Chapter 3.6

The Maximum Principle

[75/100%]
Show
LaTeX
Problem 3.266

The maximum of ∣f(z)∣\left|f(z)\right| on the disk ∣z∣≤r\left|z\right|\leq r is M(r)M(r).

?
Problem 3.267

The maximum M(r)M(r) is monotone increasing with rr unless f(z)f(z) is a constant.

?
Problem 3.268

We assume that the function f(z)f(z) is regular in the simply connected region ∣z∣>R\left|z\right|>R. M(r)M(r) denotes the maximum of ∣f(z)∣\left|f(z)\right| on the circle ∣z∣=r\left|z\right|=r, r>Rr>R. Then M(r)M(r) is also the maximum of ∣f(z)∣\left|f(z)\right| for ∣z∣≥r\left|z\right|\geq r and M(r)M(r) is monotone decreasing unless f(z)f(z) is a constant.

?
Problem 3.269

Let f(z)f(z) denote a polynomial of degree nn; then

M(r1)r1n≥M(r2)r2n,0<r1<r2. \frac{M(r_{1})}{r_{1}^{n}} \geq \frac{M(r_{2})}{r_{2}^{n}}, \quad 0<r_{1}<r_{2}.

Equality is attained only if the polynomial is of the form czncz^{n}.

?
Problem 3.270

Suppose that f(z)f(z) is a polynomial of degree nn and that

∣f(z)∣≤M \left|f(z)\right| \leq M

on the real interval −1≤z≤1-1\leq z\leq 1. Then we have

∣f(z)∣≤M(a+b)n \left|f(z)\right| \leq M(a+b)^{n}

for any zz outside this interval; aa and bb are the semi-axes of the ellipse through zz and with foci −1-1 and 11. What does the proposition imply for z→∞z\to \infty?

?
Problem 3.271

We assume that f(z)f(z) is a polynomial of degree nn, that E1E_{1} and E2E_{2} denote two homofocal ellipses with semi-axes a1,b1a_{1},b_{1} and a2,b2a_{2},b_{2}, a1<a2a_{1}<a_{2}, b1<b2b_{1}<b_{2}. The maximum of ∣f(z)∣\left|f(z)\right| on E1E_{1} and E2E_{2} is denoted by M1M_{1} and M2M_{2} resp.; then

M1(a1+b1)n≥M2(a2+b2)n. \frac{M_{1}}{(a_{1}+b_{1})^{n}} \geq \frac{M_{2}}{(a_{2}+b_{2})^{n}}.

Derive problems 269 and 270 from this proposition.

?
Problem 3.272

If an analytic function is regular in a closed disk and not a constant its absolute value at the center of the disk is smaller than the arithmetic mean of its absolute value on the boundary of the disk.

?
Problem 3.273

If the absolute value of an analytic function f(z)f(z) is constant in an open set of the zz-plane (e.g.in a disk) f(z)f(z) must be a constant.

?
Problem 3.274

We suppose that the functions φ(z)\varphi (z) and ψ(z)\psi (z) are regular in the closed disk ∣z∣≤1\left|z\right|\leq 1 and non-zero in the open disk ∣z∣<1\left|z\right|<1 and that, besides, φ(0)\varphi (0) and ψ(0)\psi (0) are real and positive. If φ(z)\varphi (z) and ψ(z)\psi (z) have the same modulus on the circle ∣z∣=1\left|z\right|=1 then φ(z)=ψ(z)\varphi (z)=\psi (z) identically.

?
Problem 3.275

The function f(z)f(z) is regular and single-valued in the interior of the closed domain D\mathfrak {D} and continuous in D\mathfrak {D}, boundary included; the maximum of ∣f(z)∣\left|f(z)\right| on the boundary of D\mathfrak {D} is called MM. Under these conditions we have the inequality

∣f(z)∣<M \left|f(z)\right| < M

in the interior of D\mathfrak {D} unless f(z)f(z) is a constant. [This statement is stronger than problem 135.]

?
Problem 3.276

Let D\mathfrak {D} denote a domain, ζ\zeta be an inner point of D\mathfrak {D} and B\mathfrak {B} be the set of those boundary points of D\mathfrak {D} whose distance to ζ\zeta does not exceed ϱ\varrho. The circle of radius ϱ\varrho and center ζ\zeta is assumed to have an arc that does not belong to D\mathfrak {D} and the length of which is not smaller than 2πϱn\dfrac {2\pi \varrho }{n}, nn integer.

We suppose that the function f(z)f(z) is regular and single-valued in the interior of D\mathfrak {D} and continuous on the boundary and that, in particular, ∣f(z)∣≤a\left|f(z)\right|\leq a at the points of B\mathfrak {B} and ∣f(z)∣≤A\left|f(z)\right|\leq A at the remaining boundary points of D\mathfrak {D}; a<Aa<A. Then

∣f(ζ)∣≤a1nA1−1n. \left|f(\zeta )\right| \leq a^{\frac{1}{n}}A^{1-\frac{1}{n}}.

[Examine the product ∏ν=0n−1f[ζ+(z−ζ)ω−ν]\displaystyle \prod_{\nu =0}^{n-1} f[\zeta +(z-\zeta )\omega^{-\nu }] in a suitable domain, ω=e2πi/n\omega =e^{2\pi i/n}.]

?
Problem 3.277

We assume that the function f(z)f(z) is regular and bounded in the sector 0<arg⁡z<α0<\arg z<\alpha, continuous on the real axis and that lim⁡x→∞f(x)=0\lim \limits_{x\to \infty } f(x)=0, xx real, x>0x>0. Then the limit relation

lim⁡∣z∣→∞f(z)=0 \lim _{\left|z\right|\to \infty } f(z) = 0

holds uniformly in any sector 0≤arg⁡z≤α−ε<α0\leq \arg z\leq \alpha -\varepsilon <\alpha.

