Open Access
Issue
Wuhan Univ. J. Nat. Sci.
Volume 27, Number 4, August 2022
Page(s) 281 - 286
DOI https://doi.org/10.1051/wujns/2022274281
Published online 26 September 2022

© Wuhan University 2022

Licence Creative CommonsThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

0 Introduction and Main Results

The setting for this paper is n-dimensional Eucli-dean spaces Rn. Let K n denote the set of convex bodies (compact, convex subsets with non-empty interiors) in Euclidean space Rn. Ko n denotes the set of convex bodies containing the origin in their interiors. V(K) denotes the n-dimensional volume of a body K, B the standard unit ball, and V(B)=ωn. Let Sn-1 denote the unit sphere in Rn.

Blaschke[1] considered the classical width-integral of convex bodies first and Hadwiger[2] studied it further. In 1975 Lutwak[3] introduced the i-th width-integral of convex bodies. In 1977, Lutwak[4] generalized the i-th width-integral to the mixed width-integral of convex bodies. In 2016, Feng[5] gave the definitions of mixed width-integral and the general i-th width-integral of convex bodies. In 2017, Zhou[6] defined the general Lp-mixed width-integral of convex bodies. For the more results of the mixed width-integral of convex bodies, we refer the interested reader to Refs. [7-14].

In this paper, we first establish the Brunn-Minkowski type inequality for the i-th general Lp-mixed width-integral of convex bodies.

Theorem 1   Let M,N,K,LKon, τ[-1,1], p>0,M and N have similar general Lp-width,for KM, LN, n-p<i<n or MK, NL, i>n , we have

[ B p , i ( τ ) ( M + p N ) - B p , i ( τ ) ( K + p L ) ] p n - i [ B p , i ( τ ) ( M ) - B p , i ( τ ) ( K ) ] p n - i + [ B p , i ( τ ) ( N ) - B p , i ( τ ) ( L ) ] p n - i (1)

for KM,LN, 0i<n-p, we have

[ B p , i ( τ ) ( M + p N ) - B p , i ( τ ) ( K + p L ) ] p n - i [ B p , i ( τ ) ( M ) - B p , i ( τ ) ( K ) ] p n - i + [ B p , i ( τ ) ( N ) - B p , i ( τ ) ( L ) ] p n - i (2)

the equality holds in (1) or (2) if and only if K  and L have similar general Lp-width, and (Bp,i(τ)(M),Bp,i(τ)(K))=c(Bp,i(τ)(N),Bp,i(τ)(L)) , where c is a constant.

We also establish two cyclic inequalities for the differences of i-th general Lp-mixed width-integral of convex bodies.

Theorem 2   Let M, KKon , KM, τ[-1,1], p>0, M has constant general Lp-width, for 0i<j<kn, i,j,kR, we get

[ B p , j ( τ ) ( M ) - B p , j ( τ ) ( K ) ] k - i [ B p , i ( τ ) ( M ) - B p , i ( τ ) ( K ) ] k - j [ B p , k ( τ ) ( M ) - B p , k ( τ ) ( K ) ] j - i (3)

with equality if and only if K has constant general Lp-width.

Theorem 3   Let M,N,K,LKon, KM, LN, τ[-1,1], p>0, M and N have similar general Lp-width, for 0i<j<kn, i,j,kR, we can obtain

[ B p , j ( τ ) ( M , N ) - B p , j ( τ ) ( K , L ) ] k - i [ B p , i ( τ ) ( M , N ) - B p , i ( τ ) ( K , L ) ] k - j [ B p , k ( τ ) ( M , N ) - B p , k ( τ ) ( K , L ) ] j - i (4)

with equality if and only if (bp(τ)(M,u),bp(τ)(K,u))=c(bp(τ)(N,u),bp(τ)(L,u)) , where c is a constant, and K and L have similar general Lp -width.

Meanwhile, we establish a cyclic Brunn-Minkowski inequality for the i-th general Lp-mixed width-integral of convex bodies.

