CC BY
26DOI: 10.15393/j3.art.20014.2689 Issues of Analysis. Vol. 3(21), No. 2, 2014
K. F. Amqzqva, E. G. Ganenkqya
ABOUT PLANAR (a, ft)-ACCESSIBLE DOMAINS1
Abstract. The article is devoted to the class Aof all (a, ¡3)— accessible with respect to the origin domains D, a, 3 € [0,1), possessing the property p = min where p € (0, is a fi-
p£&D
xed number. We find the maximal set of points a such that all domains D € Aare (y, 5)-accessible with respect to a, Y € [0; a], 5 € [0; 3]. This set is proved to be the closed disc of center 0 and radius p sin where p = min {a — y, 3 — 5} .
Keywords: a-accessible domain, (a, 3)-accessible domain, cone condition.
2010 Mathematical Subject Classification: 52A30, 03E15.
In [1] the notion of a-accessible domain was introduced. Let a E [0,1) be a fixed number. A domain D C Rn, 0 E D, is called a-accessible if for every point p E dD there exists a number r = r(p) > 0 such that the cone
( ( p \ an 1
K+(p, a, r) = < x E Rn : IIxN < r, [ x — p,^—^ > ^x — p\\ cos — >
I V m\J 2 j
is included in Rn\D.
In the case n = 2 these domains have been studied earlier by J. Stankie-wicz [2—4], D. A. Brannan and W. E. Kirwan [5], W. Ma and D. Minda [6], T. Sugawa [7] and others as a generalization of starlike domains. It was noted (see, for example, [7—10]) that in the planar case it is possible to consider domains, possessing a more general property. In [9, 10] such domains were called (a, [3)-accessible.
1 This work was supported by the Russian Foundation for Basic Research (project No. 14-01-00510) and by Program of Strategic Development of Petrozavodsk State University.
© Amozova K. F., Ganenkova E. G., 2014
Definition 1. [9], [10] Let a, ft £ [0,1), D c C, a £ D. A domain D is called (a, ft)-accessible with respect to a if for every point p £ dD there exists a number r = r(p) > 0 such that the cone
K + (p, a, a, ft, r) =
f ftn an 1
= < z £ C : —— < Arg(z — p) — Arg(p — a) < , \z — P\ < r >
is contained in C\D.
The notion of (a, ft)-accessible domain is a generalization of the notion of a-accessible domain. They are equal if a = ft.
In the article we choose values of arguments so that their difference belongs to (—n; n].
It was shown in [1] and [9, 10] that a- and (a, ft)-accessible domains satisfy the so-called "cone condition", i. e. such domains are also conically accessible from the interior.
The problem of characterization all domains with the "cone condition" is very hard. a- and (a, ft)-accessible domains are only special, but important cases of such domains. For a-accessible domains the axis of the symmetry of the cone is radial, for (a, ft)-accessible domains D the angle between this axis and the vector p — a, p £ dD, is fixed.
The following criterion of (a, ft)-accessibility is needed for the sequel.
Theorem A. [9, 10] Let D c C, dD be smooth, n(p) be an outward normal to the domain D at a point p £ dD, a, ft £ (0; 1). Then the domain D is (a, ft)-accessible with respect to the origin if and only if
(1 — ft )n . , ... (1 — a)n , .
< Arg(p) — Arg(n(p)) < -—(1)
for every p £ dD.
Let us note that Theorem A can be applied locally. In particular, if r is an open smooth curve, r c dD, a point p £ r, then from the proof of Theorem A it follows that the unbounded cone
K+ (p, 0, a, ft) = jz £ C : — < Arg(z — p) — Arg p < ^
is included in C\D.
Consider the class A^, containing all (a, fl)-accessible with respect to the origin domains D, possessing the property min \p\ = p, where
pedD
p e (0, to) is a fixed number. Let 7 e [0; a], S e [0; fl], D e Aa'3. By
iiD)S denote the set of all points a e D such that the domain D is (7, S)-accessible with respect to a.
In the present paper we find the maximal set, containing in WD6 for all domains D from A.
By B [0, R] denote the disc {z e C : \z\ < R}.
Theorem 1. If a, fl e [0; 1), 7 e [0; a], S e [0;fl] then
n WD6 = B
deAp'3
n • Pn 0, p sin —
. ,H 2 J
where p = min {a — y, fl — S} .
Proof. Let us show that the set P| WD is a disc. If D e Aap'3, then
DeA°p'3
a domain U(D), obtaining by a rotation of D with respect to the origin, also belongs to A^'3. Therefore the set U| WUd) , where the intersection
extends over all rotation transformations of D, is a disc with center at the origin. Hence,
n WD6 = n OU'6
U (D)
DeAp'3 U
is a disc too.