?
Problem 3.278

We denote by MM a positive constant and by R\mathfrak {R} a connected open region. The analytic function f(z)f(z) is assumed to have the following properties:

  1. f(z)f(z) is regular at every point of R\mathfrak {R};

  2. f(z)f(z) is single-valued in R\mathfrak {R};

  3. each boundary point of R\mathfrak {R} has for every positive number ε\varepsilon a neighbourhood such that at every point zz of R\mathfrak {R} in this neighbourhood the inequality

    ∣f(z)∣<M+ε \left|f(z)\right| < M+\varepsilon

    is satisfied. These conditions imply

∣f(z)∣≤Mfor z in R \left|f(z)\right| \leq M \quad \text{for } z \text{ in } \mathfrak {R}

and even ∣f(z)∣<M\left|f(z)\right|<M unless f(z)f(z) is a constant. [Stronger than problem 275.]

?
Problem 3.279

Let f(z)f(z) be regular and bounded on the disk ∣z∣<1\left|z\right|<1 and let

lim⁡r→1f(reiϑ)=0 \lim _{r\to 1} f(re^{i\vartheta }) = 0

hold uniformly in a sector α≤ϑ≤β\alpha \leq \vartheta \leq \beta, α<β\alpha <\beta. Then f(z)f(z) vanishes identically.

?
Problem 3.280

The function f(z)f(z) is assumed to be regular and ∣f(z)∣<1\left|f(z)\right|<1 in the disk ∣z∣<1\left|z\right|<1. If f(0)=0f(0)=0 either the stricter inequality ∣f(z)∣<∣z∣\left|f(z)\right|<\left|z\right| holds for z≠0z\neq 0 or f(z)=eiαzf(z)=e^{i\alpha }z, α=\alpha = real.

?
Problem 3.281

We denote by z=φ(ζ)z=\varphi (\zeta ) and w=ψ(ζ)w=\psi (\zeta ) two schlicht maps of the unit disk ∣ζ∣<1\left|\zeta \right|<1 into the regions R\mathfrak {R} and S\mathfrak {S} of the zz- and ww-plane resp.; the images of the origin ζ=0\zeta =0 are the points z=z0z=z_{0} and w=w0w=w_{0} resp. In addition let 0<ϱ<10<\varrho <1 and r\mathfrak {r} and s\mathfrak {s} be the images of the disk ∣ζ∣≤ϱ\left|\zeta \right|\leq \varrho under the above mentioned maps. We assume that w=f(z)w=f(z) is a regular analytic function in R\mathfrak {R} the range of which belongs to S\mathfrak {S} and that f(z0)=w0f(z_{0})=w_{0}. Then f(z)f(z) assumes in the subdomain r\mathfrak {r} of R\mathfrak {R} values that belong to the subdomain s\mathfrak {s} of S\mathfrak {S}. These values are in the interior of s\mathfrak {s} unless f(z)f(z) is a schlicht function that maps the region R\mathfrak {R} onto the region S\mathfrak {S}.

?
Problem 3.282

Let f(z)f(z) be regular and ∣f(z)∣<1\left|f(z)\right|<1 for ∣z∣<1\left|z\right|<1. Then

∣f(z)−f(0)∣≤∣z∣1−∣f(0)∣21−∣f(0)∣∣z∣,0<∣z∣<1. \left|f(z)-f(0)\right| \leq \left|z\right|\frac{1-\left|f(0)\right|^{2}}{1-\left|f(0)\right|\left|z\right|}, \quad 0<\left|z\right|<1.

The relation will be an equality only for the linear function

f(z)=eiαz+w01+w0‾eiαz,α real. f(z) = \frac{e^{i\alpha }z+w_{0}}{1+\overline{w_{0}}e^{i\alpha }z}, \quad \alpha \text{ real}.
?
Problem 3.283

Suppose that f(z)f(z) is regular for ∣z∣<R\left|z\right|<R and that A(r)A(r) denotes the maximum of the real part of f(z)f(z) for ∣z∣≤r\left|z\right|\leq r, 0≤r<R0\leq r<R. Then we have the inequality

A(r)≤R−rR+rA(0)+2rR+rA(R),0<r<R, A(r) \leq \frac{R-r}{R+r}A(0) + \frac{2r}{R+r}A(R), \quad 0<r<R,

where lim⁡r→R−0A(r)=A(R)\lim \limits_{r\to R-0} A(r)=A(R) [A(r)A(r) increases monotonically with rr, problem 313]. There is equality only for the linear function

f(z)=Rw0+[w0‾−2A(R)]eiαzR−eiαz,α real. f(z) = \frac{Rw_{0}+[\overline{w_{0}}-2A(R)]e^{i\alpha }z}{R-e^{i\alpha }z}, \quad \alpha \text{ real}.
?
Problem 3.284

(continued). The maximum M(r)M(r) of the absolute value of f(z)f(z) in the disk ∣z∣≤r\left|z\right|\leq r is restricted by

M(r)≤M(0)+2rR−r[A(R)−A(0)]≤R+rR−rM(0)+2rR−rA(R). M(r) \leq M(0) + \frac{2r}{R-r}[A(R)-A(0)] \leq \frac{R+r}{R-r}M(0) + \frac{2r}{R-r}A(R).
?
Problem 3.285

We assume that f(z)f(z) is regular, non-zero and bounded for ∣z∣<R\left|z\right|<R. Then

M(r)≤M(0)R−rR+rM(R)2rR+r,0<r<R, M(r) \leq M(0)^{\frac{R-r}{R+r}}M(R)^{\frac{2r}{R+r}}, \quad 0<r<R,

where lim⁡r→R−0M(r)=M(R)\lim \limits_{r\to R-0} M(r) = M(R).

?
Problem 3.286

The functions f1(z),f2(z),f3(z),…,fn(z),…f_{1}(z),f_{2}(z),f_{3}(z),\ldots ,f_{n}(z),\ldots are supposed to be regular, non-zero, and of absolute value smaller than 11, for ∣z∣<1\left|z\right|<1. If the series

f1(0)+f2(0)+f3(0)+⋯+fn(0)+⋯ f_{1}(0)+f_{2}(0)+f_{3}(0)+\cdots +f_{n}(0)+\cdots

is absolutely convergent, the series

[f1(z)]2+[f2(z)]2+[f3(z)]2+⋯+[fn(z)]2+⋯ [f_{1}(z)]^{2}+[f_{2}(z)]^{2}+[f_{3}(z)]^{2}+\cdots +[f_{n}(z)]^{2}+\cdots

is absolutely convergent for ∣z∣≤13\left|z\right|\leq \dfrac {1}{3}.