Theorem 4   Let K, LKon, p>0, τ[-1,1], if j<n-p and i<j<k, we can obtain

B p , j ( τ ) ( K + p L ) p n - j B p , i ( τ ) ( K ) p ( k - j ) ( k - i ) ( n - j ) B p , k ( τ ) ( K ) p ( j - i ) ( k - i ) ( n - j ) + B p , i ( τ ) ( L ) p ( k - j ) ( k - i ) ( n - j ) B p , k ( τ ) ( L ) p ( j - i ) ( k - i ) ( n - j ) (5)

with equality if and only if K and L both have constant general Lp-width. If n-p<j<n and ji<k, or j>n and ij<k, the inequality is reversed.

1 Preliminaries

1.1 Support Function and Firey Lp-Combination

If KKn, the support function, hK=h(K,): Rn(-,), is defined by[15,16]

h ( K , x ) = m a x { x y : y K } , x R n

where xy denotes the standard inner product of x and y.

For K,LKon, p1 and λ, μ0 (not both zero), the Firey Lp-combination λK+pμLKon of K and L is defined by [17]

h ( λ K + p μ L , ) p = λ h ( K , ) p + μ h ( L , ) p (6)

where the operation "+p" is called Firey addition and λK denotes the Firey scalar multiplication.

1.2 General Lp-Mixed Width-Integral of Order i

For τ[-1,1] and p>0, the general Lp-mixed width-integral Bp(τ)(K1,,Kn) of K1,,KnKon is defined by [6]

B p ( τ ) ( K 1 , , K n ) = 1 n S n - 1 b p ( τ ) ( K 1 , u ) b p ( τ ) ( K n , u ) d S ( u ) (7)

where bp(τ)(K,u)=(f1(τ)hp(K,u)+f2(τ)hp(K,-u))1p for any uSn-1, and f1(τ), f2(τ) are chosen as follows:

f 1 ( τ ) = ( 1 + τ ) 2 p ( 1 + τ ) 2 p + ( 1 - τ ) 2 p ,   f 2 ( τ ) = ( 1 - τ ) 2 p ( 1 + τ ) 2 p + ( 1 - τ ) 2 p

f 1 ( τ ) and f2(τ) satisfy

f 1 ( τ ) + f 2 ( τ ) = 1 ;   f 1 ( - τ ) = f 2 ( τ ) ; f 2 ( - τ ) = f 1 ( τ ) .

K and L are said to have similar general Lp-width if there exists a constant λ>0 such that bp(τ)(K,u)=λbp(τ)(L,u) for all uSn-1. If bp(τ)(K,u)=bp(τ)(L,u) for all uSn-1, then we call K and L have the same general Lp-width. If bp(τ)(K,u) is a constant for all uSn-1, we call K has the constant general Lp-width.

Taking K1==Kn-i=K and Kn-i+1==Kn=L in (7), the general Lp-mixed width-integral Bp,i(τ)(K,L) of K,LKon is given by

B p , i ( τ ) ( K , L ) = 1 n S n - 1 b p ( τ ) ( K , u ) n - i b p ( τ ) ( L , u ) i d S ( u ) (8)

Further, let L=B in (8), since bp(τ)(B,u)=1, and write Bp,i(τ)(K) for Bp,i(τ)(K,B), we get

B p , i ( τ ) ( K ) = 1 n S n - 1 b p ( τ ) ( K , u ) n - i d S ( u ) (9)

where Bp,i(τ)(K) is called the i-th general Lp-mixed width-integral of KKon. If i=0, we write Bp,0(τ)(K)=Bp(τ)(K), Bp(τ)(K) is called the general Lp-width-integral of KKon.

2 Proofs of Theorems

In this section, we give the proofs of the Theorems 1-4. The proof of Theorem 1 requires the following lemmas.

Lemma 1[6]If K,LKon, τ[-1,1] and p>0, for in-pjn and ij, then

( B p , i ( τ ) ( K + p L ) B p , j ( τ ) ( K + p L ) ) p j - i ( B p , i ( τ ) ( K ) B p , j ( τ ) ( K ) ) p j - i + ( B p , i ( τ ) ( L ) B p , j ( τ ) ( L ) ) p j - i (10)

for jnin-p and ij, then

( B p , i ( τ ) ( K + p L ) B p , j ( τ ) ( K + p L ) ) p j - i ( B p , i ( τ ) ( K ) B p , j ( τ ) ( K ) ) p j - i + ( B p , i ( τ ) ( L ) B p , j ( τ ) ( L ) ) p j - i (11)

with equality in every inequality if and only if K and L have similar general Lp-width.