Let a,fl e [0; 1), D e Ap3. Fix p e dD. Since the domain D is (a,fl)-accessible with respect to the origin, then K+ (p, 0,a,fl,r) C C\D for some r > 0. Let us show that for all points a from the intersection of the domain D and the cone
K_(p, 0, a — y, fl — S) =
= {z e C : - < Arg(z - p) — Arg(—p) < }
the following inclusion holds
K+ (p, a,Y, S,r) C K+(p, 0,a,fl,r).
p
Take a G K-(p, 0,a — y, fi — S) H D, then (fi — S)n
— Arg(a — p) — Arg(—p) —
(a — y )n
Let ^ G K+ (p, a, y,S,t), z = p, this means that \z — p\ < r and
Sn Yn
—2 < Arg(z — p) — Arg(p — a) < -j-■
fifl.^
(2)
(3)
Figure 1: The domain D, case | Arg(z — p) — Argp\ > | Arg(p — a) — Argp\
If \ Arg(z — p) — Argp\ > \ Arg(p — a) — Argp\ (see fig. 1), then, by (2)
and (3),
fin Sn (fi — S)n
2
2
< Arg(z — p) — Arg p =
= (Arg(z — p) — Arg(p — a)) + (Arg(p — a) — Argp) < Yn (a — y )n an
- ~2 + 2 = T"'
and therefore z G K+(p, 0, a, fi, r). In the case
\ Arg(z — p) — Argp\ < \ Arg(p — a) — Argp\ (see fig. 2), we have z G K+ (p, 0,a — y, fi — S,r) and therefore z G K+(p, 0, a, fi, r).
Figure 2: The domain D, case \ Arg(z — p) — Argp\ < \ Arg(p — a) — Argp\
Consequently for both variants
K+ (p,a,Y,S,r) C K+(p, 0,a,fl,r) C C\D. Inscribe a disc with center at the origin into the set
K-(p, 0, a — - S) n D
and find its radius R(p). Denote by y a point from dK- (p, 0,a — y, fl — S) such that
\y\ = dist(0, dK- (p, 0, a — j,fl — S)).
Then, from the right triangle 0,y,p, we obtain R(p) = \y\ = \p\ sin P—,
2
where p = min{a — y, fl — S}. Put
R = min R(p) = p sin P—.
pedD 2
Then the disc B [0, R] is contained in K- (p, 0,a — Y,fl — S) n D for all p e dD. Thus B [0, R] c &Df for all D e Aa?.
Let us prove that it is impossible to enlarge the constant R in the last inclusion. For this aim we find a domain D0 e Ap? such that B [0, R] c
C &DS, but for every e > 0
B [0, R + e] £ Wd6 .
a) Let us begin with the case a, / £ (0; 1), y £ [0; a], S £ [0; /3]. Consider the simply connected domain D0, 0 £ D0, bounded by the logarithmic spirals
, n • j- (1 — a)n n
la (p) = pelLp e* tg-^-, 0 < p < 2,
, , • i (1 — n
lp(p) = pel* e— , -2 < p < 0,
and the circle
l(p) = pe2 tg ^ e^, -n<p < n, if a > / or l(p) = pe2 tg (1 2))T eiif, -n < p < n, if a < / (see fig. 3).
Figure 3: The domain Do, case a, ft £ (0; 1)
We will consider the case a > / only. The proof for a < / is analogous. Let us check that D0 £ Aa>P. Note that min \p\ = \la(0)| = \lp(0)| = p.
p pedD
Show that domain D0 is (a, 3)-accessible with respect to the origin. Take p £ dD0. Prove that the unbounded cone K+(p, 0,a,3) is contained in C\D0. Denote by n(p) the outward normal to the domain D0 at the point p if such a normal exists. Divide the proof into six cases.
a 1) Let p = la(p), p G (0; J) ■ Then
Argp — Arg n(p) = Argp — (Arg l'a(p) — Jj
(1-a)n i (1 — a)n\\ n
= p — arg (pe^e*tg ^ ^ + tg (1 f^) J + J =
1 n ( an) n (1 — a)n
+ - = — arctg tg— + - = -—.