?
Problem 3.287

We assume that the function f(z)f(z) is regular, has positive real part for ∣z∣<1\left|z\right|<1 and that f(0)f(0) is real. Then we have

f(0)1−∣z∣1+∣z∣≤\ref(z)≤f(0)1+∣z∣1−∣z∣,∣\imf(z)∣≤f(0)2∣z∣1−∣z∣2, f(0)\frac{1-\left|z\right|}{1+\left|z\right|} \leq \re f(z) \leq f(0)\frac{1+\left|z\right|}{1-\left|z\right|}, \quad \left|\im f(z)\right| \leq f(0)\frac{2\left|z\right|}{1-\left|z\right|^{2}}, f(0)1−∣z∣1+∣z∣≤∣f(z)∣≤f(0)1+∣z∣1−∣z∣,0<∣z∣<1. f(0)\frac{1-\left|z\right|}{1+\left|z\right|} \leq \left|f(z)\right| \leq f(0)\frac{1+\left|z\right|}{1-\left|z\right|}, \quad 0<\left|z\right|<1.

There is equality only if

f(z)=w01+eiαz1−eiαz,w0,α real,  w0>0. f(z) = w_{0}\frac{1+e^{i\alpha }z}{1-e^{i\alpha }z}, \quad w_{0},\alpha \text{ real}, \; w_{0}>0.
?
Problem 3.288

Let f(z)f(z) be a regular function with ∣\ref(z)∣<1\left|\re f(z)\right|<1 for ∣z∣<1\left|z\right|<1. If f(0)=0f(0)=0 the stronger inequality

∣\ref(z)∣≤4πarctan⁡∣z∣,0<∣z∣<1, \left|\re f(z)\right| \leq \frac{4}{\pi }\arctan \left|z\right|, \quad 0<\left|z\right|<1,

holds; in addition

∣\imf(z)∣≤2πlog⁡1+∣z∣1−∣z∣,0<∣z∣<1. \left|\im f(z)\right| \leq \frac{2}{\pi }\log \frac{1+\left|z\right|}{1-\left|z\right|}, \quad 0<\left|z\right|<1.

Equality occurs if and only if

f(z)=2πilog⁡1+eiαz1−eiαz,α real. f(z) = \frac{2}{\pi i}\log \frac{1+e^{i\alpha }z}{1-e^{i\alpha }z}, \quad \alpha \text{ real}.
?
Problem 3.289

Suppose that the function f(z)f(z) is regular for ∣z∣<R\left|z\right|<R and that Δ\Delta denotes the oscillation of its real part in the disk, i.e.

∣\ref(z1)−\ref(z2)∣<Δ \left|\re f(z_{1})-\re f(z_{2})\right| < \Delta

for ∣z1∣<R\left|z_{1}\right|<R, ∣z2∣<R\left|z_{2}\right|<R. Then the oscillation of its real part in the smaller disk ∣z∣≤r\left|z\right|\leq r, r<Rr<R, is

∣\ref(z1)−\ref(z2)∣≤4Δπarctan⁡rR,∣z1∣≤r,  ∣z2∣≤r. \left|\re f(z_{1})-\re f(z_{2})\right| \leq \frac{4\Delta }{\pi }\arctan \frac{r}{R}, \quad \left|z_{1}\right|\leq r, \; \left|z_{2}\right|\leq r.

The oscillation of the imaginary part is restricted by the inequality

∣\imf(z1)−\imf(z2)∣≤2Δπlog⁡R+rR−r,∣z1∣≤r,  ∣z2∣≤r. \left|\im f(z_{1})-\im f(z_{2})\right| \leq \frac{2\Delta }{\pi }\log \frac{R+r}{R-r}, \quad \left|z_{1}\right|\leq r, \; \left|z_{2}\right|\leq r.
?
Problem 3.290

We denote by T\mathfrak {T} the infinite region of the zz-plane that is symmetric with respect to the xx-axis and whose points are characterized by the inequalities

x>0,−k(x)<y<k(x), x>0, \quad -k(x)<y<k(x),

where k(x)k(x) is a positive and continuous function defined for x≥0x\geq 0. Then there exists a positive function h(x)h(x) that depends only on T\mathfrak {T} and has the following properties: If F(z)F(z) is regular in T\mathfrak {T} and bounded from below, ∣F(z)∣>c\left|F(z)\right|>c, then

log⁡∣F(x)∣h(x) \frac{\log \left|F(x)\right|}{h(x)}

is bounded from above as xx increases to infinity. (The theorem is particularly interesting when k(x)k(x) converges monotonically to 00; i.e.when the region T\mathfrak {T} is ``tapering''; whereas an analytic function can increase arbitrarily fast along a ray (Part IV, problem 180) the rate of increase is limited if the function grows to infinity in some properly chosen neighbourhood of the ray. The result can be formulated with reference to the points of the modular graph somewhat vaguely so: If none should fall below a certain minimum standard none may rise arbitrarily high.)

?
Problem 3.291

Let f(z)f(z) be a regular function and ∣f(z)∣<1\left|f(z)\right|<1 in the unit disk ∣z∣<1\left|z\right|<1; in addition let f(z)f(z) be regular at z=1z=1 and f(0)=0f(0)=0, f(1)=1f(1)=1. Then f′(1)f'(1) must be real and f′(1)≥1f'(1)\geq 1.

?
Problem 3.292

We assume that the function f(z)f(z) is regular and ∣f(z)∣<1\left|f(z)\right|<1 in the unit disk ∣z∣<1\left|z\right|<1, and that, besides, f(z)f(z) is regular for z=1z=1 and f(1)=1f(1)=1. Then the derivative f′(1)f'(1) is real and

f′(1)1−∣f(z)∣2∣1−f(z)∣2≥1−∣z∣2∣1−z∣2,∣z∣<1. f'(1)\frac{1-\left|f(z)\right|^{2}}{\left|1-f(z)\right|^{2}} \geq \frac{1-\left|z\right|^{2}}{\left|1-z\right|^{2}}, \quad \left|z\right|<1.
?
Problem 3.293

Suppose that f(z)f(z) is regular and \imf(z)>0\im f(z)>0 in the upper half-plane Im(z)>0\mathrm{Im}\left(z\right)>0 and that, besides, f(z)f(z) is regular at the point z=az=a of the real axis and f(a)=bf(a)=b, bb real. Then f′(a)f'(a) is real and positive and we have the inequality

