.e., the measure with density (2π)−m/2e−∥x∥2/2(2 \pi )^{-m / 2} e^{-\left\| x\right\|^{2} / 2}.

The formula given before problem 3.29 (see problem 3.26 b)):

∫a⩽∥x∥⩽bf(∥x∥)dλm(x)=mαm∫abtm−1f(t)dt, \int _{a \leqslant \left\| x\right\| \leqslant b} f(\left\| x\right\| ) d \lambda _{m}(x)=m \alpha _{m} \int _{a}^{b} t^{m-1} f(t) d t,

where ff is a measurable nonnegative function on the interval ⟨a,b⟩\langle a, b\rangle (0⩽a<b⩽∞0 \leqslant a<b \leqslant \infty), αm=λm(Bm)=πm/2/Γ(m2+1)\alpha_{m}=\lambda_{m}\left(B^{m}\right)=\pi^{m / 2} / \Gamma \left(\frac{m}{2}+1\right).

Problem VIII.3.42

Let A⊂RmA \subset \mathbb {R}^{m} be an open set that is star-shaped with respect to the origin,

qA(x)=inf⁡{t>0∣t−1x∈A}(x∈Rm) q_{A}(x)=\inf \left\{ t>0 \mid t^{-1} x \in A\right\} \quad \left(x \in \mathbb {R}^{m}\right)

the gauge function of the set AA, and let μ\mu be a probability measure on Rm\mathbb {R}^{m} such that ∫Rm∣(x,y)∣dμ(x)<+∞\int_{\mathbb {R}^{m}}\left|(x, y)\right| d \mu (x)<+\infty for every y∈Rmy \in \mathbb {R}^{m}.

Prove that

∫0+∞μ(tA)(1−μ(tA))dt=12∫Rm∫Rm∣qA(x)−qA(y)∣dμ(x)dμ(y)<+∞ \int _{0}^{+\infty } \mu (t A)(1-\mu (t A)) d t=\frac{1}{2} \int _{\mathbb {R}^{m}} \int _{\mathbb {R}^{m}}\left|q_{A}(x)-q_{A}(y)\right| d \mu (x) d \mu (y)<+\infty
?
Problem VIII.3.43

Let the function ff be summable on the cube Q=[0,1)mQ=[0,1)^{m} and have period 1 in each variable. Prove that

?
(a)

under a shift of QQ by any vector the integral ∫Qf(x)dx\int_{Q} f(x) d x does not change, i.e., for any a∈Rma \in \mathbb {R}^{m}

∫Qf(x)dx=∫a+Qf(x)dx, where a+Q={a+y∣y∈Q} \int _{Q} f(x) d x=\int _{a+Q} f(x) d x, \quad \text{ where } \quad a+Q=\left\{ a+y \mid y \in Q\right\}
(b)

if AA is an arbitrary integer m×mm \times m matrix with determinant equal to one, then

∫Qf(x)dx=∫A(Q)f(x)dx \int _{Q} f(x) d x=\int _{A(Q)} f(x) d x

Is this statement true if the determinant of the matrix AA equals one, but its entries need not be integers?

Problem VIII.3.44

Let TT be an arbitrary convex compact body in Rm\mathbb {R}^{m}, V=λm(T)V=\lambda_{m}(T). Prove that if the (m−1)(m-1)-dimensional volume of the projection of TT onto any hyperplane is not less than SS, then diam⁡T⩽mV/S\operatorname {diam} T \leqslant m V / S.

?
Problem VIII.3.45

Let T⊂RmT \subset \mathbb {R}^{m} be a convex compact body that is centrally symmetric with respect to the origin, and let E\mathscr {E} be an ellipsoid of maximal volume contained in TT:

E⊂T,λm(E)=sup⁡{λm(E)∣E⊂T,E− an ellipsoid }. \mathscr {E} \subset T, \quad \lambda _{m}(\mathscr {E})=\sup \left\{ \lambda _{m}(E) \mid E \subset T, \quad E-\text{ an ellipsoid }\right\} .

Prove that T⊂mET \subset \sqrt{m} \mathscr {E}.

?
Problem VIII.3.46

Let K⊂CK \subset \mathbb {C} be a closed bounded set of positive (planar) measure. Define the function φ\varphi by the equality

φ(z)=∬Kdξdηz−(ξ+iη)(z∈C) \varphi (z)=\iint _{K} \frac{d \xi d \eta }{z-(\xi +i \eta )} \quad (z \in \mathbb {C})

Prove that

?
(a)

φ∈C(C),zφ(z)→λ2(K)\varphi \in C(\mathbb {C}), z \varphi (z) \rightarrow \lambda_{2}(K) as z→∞z \rightarrow \infty;

(b)

the function φ\varphi is holomorphic outside KK.

Problem VIII.3.47

Identifying C2\mathbb {C}^{2} with R4\mathbb {R}^{4} in the natural way, consider on the sphere S3S^{3} the functions φm,n(m,n=0,1,2,…)\varphi_{m, n}(m, n=0,1,2, \ldots ), defined by the equalities φm,n(ξ1,ξ2)=ξ1mξ2n\varphi_{m, n}\left(\xi_{1}, \xi_{2}\right)=\xi_{1}^{m} \xi_{2}^{n}.

?
(a)

Compute the integrals

∫S3∣φm,n(ξ1,ξ2)∣2dμ3(ξ1,ξ2)(m,n=0,1,2,…) \int _{S^{3}}\left|\varphi _{m, n}\left(\xi _{1}, \xi _{2}\right)\right|^{2} d \mu _{3}\left(\xi _{1}, \xi _{2}\right) \quad (m, n=0,1,2, \ldots )
(b)

prove that the system {φm,n}m,n=0∞\left\{ \varphi_{m, n}\right\}_{m, n=0}^{\infty } is orthogonal;

(c)

prove that if the series ∑m,n⩾0am,nφm,n\sum_{m, n \geqslant 0} a_{m, n} \varphi_{m, n} converges uniformly on B4‾\overline{B^{4}} and its sum equals ff, then the following generalization of the Cauchy integral formula holds:

f(ξ1,ξ2)=12π2∫S3f(z1,z2)(1−ξ1zˉ1−ξ2zˉ2)2dμ3(z1,z2) f\left(\xi _{1}, \xi _{2}\right)=\frac{1}{2 \pi ^{2}} \int _{S^{3}} \frac{f\left(z_{1}, z_{2}\right)}{\left(1-\xi _{1} \bar{z}_{1}-\xi _{2} \bar{z}_{2}\right)^{2}} d \mu _{3}\left(z_{1}, z_{2}\right)

where (ξ1,ξ2)∈B4,B4={(ξ1,ξ2)∈C2∣∣ξ1∣2+∣ξ2∣2<1}\left(\xi_{1}, \xi_{2}\right) \in B^{4}, B^{4}=\left\{ \left(\xi_{1}, \xi_{2}\right) \in \mathbb {C}^{2} \mid \left|\xi_{1}\right|^{2}+\left|\xi_{2}\right|^{2}<1\right\}.