Lemma 2[18,19]Let x=(x1,,xn) and y=(y1,,yn) be two series of non-negative real numbers, and x1pi=2nxip, y1pi=2nyip, for p>1, then

( x 1 p - i = 2 n x i p ) 1 p + ( y 1 p - i = 2 n y i p ) 1 p ( ( x 1 + y 1 ) p - i = 2 n ( x i + y i ) p ) 1 p (12)

for p<0 or 0<p<1,

( x 1 p - i = 2 n x i p ) 1 p + ( y 1 p - i = 2 n y i p ) 1 p ( ( x 1 + y 1 ) p - i = 2 n ( x i + y i ) p ) 1 p (13)

with equality in every inequality if and only if x and y are proportional.

Proof of Theorem 1   Let j=n in Lemma 1, since Bp,n(τ)(K+pL)=ωn is a constant, for in-p, we immediately obtain

B p , i ( τ ) ( K + p L ) p n - i B p , i ( τ ) ( K ) p n - i + B p , L ( τ ) ( L ) p n - i (14)

and for n-p<i<n,

B p , i ( τ ) ( K + p L ) p n - i B p , i ( τ ) ( K ) p n - i + B p , L ( τ ) ( L ) p n - i (15)

the equality holds if and only if K and L have similar general Lp-width, which is just Corollary 3.

Since M, N, K, LKon, M and N have similar general Lp-width, for n-p<i<n, we have

B p , i ( τ ) ( K + p L ) p n - i B p , i ( τ ) ( K ) p n - i + B p , i ( τ ) ( L ) p n - i (16)

B p , i ( τ ) ( M + p N ) p n - i = B p , i ( τ ) ( M ) p n - i + B p , i ( τ ) ( N ) p n - i (17)

By the definition of the i-th general Lp-mixed width-integral of convex bodies, we know that Bp,i(τ)(K)Bp,i(τ)(M) and Bp,i(τ)(L)Bp,i(τ)(N) if KM, LN for n-p<i<n; Bp,i(τ)(M)>Bp,i(τ)(K) and Bp,i(τ)(N)>Bp,i(τ)(L) if MK, NL for i>n. According to (13), combining (16) with (17), we can obtain

[ B p , i ( τ ) ( M + p N ) - B p , i ( τ ) ( K + p L ) ] p n - i [ ( B p , i ( τ ) ( M ) p n - i + B p , i ( τ ) ( N ) p n - i ) n - i p - ( B p , i ( τ ) ( K ) p n - i + B p , i ( τ ) ( L ) p n - i ) n - i p ] p n - i

[ B p , i ( τ ) ( M ) - B p , i ( τ ) ( K ) ] p n - i + [ B p , i ( τ ) ( N ) - B p , i ( τ ) ( L ) ] p n - i (18)

the equality holds if and only if (Bp,i(τ)(M),Bp,i(τ)(K)) is proportional to (Bp,i(τ)(N),Bp,i(τ)(L)), and K and L have similar general Lp-width. The proof of inequality (2) is similar. This proves the theorem.

The proof of Theorem 2-3 requires the following lemma.

Lemma 3[20]Suppose that fi , gi (i= 1,2) are non-negative continuous functions on Sn-1 such that Sn-1f1p(ξ)dξSn-1f2p(ξ)dξ and Sn-1g1q(ξ)dξSn-1g2q(ξ)dξ for p>1, 1p+1q=1 , and for all ξSn-1, f1p(ξ)=λg1q(ξ) where λ is a constant, then

( S n - 1 ( f 1 p ( ξ ) - f 2 p ( ξ ) ) d ξ ) 1 p ( S n - 1 ( g 1 q ( ξ ) - g 2 q ( ξ ) ) d ξ ) 1 q S n - 1 ( f 1 ( ξ ) g 1 ( ξ ) - f 2 ( ξ ) g 2 ( ξ ) ) d ξ (19)

with equality if and only if f2p(ξ)=λg2q(ξ) for any ξSn-1.

Proof of Theorem 2   Suppose that 0i<j<kn, let

f 1 = ( b p ( τ ) ( M , u ) n - k ) 1 λ , f 2 = ( b p ( τ ) ( K , u ) n - k ) 1 λ ,   g 1 = ( b p ( τ ) ( M , u ) n - i ) 1 μ , g 2 = ( b p ( τ ) ( K , u ) n - i ) 1 μ ,   λ = ( k - i ) / ( j - i ) , μ = ( k - i ) / ( k - j )

in Lemma 3, since KM, M has constant general Lp-width, f1λ/g1μ=bp(τ)(M,u))i-k is a constant, according to Lemma 3, we can get the inequality (3).