\im1b−f(z)≥\im1(a−z)f′(a)for Im(z)>0. \im \frac{1}{b-f(z)} \geq \im \frac{1}{(a-z)f'(a)} \quad \text{for } \mathrm{Im}\left(z\right)>0.
?
Problem 3.294

Let the function f(z)f(z) be regular, have the zeros z1,z2,…,znz_{1},z_{2},\ldots ,z_{n} and let f(z)f(z) be bounded, ∣f(z)∣≤M\left|f(z)\right|\leq M, in the unit disk ∣z∣<1\left|z\right|<1. Then the stronger inequality

∣f(z)∣≤∣z−z11−z1‾z⋅z−z21−z2‾z⋯z−zn1−zn‾z∣M \left|f(z)\right| \leq \left|\frac{z-z_{1}}{1-\overline{z_{1}}z}\cdot \frac{z-z_{2}}{1-\overline{z_{2}}z}\cdots \frac{z-z_{n}}{1-\overline{z_{n}}z}\right|M

holds for ∣z∣<1\left|z\right|<1. We have equality either at every point or at no point of the open disk ∣z∣<1\left|z\right|<1. (Proposition 280 is a special case, n=1n=1, z1=0z_{1}=0.)

?
Problem 3.295

Let the function f(z)f(z) be regular, have the zeros z1,z2,…,znz_{1},z_{2},\ldots ,z_{n} and let f(z)f(z) be bounded, ∣f(z)∣≤M\left|f(z)\right|\leq M, in the open half-plane Re(z)>0\mathrm{Re}\left(z\right)>0. Then the stronger inequality

∣f(z)∣≤∣z1−zz1‾+z⋅z2−zz2‾+z⋯zn−zzn‾+z∣M \left|f(z)\right| \leq \left|\frac{z_{1}-z}{\overline{z_{1}}+z}\cdot \frac{z_{2}-z}{\overline{z_{2}}+z}\cdots \frac{z_{n}-z}{\overline{z_{n}}+z}\right|M

holds for all zz with Re(z)>0\mathrm{Re}\left(z\right)>0. We have equality either at every point or at no point of the right half-plane Re(z)>0\mathrm{Re}\left(z\right)>0.

?
Problem 3.296

A function that is meromorphic in a closed disk and of constant absolute value on the boundary circle is a rational function; in fact it is, up to a constant factor, the product of linear fractional functions that map the disk in question either onto the interior or the exterior of the unit circle.

?
Problem 3.297

[W. Blaschke: Sber. Naturf. Ges. Lpz. Vol. 67, p.194 (1915).] We assume that the function f(z)f(z) is regular and bounded in the disk ∣z∣<1\left|z\right|<1 and vanishes at the points z1,z2,z3,…z_{1},z_{2},z_{3},\ldots Then

(1−∣z1∣)+(1−∣z2∣)+(1−∣z3∣)+⋯ (1-\left|z_{1}\right|) + (1-\left|z_{2}\right|) + (1-\left|z_{3}\right|) + \cdots

(the sum of the distances of the zeros from the unit circle) is finite or else f(z)≡0f(z)\equiv 0.

?
Problem 3.298

We assume that the function f(z)f(z) is regular and bounded in the half-plane Re(z)>0\mathrm{Re}\left(z\right)>0 and vanishes at the points z1,z2,z3,…z_{1},z_{2},z_{3},\ldots outside the unit disk in the half-plane, i.e.∣zn∣>1\left|z_{n}\right|>1, Re(zn)>0\mathrm{Re}\left(z_{n}\right)>0, n=1,2,3,…n=1,2,3,\ldots Then the sum of the series

\re1z1+\re1z2+\re1z3+⋯ \re \frac{1}{z_{1}} + \re \frac{1}{z_{2}} + \re \frac{1}{z_{3}} + \cdots

is finite or else f(z)≡0f(z)\equiv 0.

?
Problem 3.299

The sum of the absolute values of several analytic functions attains its maximum on the boundary. Here is a more detailed statement: The functions f1(z),f2(z),f3(z),…,fn(z)f_{1}(z),f_{2}(z),f_{3}(z),\ldots ,f_{n}(z) are supposed to be regular and single-valued in the domain D\mathfrak {D}. Then the function

φ(z)=∣f1(z)∣+∣f2(z)∣+⋯+∣fn(z)∣, \varphi (z) = \left|f_{1}(z)\right| + \left|f_{2}(z)\right| + \cdots + \left|f_{n}(z)\right|,

which is continuous in D\mathfrak {D}, assumes its maximum on the boundary of D\mathfrak {D}.

?
Problem 3.300

(continued). The function φ(z)\varphi (z) assumes its maximum only on the boundary of D\mathfrak {D} unless all the functions f1(z),f2(z),…,fn(z)f_{1}(z),f_{2}(z),\ldots ,f_{n}(z) are constants.

?
Problem 3.301

In three dimensional space, the nn points P1,P2,…,PnP_{1},P_{2},\ldots ,P_{n} are given and PP denotes a variable point. The function

φ(P)=PP1‾⋅PP2‾⋯PPn‾ \varphi (P) = \overline{PP_{1}}\cdot \overline{PP_{2}}\cdots \overline{PP_{n}}

(PPν‾\overline{PP_{\nu }} is the distance between PP and PνP_{\nu }) of the point PP assumes its maximum in any domain on the boundary. (Generalization of problem 137.)

?
Problem 3.302

We assume that the functions f1(z),f2(z),…,fn(z)f_{1}(z),f_{2}(z),\ldots ,f_{n}(z) are regular and single-valued in the domain D\mathfrak {D}. Let p1,p2,…,pnp_{1},p_{2},\ldots ,p_{n} denote positive numbers. The function

φ(z)=∣f1(z)∣p1+∣f2(z)∣p2+⋯+∣fn(z)∣pn \varphi (z) = \left|f_{1}(z)\right|^{p_{1}} + \left|f_{2}(z)\right|^{p_{2}} + \cdots + \left|f_{n}(z)\right|^{p_{n}}

is continuous in D\mathfrak {D}. It reaches its maximum only on the boundary of D\mathfrak {D} unless all the functions f1(z),f2(z),…,fn(z)f_{1}(z),f_{2}(z),\ldots ,f_{n}(z) are constants.

