A Sufficient Condition for the Convexity of the Area of an Isoptic Curve of an Oval.
M. MICHALSKA(*)
ABSTRACT - The problem of convexity of the area function of ana-isoptic of the convex curve C is considered.
1. Introduction.
An a-isoptic Ca of a plane, closed, convex curve C is a set of those points in the plane from which the curve C is seen under a fixed angle p2a. In [1] the authors discussed the problem of convexity of a-isoptic curves. We will study the problem of convexity of the area function.
Letp(t),t[ 0 , 2p] be a support function of the curveC. It is known [1], [2] that the area A(a) of a-isoptic of C is given by formula
A(a)4
0 2p 1
sin2a[p2(t)2p(t1a)(p(t) cosa1p.
(t) sina) ]dt. (1.1)
Letp(t)4a01
!
n41 Q
(ancosnt1bnsinnt) be the Fourier series of the sup- port function. We will assume that p(t)C2[ 0 , 2p] and p(t),p.
(t) L2[ 0 , 2p]. Moreover the curvature radiusR(t)4p(t)1p..
(t)D0 and the Fourier series of R(t)4a01
!
n41 Q
(12n2)(ancosnt1bnsinnt) is convergent.
(*) Indirizzo dell’A.: Institute of Mathematics, Maria Curie-SklCodowska Uni- versity, pl. M. Curie-SklCodowska 1, 20-031 Lublin, Poland; e-mail: daisyHgo- lem.umcs.lublin.pl
Let f(x)An
!
4 2Q Q
cneinx, g(x)An
!
4 2Q Q
dneinxL2 where c2n4cn are complex numbers then we have ([5], p. 37, theorem 1.12)
THEOREM 1.1. Suppose that f and g are in L2and have coefficients cn and dn respectively. Then
1 2p
0 2p
f(x2t)g(t)dt4n
!
4 2Q Q
cndneinx (1.2)
for all x, and the series on the right converges absolutely and uniformly.
In particular 1 2p
0 2p
f(x1t) f(t)dt4n
!
4 2Q Q
NcnN2einx, (1.3)
1 2p
0 2p
f(t)g(t)dt4n
!
4 2Q Q
cnd2n, (1.4)
1 2p
0 2p
Nf(t)N2dt4n4 2Q
!
Q NcnN2. (1.5)If f(t)4a01
!
n41 Q
(ancosnt1bnsinnt) is a real valued function and g4f8 then we have for all xR
0 2p
f2(t)dt42pa021p
!
n41 Q
(an21bn2) , (1.6)
0 2p
f(t1x) f(t)dt42pa021p
!
n41 Q
(an21bn2) cosnx, (1.7)
0 2p
f(t1x) f8(t)dt4p
!
n41 Q
n(an21bn2) sinnx (1.8)
which, with formula (1.1), give the following formula for the area of the a-isoptic
(1.9) A(a)4 4 1
sin2a
k
2pa021pn!
Q41(an21bn2)2cosag
2pa021pn!
4Q1(an21bn2) cosnah
22sinaQp
!
n41 Q
n(an21bn2) sinna
l
44 p
sin2a
k
2a02( 12cosa)1n!
4Q1(an21bn2)(12cosnacosa2nsinasinna)l
44 p
sin2a
k
2a02(12cosa)1n!
40 Q(an112 1bn112 )
g
11n2 cos (n12)a2n122 cosna
hl
.2. The main result.
THEOREM 2.1. Let p(t)4a01
!
n41 Q
(ancosnt1bnsinnt)be the Fouri- er series of the support function of C.If Fourier coefficients of p(t)are such that the inequality holds
a02F
!
n40 Q
(n(n12 ) )2(an112 1bn112 ) (2.1)
then A(a) is a convex function.
PROOF. We will show thatA9(t)F0 . From the above considerations we get
A8(a)
p 42a02 ( 12cosa)2 sin3a 1 1
!
n40 Q
(an2111bn211) 1
sin3a
y
n(n212 )sina( sinna2sin (n12 )a)222 cosa
g
11 n2 cos (n12 )a2n122 cosna
hl
and by some calculations (2.2) A9(a)4 p
sin4a
m
2a02( 12cosa)2( 22cosa)11
!
n40 Q
(an12 1
1bn12 1)[n3sin3asin (n11 )a13n2sin2acosna2 26nsinacosasinna12( 112 cos2a)( 12cosna) ]
n
.If the condition (2.1) holds, then A9(a)
p F 1
sin4a
!
n40 Q
(an2111bn211)]2n2(n12 )2( 12cosa)2( 22cosa)1 1n3sin3asin (n11 )a13n2sin2acosna26nsinacosasinna1 12( 112 cos2a)( 12cosna)(
4 1 sin4a
!