1 n (
= p — p — arctg —(1-a)n + j = — arctg (tg
tg (i-2")n 1 2 - —&VU6 2/2 2
By Theorem A, for such p we have K+ (p, 0, a, fi) C C\D0■
a 2) If p = lp (p), p G (p0; 0), where p0 is the solution of the equation l(po) = lp (po )■ Then
Argp — Arg n(p) = Argp — (Arg l'p(p) — = *>-««! "^ " tg )) + 2 =
1 n (1 — fi )n
= p — p — n + arctg ¡gn-p + 2 =--— ■ (4)
Consequently, by Theorem A, K+(p, 0,a,fi) C C\D0■
(n 1
a 3) Let p = l(p), p G (—n; p0) U ( —; n ■ In this case
2
K+(p, 0, a, fi) C K+ (p, 0,1,1) C C\D0■
a 4) Consider
(n) (n) . n (i-q)n
p = l\ 2/ = H 2/ = pie 2
Since arg l'a = n + 02", arg l^"2) = n (here and below in a 5), a 6),
b3), b 5) we consider one-sided derivatives), and arga l (p), arg l (p) are monotone we get
K+(p, 0, a, fi) C K+ (p, 0,1,1) C C\D0■
an
a 5) Let p = la (0) = lp (0) = p■ Since arg l'a (0) = —— and arg lp (0) =
2
Bn
= n — -—, then K+ (p, 0, a, fi) C C\D0 and for all £i ,£2 > 0, £i2 + £22 = 0, 2
K+(p, 0, a + £1 ,fi + £2) H D0 =
Moreover, both rays of dK+ (p, 0,a + ei, fl + e2) intersect the domain Do.
n
a 6) The last case is p = l(po) = l?(po). Here arg l'(po) = po + — and
2
arg l? (po) = po + n — Thus,
K+(p, 0,a,ß) c K+ (p, 0,1,1) c C\Do.
Consequently K+ (p, 0,a,fl) C C\Do for all p e dDo. Therefore Do e
rz Aa,ß ^ p
Now we will show that B
in .
Do
0, p sin —
. ,H 2 J
is the maximal disc, contained
Let t E (0,1 - p). Fix
a* £ dB
0, p sin
(p + t)n
n K-(p, 0, p + t,p + t),
such that Im a* > 0 if p = ß — S and Im a* < 0 if p = a — j (see fig. 4).
Figure 4: Intersection Do and K+(p,a* ,j,S), case a, ß £ (0; 1)
2
Further we consider the case p = ft — S only, because the proof for the
a,
case p = a — 7 is analogous. Let us show that a* £ . By definition of
n
Arg p — Arg(p — a*) = (p + t)2. (5)
Denote by l the ray, consisting of points w £ dK+ (p, a* ,Y,S), w = p, such that
Sn
Arg(w — p) — Arg(p — a* ) = ——. (6) By (5) and (6), for every w £ l
Arg(w — p) — Arg(p) = (Arg(w — p) — Arg(p — a* )) + (Arg(p — a*) — Arg p) =
Sn (p + t)n (ft + t)n
2 2 2
Consequently, l is one of the ray of dK+ (p, 0,a, ft + t). As it was proved above (see a5)) for every t > 0
l n Do =
Therefore a* £ fi^o. Since t is arbitrary positive, we obtain that
B [0, p sin — 2
is the maximal disc, contained in .
' Do
b) Let a = 0, ft £ (0; 1), S £ [0;ft]. Consider the domain Do C C, 0 £ Do, bounded by the logarithmic spiral
, (1-SW n
lp(p) = pe^ , — 2 < p < 0,
the circle
l(p) = pe2 tg e1^. p £ (—n. — 2J
n 4. (1-^)n / ni
l(p)= pe 2 elif, p £ (—n, — U [0,n],
2
and the segment p; pe2 tg ( ^ (see fig. 5). Show that D0 £ A0'p.
p
Let us prove that domain D0 is (0, ft)-accessible with respect to the origin. Fix p £ dDo. Show that K+ (p, 0, 0, ft) C C\Do.
Figure 5: The domain Do, case a = 0, ft £ (0; 1)
b 1) If p = lp(p), p £ , then (4) is true (see a1)) and
K+(p, 0, 0,3) c K+ (p, 0, a, 3) C C\D0. b2) If p = l(p), p £ (—n; — 2) U (0; n) , then
K+ (p, 0, 0,3) C K+(p, 0,1,1) C C\D0.
b 3) Let p = lp (0) = p. In this case arg l^ (0) = n--. In addition
arg lp (p) is monotone. Hence, K+(p, 0, 0,3) C C\D0.
\ ( n t (1 — \ N
b 4) Consider the case p £ ( p; pe2 tg 2 j . By b3), we have
K+(p, 0, 0,3) C K+(p, 0, 0,3) C C\D0. b5) Let p = l (— 2) = lp (— 2) = —pie2tg (1 2>)n. Since
arg . (-|)=0andargl/ (-^^ - 1
and arg l/ (v) is monotone, we get
K+ (p, 0, 0,fl) C K+(p, 0,1,1) C C\Do.