?
Problem 3.303

The function f(z)f(z) is supposed to be regular in the multiply connected closed domain D\mathfrak {D} and ∣f(z)∣\left|f(z)\right| single-valued in D\mathfrak {D}. [f(z)f(z) is not necessarily single-valued.] The absolute value ∣f(z)∣\left|f(z)\right| attains its maximum at a boundary point of D\mathfrak {D}. The maximum cannot be attained at an inner point of D\mathfrak {D} unless f(z)f(z) is a constant.

?
Problem 3.304

Let the function f(z)f(z) be regular in the disk ∣z∣<R\left|z\right|<R. Suppose

0<r1<r2<r3<R. 0 < r_{1} < r_{2} < r_{3} < R.

Then

log⁡M(r2)≤log⁡r2−log⁡r1log⁡r3−log⁡r1log⁡M(r3)+log⁡r3−log⁡r2log⁡r3−log⁡r1log⁡M(r1). \log M(r_{2}) \leq \frac{\log r_{2}-\log r_{1}}{\log r_{3}-\log r_{1}}\log M(r_{3}) + \frac{\log r_{3}-\log r_{2}}{\log r_{3}-\log r_{1}}\log M(r_{1}).

This means that in an orthogonal system of coordinates the graph of log⁡M(r)\log M(r) as a function of log⁡r\log r appears as a convex curve. (Hadamard's three circle theorem.) [Examine zαf(z)z^{\alpha }f(z) with suitably chosen α\alpha.]

?
Problem 3.305

(continued). The function log⁡M(r)\log M(r) is a strictly convex function of log⁡r\log r unless f(z)f(z) is of the form azαaz^{\alpha }, a,αa,\alpha constants, α\alpha real: this is the only type of function for which the inequality in problem 304 becomes an equality.

?
Problem 3.306

Suppose that the function f(z)f(z) is regular for ∣z∣<R\left|z\right|<R but not of the form czncz^{n}, cc constant, nn integer; let

I2(r)=12π∫02π∣f(reiϑ)∣2 dϑ I_{2}(r) = \frac{1}{2\pi }\int _{0}^{2\pi } \left|f(re^{i\vartheta })\right|^{2}\, d\vartheta

denote the arithmetic mean of ∣f(z)∣2\left|f(z)\right|^{2} on the circle ∣z∣=r\left|z\right|=r, r<Rr<R. The function I2(r)I_{2}(r) is monotone increasing with rr and log⁡I2(r)\log I_{2}(r) is a strictly convex function of log⁡r\log r.

?
Problem 3.307

Let f(z)f(z) be regular for ∣z∣<R\left|z\right|<R;

G(r)=e12π∫02πlog⁡∣f(reiϑ)∣ dϑ \mathfrak {G}(r) = e^{\frac{1}{2\pi }\int _{0}^{2\pi } \log \left|f(re^{i\vartheta })\right|\, d\vartheta }

denotes the geometric mean of ∣f(z)∣\left|f(z)\right| on the circle ∣z∣=r\left|z\right|=r, r<Rr<R. The function G(r)\mathfrak {G}(r) is monotone increasing with rr and a convex function of log⁡r\log r (in the wide sense).

?
Problem 3.308

The function f(z)f(z) is supposed to be regular for ∣z∣<R\left|z\right|<R and not a constant. We put

I(r)=12π∫02π∣f(reiϑ)∣ dϑ,r<R. I(r) = \frac{1}{2\pi }\int _{0}^{2\pi } \left|f(re^{i\vartheta })\right|\, d\vartheta , \quad r<R.

The function I(r)I(r) is monotone increasing with rr and log⁡I(r)\log I(r) is a convex function of log⁡r\log r. [Problems 299, 304.]

?
Problem 3.309

Let f(z)f(z) be regular and not a constant in the disk ∣z∣<R\left|z\right|<R. The function w=f(z)w=f(z) maps the circle ∣z∣=r\left|z\right|=r, r<Rr<R, in the zz-plane onto a curve in the ww-plane with length l(r)l(r). The ratio l(r)2πr\dfrac {l(r)}{2\pi r} is monotone increasing with rr.

?
Problem 3.310

We suppose that f(z)f(z) is regular and not constant for ∣z∣<R\left|z\right|<R and that pp is a positive number. We define

Ip(r)=12π∫02π∣f(reiϑ)∣p dϑ,r<R. I_{p}(r) = \frac{1}{2\pi }\int _{0}^{2\pi } \left|f(re^{i\vartheta })\right|^{p}\, d\vartheta , \quad r<R.

The function Ip(r)I_{p}(r) is monotone increasing with rr and log⁡Ip(r)\log I_{p}(r) is a convex function of log⁡r\log r. (Cf.problems 306, 308, 307, 267 and 304, for an analogous case see Part IV, problem 19.)

?
Problem 3.311

An analytic function that is regular in the closed disk R\mathfrak {R} cannot assume real values at all the boundary points of R\mathfrak {R} except if it is a real constant.

?
Problem 3.312

We denote by u(x,y)u(x,y), z=x+iyz=x+iy, a harmonic function that is regular in the disk (x−x0)2+(y−y0)2≤r2(x-x_{0})^{2}+(y-y_{0})^{2}\leq r^{2}. The value of u(x,y)u(x,y) at the center is

u(x0,y0)=12π∫02πu(x0+rcos⁡ϑ,y0+rsin⁡ϑ) dϑ[problem 118], u(x_{0},y_{0}) = \frac{1}{2\pi }\int _{0}^{2\pi } u(x_{0}+r\cos \vartheta ,y_{0}+r\sin \vartheta )\, d\vartheta \quad [\text{problem 118}],

consequently

∣u(x0,y0)∣≤12π∫02π∣u(x0+rcos⁡ϑ,y0+rsin⁡ϑ)∣ dϑ. \left|u(x_{0},y_{0})\right| \leq \frac{1}{2\pi }\int _{0}^{2\pi } \left|u(x_{0}+r\cos \vartheta ,y_{0}+r\sin \vartheta )\right|\, d\vartheta .