n40 Q
(an211
1bn211)Mn(a), where
(2.3) Mn(a)42n2(n12 )2( 12cosa)2( 22cosa)1n3sin3asin (n11 )a1 13n2sin2acosna26nsinacosasinna12( 112 cos2a)( 12cosna) for abbreviation. One can see that Mn( 0 )40 for nN andM0(a)40 . We will show that Mn8(a)F0 . We have
Mn8(a)42(n414n314n2) sina( 12cosa)( 523 cosa)1
1n4sin3acos (n11 )a14n3sin3acos (n11 )a1 18nsin2asinna28 sinacosa( 12cosna) . Ifa[ 0 , p) then sinaF0 andNsinnaN Gnsinaandn414n3F5n2for any positive integer n. Moreover, we have
Mn8(a)Fn4sina( 12cosa)[ 2( 523 cosa)2( 11cosa) ]1
14n3sina( 12cosa)[ 2( 523 cosa)2( 11cosa) ]1 18n2sina( 12cosa)( 523 cosa)2
28n2sin3a28 sinacosaQ2 sin2na 2 F F(n414n3) sina( 12cosa)[ 927 cosa]1
18n2sina( 12cosa)[ 523 cosa212cosa]2 216n2sinacosasin2 a
2 F5n2Q2 sinasin2 a
2( 927 cosa)216n2sinacosasin2 a 2 4 42n2sinasin2 a
2[ 45235 cosa28 cosa]4
42n2sinasin2 a
2( 45243 cosa)F0 .
The functionMn(a) is not decreasing for eacha[ 0 ,p) and Mn( 0 )40 . This means thatMn(a)F0 for eacha[ 0 ,p) and each positive integer n and M0(a)40 . This completes the proof of the inequality
A9(a)
p F 1
sin4a
!
n40 Q
(an2111bn211)Mn(a)F0 for each a[ 0 ,p). r
EXAMPLE 1. Let p(t)4a1bcos 3 t , a,bD0 then the curvature radiusR(t)4a28bcos 3t. If aD8b then the curve C is convex. From (2.1) the area function A(a) is convex if a2F( 8b)2.
EXAMPLE 2. Let p(t)4a1bcos 3t1ccos 5t then from (2.1) a2F F( 8b)21( 24c)2andR(t)4a28bcos 3t224ccos 5tD0 foraD8NbN 1 124NcN. We have
a2D( 8NbN 124NcN)2D( 8b)21( 24c)2.
For any convex curve C with support function p(t)4a1bcos 3t1 1ccos 5t the area function A(a) is convex.
EXAMPLE 3. Consider a support function p(t)4 1
12asint, t[ 0 , 2p). Then the curvature radius of the corresponding oval is R(t)4123asint12a
2
( 12asint)3 F0 for 0E NaN E12. We will find the Fourier series of p(t)4a01
!
n41 Q
(ancosnt1bnsinnt).
1
12asint4
!
n40 Q
(asint)n, where
sinmt4
. ` / `
´
(21 )k 22k
y
2!
j40 k21
g
2jkh
(21 )jcos 2(k2j)t1(21 )kg
2kkh z
for m42k(21 )k 22k
y !
j40
k
g
2k1j 1h
(21 )jsin ( 2k22j11 )tz
for m42k11 ,hence we have 1
12asint 4
!
k40 Q
a2k
g
2kkh
22k 1
!
k40 Q
(21 )k11cos 2(k11 )t
!
j4k Q a2j12
22j11
g
2jj21k2h
1
!
k40 Q
(21 )ksin ( 2k11 )t
!
j4k Q a2j11
22j
g
2jj21k1h
.The Fourier coefficients of p(t) are equal to
an4
. ` / `
´
k
!
40 Qa2k
g
2kkh
22k for n40 (21 )k11
!
j4k Q a2j12
22j11
g
2jj21k2h
for n42k120 for n42k11 ,
bn4
. /
´
0 for n42k (21 )k
!
j4k Q a2j11
22j
g
2jj21k1h
for n42k11 .Using the inequalities
k2pn
g
neh
ne12n11 1En!Ek2png
neh
ne12n1 ,g
11n1h
nEeEg
11 n1h
n112,we have
g
2jj1112h
22j11 E 2
kpkj11 G 2 kpkk11 ,
g
2jj1111h
22j E 2
kpkj G 2
kpkk , for j4k,k11 ,R
and
Na2k12N 4
!
j4k Q a2j12
22j11
g
2jj21k2h
Gj!
4k Q a2j1222j11
g
2jj1112h
Gj4k!