Summarizing the proved above we obtain that K+ (p, 0, 0, fl) C C\D0 for all p G dD0. Since, in addition min \p\ = \l/(0)| = p, we conclude
pedDo
that Do G A03.
Let us check that B [0, 0] = {0} is the maximal disc, contained in • Suppose that for some r > 0
B [0, r] C q0dO •
Let z0 G B [0, r] and Im z0 < 0. Then, by construction of D0,
K+ (p,z0, 0, 0) n D0 = see fig. 3. Since K+(p,z0, 0, 0) C K+(p,z0, 0,5), then
K+(p,Z0, 0,5) n D0 =
Therefore, z0 G &D) • This contradiction shows that
B [0, r] g ftDO
for every r > 0.
c) In the case [ = 0, a G (0; 1), 7 G [0; a], we consider the domain D0 c C, 0 G D0, bounded by the logarithmic spiral
la(v) = peeV, 0 < v < ^,
the circle
n j- (1 — a)n
l(v)= pe2eiV, v G (-n, 0] U
7T
L2
and the segment p; pen tg ( 2 ) (see fig. 6).
Arguing as in case b), taking z0, Im z0 > 0, we prove that D0 G A and B [0, 0] is the maximal disc, containing in fi^j0.
a,0
p
Figure 6: The domain Do, case a € (0; 1), ( = 0
d) Let a = fl = 0. In this case, the class of (0, 0)-accessible domains coincides with the class of 0-accessible domains and the class of starlike with respect to the origin domains (see [11,12]).
Consider domain D0 = C\lp e A°p'°, where lp = {pt,t > 1}. Then the set W°D°0 consists of all points a e D0 such that K+ (p, a, 0, 0) C lp for every p = pr, t > 1. Consequently, W°D°0 = {pk, k < 1}. Therefore, for all £ > 0
B [0,£] ^ WD0 and n WD0 = {0}.
deAP'0
□
References
[1] Liczberski P., Starkov V. V. Domains in R" with conical accessible boundary. J. Math. Anal. Appl., 2013, vol. 408, no. 2, pp. 547-560.
[2] Stankiewicz J. Quelques problèmes extremaux dans les classes des fonctions a-angulairement etoilees. Ann. Univ. Mariae Curie-Sklodowska, Sectio A, 1966, vol. XX, pp. 59-75.
[3] Stankiewicz J. Some remarks concerning starlike functions. Bulletin de l'academie Polonaise des sciences. Serie des sciences math., astr. et phys., 1970, vol. XVIII, no. 3, pp. 143-146.
[4] Stankiewicz J. On a family of starlike functions. Ann. Univ. Mariae Curie-Sklodowska, Sectio A, 1968/1969/1970, vol. XXII/XXIII/XXIV, pp. 175181.
[5] Brannan D. A. and Kirwan W. E. On some classes of bounded univalent functions. J. London Math. Soc., 1969, vol. 2, no. 1, pp. 431-443.
[6] Ma W. and Minda D. An internal geometric characterization of strongly starlike functions. Ann. Univ. Mariae Curie-Sklodowska, Sectio A, 1991, vol. XLV, pp. 89-97.
[7] Sugawa T. A self-duality of strong starlikeness. Kodai Math. J., 2005, vol. 28, pp. 382-389. DOI: 10.2996%2Fkmj%2F1123767018.
[8] Lecko A. and Stankiewicz J. An internal geometric characterization of some subclasses of starlike functions. XVIth Rolf Nevanlinna Colloquium, Joensuu, 1995 (I. Laine and O. Martio, eds.), Walter de Gruyter, Berlin, 1997, pp. 231-238.
[9] Anikiev A. N. Plane domains with special cone condition. Russian Mathematics, 2014, vol. 58, no. 2, pp. 62-63. DOI: 10.3103/S1066369X14020108.
[10] Anikiev A. N. Plane domains with special cone condition. Issues of Analysis, 2014, vol. 3 (21), no. 1, pp. 16-31. DOI: 10.15393/j3.art.2014.2609.
[11] Amozova K. F., Starkov V. V. a-accessible domains, a nonsmooth case. Izv. Sarat. Univ. N.S. Ser. Math. Mech. Inform., 2013, vol. 13, no. 3, pp. 3-8.
[12] Amozova K. F. Sufficient conditions of a-accessibility of domain in nonsmooth case. Issues of Analysis, 2013, vol. 2 (20), no. 1, pp. 3-13.
Received September 3, 2014.
Petrozavodsk State University,
Lenin Avenue, 33, 185910 Petrozavodsk, Russia.
E-mail: amokira@rambler.ru, g_ek@inbox.ru