If a harmonic function is regular in a closed disk its absolute value at the center does not exceed the arithmetic mean of its absolute values on the boundary. Under what conditions is there equality?

?
Problem 3.313

A harmonic function is supposed to be regular and single-valued in a domain D\mathfrak {D}. Then it attains its maximum and its minimum on the boundary and only on the boundary unless it is a constant.

?
Problem 3.314

A harmonic function that is regular in the domain D\mathfrak {D} and vanishes at all the boundary points of D\mathfrak {D} is identically zero.

?
Problem 3.315

The equilibrium described in solution 31 is not stable.

?
Problem 3.316

A harmonic function is assumed to be single-valued in the domain D\mathfrak {D} and regular with the exception of finitely many points at which it becomes −∞-\infty (i.e.the function converges to −∞-\infty as zz approaches such a point). Then it assumes its maximum on the boundary of D\mathfrak {D}.

?
Problem 3.317

Let the function f(z)f(z) be regular in the disk ∣z∣≤R\left|z\right|\leq R. The function w=f(z)w=f(z) maps the disk onto a certain domain of the ww-plane that is assumed to be star-shaped with respect to w=0w=0. Let f(0)=0f(0)=0. Then the images of the concentric circles ∣z∣=r\left|z\right|=r, r<Rr<R, are also star-shaped with respect to the origin.

?
Problem 3.318

Suppose that the function f(z)f(z) is regular in the disk ∣z∣≤R\left|z\right|\leq R and that it maps the disk onto a convex domain of the ww-plane. Then the image of an arbitrary circle in the open disk ∣z∣<R\left|z\right|<R is also convex.

?
Problem 3.319

Let u1(x,y),u2(x,y),…,un(x,y)u_{1}(x,y),u_{2}(x,y),\ldots ,u_{n}(x,y), z=x+iyz=x+iy, be regular harmonic functions in a domain D\mathfrak {D}. The continuous function

∣u1(x,y)∣+∣u2(x,y)∣+⋯+∣un(x,y)∣ \left|u_{1}(x,y)\right| + \left|u_{2}(x,y)\right| + \cdots + \left|u_{n}(x,y)\right|

assumes its maximum on the boundary of D\mathfrak {D}.

?
Problem 3.320

Consider a regular harmonic function in the disk ∣z∣<R\left|z\right|<R. We denote by A(r)A(r) its maximum on the circle ∣z∣=r\left|z\right|=r, r<Rr<R. When 0<r1<r2<r3<R0<r_{1}<r_{2}<r_{3}<R we have

A(r2)≤log⁡r2−log⁡r1log⁡r3−log⁡r1A(r3)+log⁡r3−log⁡r2log⁡r3−log⁡r1A(r1), A(r_{2}) \leq \frac{\log r_{2}-\log r_{1}}{\log r_{3}-\log r_{1}}A(r_{3}) + \frac{\log r_{3}-\log r_{2}}{\log r_{3}-\log r_{1}}A(r_{1}),

i.e.A(r)A(r) is a convex function of log⁡r\log r.

?
Problem 3.321

Deduce Hadamard's three circle theorem 304 from problem 320 and vice versa problem 320 from Hadamard's three circle theorem.

?
Problem 3.322

Let α\alpha be given, 0<α<π20<\alpha <\dfrac {\pi }{2}. The function f(z)f(z) is assumed to be regular in the sector −α≤ϑ≤α-\alpha \leq \vartheta \leq \alpha (z=reiϑz=re^{i\vartheta }). In addition f(z)f(z) has the following properties:

  1. there exist two positive constants AA and BB such that

    ∣f(z)∣<AeB∣z∣for−α≤ϑ≤α,  z=reiϑ; \left|f(z)\right| < Ae^{B\left|z\right|} \quad \text{for} \quad -\alpha \leq \vartheta \leq \alpha , \; z=re^{i\vartheta };
  2. ∣f(z)∣≤1\left|f(z)\right|\leq 1 for ϑ=−α\vartheta =-\alpha and ϑ=α\vartheta =\alpha (i.e.on the boundary of the sector considered). Then the inequality ∣f(z)∣≤1\left|f(z)\right|\leq 1 holds in the entire sector.

?
Problem 3.323

Condition (1) of problem 322 can be generalized insofar as the inequality

∣f(z)∣<AeB∣z∣ \left|f(z)\right| < Ae^{B\left|z\right|}

need not be required to hold in the entire sector but only on the arcs of the circles ∣z∣=r1,∣z∣=r2,…,∣z∣=rn,…\left|z\right|=r_{1},\left|z\right|=r_{2},\ldots ,\left|z\right|=r_{n},\ldots intercepted by the rays ϑ=−α\vartheta =-\alpha, ϑ=α\vartheta =\alpha, lim⁡n→∞rn=∞\lim \limits_{n\to \infty } r_{n}=\infty. The conclusion, namely ∣f(z)∣≤1\left|f(z)\right|\leq 1 in the entire sector, remains the same. What more general curves can replace the circular arcs?

?
Problem 3.324

We modify condition (2) of problem 322 in the following way: There exist in the sector −α≤ϑ≤α-\alpha \leq \vartheta \leq \alpha two curves Γ1\Gamma_{1} and Γ2\Gamma_{2} connecting the points z=0z=0 and z=∞z=\infty that do not intersect and along which ∣f(z)∣≤1\left|f(z)\right|\leq 1. This modified condition together with condition (1) as stated in problem 322 implies the inequality ∣f(z)∣≤1\left|f(z)\right|\leq 1 in the domain bounded by Γ1\Gamma_{1} and Γ2\Gamma_{2}.

?
Problem 3.325

The function f(z)f(z) is assumed to be regular in the half-plane Re(z)≥0\mathrm{Re}\left(z\right)\geq 0 and to satisfy the following conditions:

  1. there exist two constants AA and BB, A>0A>0, B>0B>0 such that in the entire half-plane

    ∣f(z)∣≤AeB∣z∣; \left|f(z)\right| \leq Ae^{B\left|z\right|};
  2. we have for r≥0r\geq 0

    ∣f(ir)∣≤1,∣f(−ir)∣≤1; \left|f(ir)\right| \leq 1, \quad \left|f(-ir)\right| \leq 1;

\limsup _{r\to +\infty } \frac{\log \left|f(r)\right|}{r} \leq 0.