By Lemma 3, the equality in (3) holds if and only if bp(τ)(K,u))i-k=bp(τ)(M,u))i-k for all uSn-1, that means K has constant general Lp-width. This proves the theorem.

Proof of Theorem 3   We can prove Theorem 3 by Lemma 3 as well. Suppose that 0i<j<kn,λ=(k-i)/(j-i) and μ=(k-i)/(k-j), let

f 1 = ( b p ( τ ) ( M , u ) n - k b p ( τ ) ( N , u ) k ) 1 λ ,   f 2 = ( b p ( τ ) ( K , u ) n - k b p ( τ ) ( L , u ) k ) 1 λ ,   g 1 = ( b p ( τ ) ( M , u ) n - i b p ( τ ) ( N , u ) i ) 1 μ ,   g 2 = ( b p ( τ ) ( K , u ) n - i b p ( τ ) ( L , u ) i ) 1 μ

in Lemma 3, since KM, LN, M and N have similar general Lp-width, f1λ/g1μ=(bp(τ)(N,u)/bp(τ)(M,u))k-i is a constant. We get the consequence.

By Lemma 3, the equality in (4) holds if and only if (bp(τ)(L,u)/bp(τ)(K,u))k-i=(bp(τ)(N,u)/bp(τ)(M,u))k-i for all uSn-1, that means K and L have similar general Lp-width. This proves the theorem.

Taking i=0, j=1, k=n in Theorem 3, we obtain

Corollary 1   Let M,N,K,LKon, τ[-1,1], p>0, M and N have similar general Lp-width, for KM, LN, one gets

[ B p , 1 ( τ ) ( M , N ) - B p , 1 ( τ ) ( K , L ) ] n [ B p ( τ ) ( M ) - B p ( τ ) ( K ) ] n - 1 [ B p ( τ ) ( N ) - B p ( τ ) ( L ) ] (20)

with equality if and only if (bp(τ)(M,u),bp(τ)(K,u))=c(bp(τ)(N,u),bp(τ)(L,u)) , where c is a constant.

Proof of Theorem 4   If j<n-p, combined (6) with (9), according to the Minkowski's inequality[21], it follows that

B p , j ( τ ) ( K + p L ) = 1 n S n - 1 b p ( τ ) ( K + p L , u ) n - j d S ( u ) = 1 n S n - 1 ( f 1 ( τ ) h p ( K + p L , u ) + f 2 ( τ ) h p ( K + p L , - u ) ) n - j p d S ( u ) = 1 n S n - 1 ( f 1 ( τ ) h p ( K , u ) + f 2 ( τ ) h p ( K , - u ) + f 1 ( τ ) h p ( L , u ) + f 2 ( τ ) h p ( L , - u ) ) n - j p d S ( u ) = 1 n S n - 1 ( b p ( τ ) ( K , u ) p + b p ( τ ) ( L , u ) p ) n - j p d S ( u ) = 1 n S n - 1 ( b p ( τ ) ( K , u ) p ( k - j ) ( n - i ) ( k - i ) ( n - j ) b p ( τ ) ( K , u ) p ( j - i ) ( n - k ) ( k - i ) ( n - j ) + b p ( τ ) ( L , u ) p ( k - j ) ( n - i ) ( k - i ) ( n - j ) b p ( τ ) ( L , u ) p ( j - i ) ( n - k ) ( k - i ) ( n - j ) ) n - j p d S ( u ) [ ( 1 n S n - 1 b p ( τ ) ( K , u ) ( k - j ) ( n - i ) k - i b p ( τ ) ( K , u ) ( j - i ) ( n - k ) k - i d S ( u ) ) p n - j + ( 1 n S n - 1 b p ( τ ) ( L , u ) ( k - j ) ( n - i ) k - i b p ( τ ) ( L , u ) ( j - i ) ( n - k ) k - i d S ( u ) ) p n - j ] n - j p (21)