Q 2a2j12
k
p(k11 )4 2a2k12
k
p(k11 ) 1 12a2,Nb2k11N 4
!
j4k Q a2j11
22j
g
2jj21k1h
Gj4k!
Q a2j1122j
g
2jj1111h
GG
!
j4k Q 2a2j11
kpk 4 2a2k11 kpk
1 12a2 . We can estimate the right hand side in the formula (2.1)
n
!
40 Q(n(n12 ) )2(an211
1bn211
)4
4
!
k40 Q
( 2k11 )2( 2k13 )2a2k2121
!
k40 Q
( 2k)2( 2k12 )2b2k211E
E
!
k40 Q
( 2k11 )2( 2k13 )2
u k
2p(ka2k1211 ) 1 12a2
v
211
!
k40 Q
( 2k)2( 2k12 )2
g
2ak2kpk11 1 12a2h
2GG 1
p( 12a2)2
!
k40 Q
[
64k(k11 )2a4k1218( 2k11 )( 2k13 )2a4k14]
44 1
p( 12a2)2
!
k40 Q
[ (a4k15)R24(a4k14)9 23(a4k13)8 23a4k121
1(a4k17)R24(a4k16)9 23(a4k15)8 23a4k14]4
8a4( 924a21a4) p( 12a2)6 .
On the other hand
a024
u !
k40 Q
a2k
g
2kkh
22k
v
24u
11!
k41 Q
a2k
g
2kkh
22k
v
2D1 .Comparing both sides we obtain that if 0E NaN E
o
2kp2215p18p17223p then the condition (2.1) holds.PROPOSITION 2.1. Let p(t)be the support function of the curve C.If there existseD0such that fora[ 0 ,e]Fourier coefficients of p(t)sat- isfy inequalities
(2.4)
!
n40 Q
(an112 1bn112 )[n3sin3asin (n11 )a13n2sin2acosna2 26nsinacosasinna12( 112 cos2a)( 12cosna) ]F0 and
a02FC(e)
!
n40 Q
(n313n216n)(an2111bn211) , (2.5)
where C(e)42( 1 1
2cose)2( 22cose) then A(a) is a convex function.
PROOF. Ifa[0,e] then from the assumption (2.4) we haveA9(a)F0.
For a(e,p) we have
n3sin3asin (n11 )a13n2sin2acosna2
26nsinacosasinna12( 112 cos2a)( 12cosna)F 2n323n226n and if a0 satisfy (2.5) then we obtain
A9(a)
p 42a02( 12cosa)2( 22cosa)
sin4a 1
1 1 sin4a
!
n40 Q
(an211
1bn211)[n3sin3asin (n11 )a1
13n2sin2acosna26nsinacosasinna12( 112 cos2a)( 12cosna) ] F 1
sin4a
y
C(e)!
n40 Q
(n313n216n)(an211
1bn211
) 1 C(e)2 2
!
n40 Q
(an12 1
1bn12 1)(n313n216n)
l
F0 . rIn particular, if the Fourier coefficients of p(t) are such that )n0N (nDn0 an4bn40
then p(t) satisfy (2.4) for some e0D0 . In fact if we put Nn11(a)4 1
sin4a[n3sin3asin (n11 )a1
13n2sin2acosna26nsinacosasinna12( 112 cos2a)( 12cosna) ] in (2.2) we have lim
aK01
Nn11(a)4n
2(n12 )2
4 D0 for each 0EnGn0 and N1f0 .Nn11(a) is a continuous function and therefore existsenD0 such that Nn11(a)D0 fora[ 0 ,en] we definee04min]e1,e2,R,en0(. If for e0 conditon (2.5) holds then A(a) is a convex function.
The author would like to thank the Referee of this paper for his re- marks which improve this paper.
R E F E R E N C E S
[1] W. CIES´LAK- A. MIERNOWSKI - W. MOZGAWA, Isoptics of a Closed Strictly Convex Curve, Lect. Notes in Math., 1481(1991), pp. 28-35.
[2] W. CIES´LAK- A. MIERNOWSKI - W. MOZGAWA, Isoptics of a Closed Strictly Convex Curve II, Rend. Sem. Mat. Univ. Padova,96 (1996), pp. 37-49.
[3] H. GROEMER, Geometric applications of Fourier series and spherical har- monics, Encyclopedia of Mathematics and its Applications, Cambridge Univ.
Press, Cambridge, 1996.
[4] M. A. HURWITZ,Sur quelques applications géométriques des séries de Fouri- er, Ann. Sci. Ecole Norm. Sup., 19 (1902), pp. 357-408.
[5] A. ZYGMUND, Trigonometric series, Cambridge Univ. Press, vol. I, II, 1968.
Manoscritto pervenuto in redazione il 9 gennaio 2003.