Then $f(z)$ is bounded by $1$ in the entire half-plane:

\left|f(z)\right| \leq 1 \quad \text{for} \quad \mathrm{Re}\left(z\right)\geq 0.

?
Problem 3.326

Let the function f(z)f(z) be regular in the half-plane Re(z)≥0\mathrm{Re}\left(z\right)\geq 0 and satisfy the following conditions:

  1. there exist two constants AA and BB, A>0A>0, B>0B>0, such that in the entire half-plane

    ∣f(z)∣<AeB∣z∣; \left|f(z)\right| < Ae^{B\left|z\right|};
  2. f(z)f(z) is bounded on the imaginary axis,

    ∣f(ir)∣≤1,∣f(−ir)∣≤1,r≥0; \left|f(ir)\right| \leq 1, \quad \left|f(-ir)\right| \leq 1, \quad r\geq 0;
  3. there exists an angle α\alpha, −π2<α<π2-\dfrac {\pi }{2}<\alpha <\dfrac {\pi }{2} such that

    lim⁡r→+∞log⁡∣f(reiα)∣r=−∞. \lim _{r\to +\infty } \frac{\log \left|f(re^{i\alpha })\right|}{r} = -\infty .

Such a function must vanish identically. [Examine the function eωzf(z)e^{\omega z}f(z), ω>0\omega >0.]

?
Problem 3.327

The function f(z)f(z) is supposed to be regular in the half-plane Re(z)≥0\mathrm{Re}\left(z\right)\geq 0 and to satisfy the conditions:

  1. there exist two constants AA and BB, A>0A>0, B>0B>0, such that on the entire half-plane

    ∣f(z)∣<AeB∣z∣; \left|f(z)\right| < Ae^{B\left|z\right|};
  2. there exist two constants CC and γ\gamma, C>0C>0, γ>0\gamma >0 such that for r≥0r\geq 0

    ∣f(±ir)∣≤Ce−γr. \left|f(\pm ir)\right| \leq Ce^{-\gamma r}.

A function satisfying these conditions must vanish identically. [Examine the function f(z)e−βzlog⁡(z+1)f(z)e^{-\beta z\log (z+1)}.]

?
Problem 3.328

The function sin⁡πz\sin \pi z is the smallest function that is analytic for Re(z)≥0\mathrm{Re}\left(z\right)\geq 0 and that vanishes at the points z=0,1,2,3,…z=0,1,2,3,\ldots More precisely, the following proposition holds:

We assume that the function f(z)f(z) is analytic in the half-plane Re(z)≥0\mathrm{Re}\left(z\right)\geq 0 and that it satisfies the conditions:

  1. there exist two constants A,BA,B, A>0A>0, B>0B>0, such that for Re(z)≥0\mathrm{Re}\left(z\right)\geq 0

    ∣f(z)∣<AeB∣z∣; \left|f(z)\right| < Ae^{B\left|z\right|};
  2. there exist two constants CC and γ\gamma, C>0C>0, γ>0\gamma >0 such that for r≥0r\geq 0

    ∣f(±ir)∣≤Ce(π−γ)r; \left|f(\pm ir)\right| \leq Ce^{(\pi -\gamma )r};
  3. f(z)f(z) has the zeros 0,1,2,…,n,…0,1,2,\ldots ,n,\ldots Such a function vanishes identically.

?
Problem 3.329

Let ω(x)\omega (x) be a positive function of the positive variable xx that increases with xx and tends to +∞+\infty as xx increases to +∞+\infty. A function f(z)f(z), regular in the half-plane Re(z)≥0\mathrm{Re}\left(z\right)\geq 0, that satisfies the inequality

∣f(z)∣≥eω(∣z∣)∣z∣forRe(z)≥0 \left|f(z)\right| \geq e^{\omega (\left|z\right|)\left|z\right|} \quad \text{for} \quad \mathrm{Re}\left(z\right)\geq 0

does not exist.

?
Problem 3.330

Suppose that the function f(z)f(z) is regular at any finite point of the sector α≤ϑ≤β\alpha \leq \vartheta \leq \beta, bounded by 11, ∣f(z)∣≤1\left|f(z)\right|\leq 1, on the two rays ϑ=α\vartheta =\alpha and ϑ=β\vartheta =\beta and that, furthermore, there exists a positive constant δ\delta such that

∣f(z)∣exp⁡(−∣z∣πβ−α−δ) \left|f(z)\right|\exp \left(-\left|z\right|^{\frac{\pi }{\beta -\alpha }-\delta }\right)

is bounded for α≤ϑ≤β\alpha \leq \vartheta \leq \beta. Then ∣f(z)∣≤1\left|f(z)\right|\leq 1 at every inner point of the sector α≤ϑ≤β\alpha \leq \vartheta \leq \beta. [Comparison function exp⁡(zπβ−α−σ)\exp \left(z^{\frac{\pi }{\beta -\alpha }-\sigma }\right).]

?
Problem 3.331

Let f(z)f(z) be regular in the sector α≤ϑ≤β\alpha \leq \vartheta \leq \beta. If ∣f(z)∣≤1\left|f(z)\right|\leq 1 on the bounding rays ϑ=α\vartheta =\alpha and ϑ=β\vartheta =\beta and if there exists for every ε>0\varepsilon >0 an r0r_{0} such that

∣f(reiϑ)∣<eerπβ−αforr>r0 \left|f(re^{i\vartheta })\right| < e^{e^{r^{\frac{\pi }{\beta -\alpha }}}} \quad \text{for} \quad r>r_{0}

in the above mentioned sector then the stronger inequality

∣f(z)∣≤1 \left|f(z)\right| \leq 1

holds in the entire sector. [Method of problem 325.]

?
Problem 3.332

The function g(z)g(z) is assumed to be an entire function, M(r)M(r) be the maximum of ∣g(z)∣\left|g(z)\right| on the circle ∣z∣=r\left|z\right|=r. If

lim⁡r→∞log⁡M(r)r=0 \lim _{r\to \infty } \frac{\log M(r)}{\sqrt{r}} = 0

then g(z)g(z) cannot be bounded along any ray. [E.g.g(z)g(z) is not bounded along the negative real axis.]

?
Problem 3.333

Suppose that the function f(z)f(z) is not a constant and that it is regular in the half-strip G\mathfrak {G} defined by the inequalities

x≥0,−π2≤y≤π2,z=x+iy. x\geq 0, \quad -\frac{\pi }{2}\leq y\leq \frac{\pi }{2}, \quad z=x+iy.

If there exist two constants AA and aa, A>0A>0, 0<a<10<a<1, such that in G\mathfrak {G}

∣f(x+iy)∣<eAeax, \left|f(x+iy)\right| < e^{Ae^{ax}},

and if

∣f(z)∣≤1 \left|f(z)\right| \leq 1

on the boundary of G\mathfrak {G} (i.e.for x=0x=0, −π2≤y≤π2-\dfrac {\pi }{2}\leq y\leq \dfrac {\pi }{2} and for x≥0x\geq 0, y=±π2y=\pm \dfrac {\pi }{2}) then f(z)f(z) satisfies the strict inequality

∣f(z)∣<1 \left|f(z)\right| < 1

in the interior of G\mathfrak {G}. [The comparison function is of the type eebze^{e^{bz}}.]

?
Problem 3.334

Let ω(x)\omega (x) have the same properties as in problem 329. Every function f(z)f(z) that is regular in the half-strip

x≥0,−π2≤y≤π2,z=x+iy x\geq 0, \quad -\frac{\pi }{2}\leq y\leq \frac{\pi }{2}, \quad z=x+iy

must satisfy the inequality

∣f(x+iy)∣<eω(x)ex \left|f(x+iy)\right| < e^{\omega (x)e^{x}}

at least at one point z=x+iyz=x+iy of the half-strip.

?
Problem 3.335

The assumptions of problem 278 are weakened insofar as (3) is satisfied in all but possibly finitely many boundary points z1,z2,…,znz_{1},z_{2},\ldots ,z_{n} of R\mathfrak {R}. An other assumption, however, is added, namely that there exists a positive number M′M' for which the inequality

∣f(z)∣<M′ \left|f(z)\right| < M'

holds everywhere in R\mathfrak {R}. (Only the case M′>MM'>M is interesting.) This modification of the hypothesis does not change the conclusion of problem 278 that under those conditions ∣f(z)∣≤M\left|f(z)\right|\leq M, ∣f(z)∣<M\left|f(z)\right|<M respectively. [In the case where the point at infinity belongs to R\mathfrak {R} and all the boundary points of R\mathfrak {R} lie in the disk ∣z∣<r\left|z\right|<r we examine the comparison function (2r)n∏ν=1n(z−zν)−1(2r)^{n}\displaystyle \prod_{\nu =1}^{n} (z-z_{\nu })^{-1}.]

?
Problem 3.336

The domain D\mathfrak {D} is supposed to lie in the half-plane Im(z)≥0\mathrm{Im}\left(z\right)\geq 0; the boundary contains a finite number of segments of the real axis; Ω\Omega denotes the sum of the angles under which the segments are seen from an inner point ζ\zeta of D\mathfrak {D}. Assume that f(z)f(z) is regular and single-valued in the interior of D\mathfrak {D} and continuous on the boundary of D\mathfrak {D}, that ∣f(z)∣≤A\left|f(z)\right|\leq A at the real boundary points, ∣f(z)∣≤a\left|f(z)\right|\leq a on the remaining boundary of D\mathfrak {D}, 0<a<A0<a<A. Then

∣f(ζ)∣≤AΩπa1−Ωπ.[Problem 57.] \left|f(\zeta )\right| \leq A^{\frac{\Omega }{\pi }}a^{1-\frac{\Omega }{\pi }}. \quad [\text{Problem 57}.]
?
Problem 3.337

The function f(z)f(z) is supposed to be regular on that piece of the Riemann surface of log⁡z\log z that covers the annulus 0<∣z∣≤10<\left|z\right|\leq 1. If f(z)f(z) is bounded and in particular if ∣f(z)∣≤1\left|f(z)\right|\leq 1 for ∣z∣=1\left|z\right|=1 then ∣f(z)∣≤1\left|f(z)\right|\leq 1 in the entire domain.

?
Problem 3.338

Let g(z)g(z) denote an entire function but not a constant; R\mathfrak {R} be a connected region on the boundary of which (more exactly: at whose boundary points different from z=∞z=\infty) ∣g(z)∣=k\left|g(z)\right|=k and in the interior of which ∣g(z)∣>k\left|g(z)\right|>k, k>0k>0. Then the point z=∞z=\infty is necessarily a boundary point of R\mathfrak {R} and g(z)g(z) is not bounded in R\mathfrak {R}.

?
Problem 3.339

Let Γ1\Gamma_{1} and Γ2\Gamma_{2} be two continuous curves that have a common starting point, extend to ∞\infty and enclose together with z=∞z=\infty a certain region R\mathfrak {R} (e.g.two rays enclosing a sector). We assume that no point of the negative real axis belongs to R\mathfrak {R}.

The function f(z)f(z) is supposed to be regular on Γ1,Γ2\Gamma_{1},\Gamma_{2} and in the enclosed region; in addition lim⁡f(z)=0\lim f(z)=0 as zz tends to ∞\infty along Γ1\Gamma_{1} and Γ2\Gamma_{2}. If f(z)f(z) is bounded in R\mathfrak {R} then lim⁡f(z)=0\lim f(z)=0 as zz goes to ∞\infty along any path in R\mathfrak {R}. [Examine log⁡zA+εlog⁡zf(z)\dfrac {\log z}{A+\varepsilon \log z}f(z).]

?
Problem 3.340

Let the curves Γ1\Gamma_{1} and Γ2\Gamma_{2} have the properties described in problem 339. Let f(z)f(z) be bounded and regular in the region between Γ1\Gamma_{1} and Γ2\Gamma_{2} and assume, in addition, that lim⁡f(z)=a\lim f(z)=a as zz tends to ∞\infty along Γ1\Gamma_{1} and lim⁡f(z)=b\lim f(z)=b as zz tends to ∞\infty along Γ2\Gamma_{2}. Then we have a=ba=b. [Consider (f(z)−a+b2)2−(a−b2)2\left(f(z)-\dfrac {a+b}{2}\right)^{2} - \left(\dfrac {a-b}{2}\right)^{2}.]

?