Since i<j<k means k-ik-j>1, using Hölder's inequality [22] we have

1 n S n - 1 b p ( τ ) ( K , u ) ( k - j ) ( n - i ) k - i b p ( τ ) ( K , u ) ( j - i ) ( n - k ) k - i d S ( u )

( 1 n S n - 1 ( b p ( τ ) ( K , u ) ( k - j ) ( n - i ) k - i ) k - i k - j d S ( u ) ) k - j k - i × ( 1 n S n - 1 ( b p ( τ ) ( K , u ) ( j - i ) ( n - k ) k - i ) k - i j - i d S ( u ) ) j - i k - i = B p , i ( τ ) ( K ) k - j k - i B p , k ( τ ) ( K ) j - i k - i (22)

Hence, we can get the following inequality

[ 1 n S n - 1 b p ( τ ) ( K , u ) ( k - j ) ( n - i ) k - i b p ( τ ) ( K , u ) ( j - i ) ( n - k ) k - i d S ( u ) ] p n - j B p , i ( τ ) ( K ) p ( k - j ) ( n - j ) ( k - i ) B p , k ( τ ) ( K ) p ( j - i ) ( n - j ) ( k - i ) (23)

Similarly, we can also obtain

[ 1 n S n - 1 b p ( τ ) ( L , u ) ( k - j ) ( n - i ) k - i b p ( τ ) ( L , u ) ( j - i ) ( n - k ) k - i d S ( u ) ] p n - j B p , i ( τ ) ( L ) p ( k - j ) ( n - j ) ( k - i ) B p , k ( τ ) ( L ) p ( j - i ) ( n - j ) ( k - i ) (24)

By (21), (23) and (24), we get

B p , j ( τ ) ( K + p L ) p n - j B p , i ( τ ) ( K ) p ( k - j ) ( k - i ) ( n - j ) B p , k ( τ ) ( K ) p ( j - i ) ( k - i ) ( n - j ) + B p , i ( τ ) ( L ) p ( k - j ) ( k - i ) ( n - j ) B p , k ( τ ) ( L ) p ( j - i ) ( k - i ) ( n - j ) (25)

From the equality condition of the Minkowski's inequality, we see that the equality (21) holds if and only if K and L have similar general Lp-width. By the equality conditions of Hölder's inequality, equality holds in (22) if and only if K has constant general Lp-width. Similary, the equality holds in (24) if and only if L has constant general Lp-width. Thus, the equality holds in (5) or its reverse if and only if both K and L have constant general Lp-width.

In particular, take L={o} in Theorem 4. Since K +{o}=K, and notice that Bp,i(τ)({o})=0, by inequality (5), we can obtain the following Corollary.

Corollary 2   Let KKon, p>0, τ[-1,1], for i<j<k, then

B p , j ( τ ) ( K ) k - i B p , i ( τ ) ( K ) k - j B p , k ( τ ) ( K ) j - i (26)

with equality if and only if K has constant general Lp-width.

Let i=j in Theorem 4, we may obtain the following Brunn-Minkowski inequality for the i-th general Lp-mixed width-integral.

Corollary 3   For K,LKon, p>0, τ[-1,1], in, if in-p, then

B p , i ( τ ) ( K + p L ) p n - i B p , i ( τ ) ( K ) p n - i + B p , L ( τ ) ( L ) p n - i (27)

with equality if and only if K and L have simliar general Lp-width. If n-p<i<n, or i>n, inequality (27) is reversed.

References

  1. Blaschke W. Vorlesungen über Integral Geometric I, II [M]. New York: Chelsea, 1949. [Google Scholar]
  2. Hadwiger H. Vorlesungen Über Inhalt, Oberflӓche und Isoperimetrie [M]. Berlin: Springer-Verlag, 1957. [Google Scholar]
  3. Lutwak E. Width-integrals of convex bodies [J]. Proc Amer Math Soc, 1975, 53(2): 435-439. [CrossRef] [MathSciNet] [Google Scholar]
  4. Lutwak E. Mixed width-integrals of convex bodies [J]. Israel J Math, 1977, 28(3): 249-253. [CrossRef] [MathSciNet] [Google Scholar]
  5. Feng Y B. General mixed width-integral of convex bodies [J]. Journal of Nonlinear Sciences and Applications, 2016, 9(6): 4226-4234. [CrossRef] [Google Scholar]
  6. Zhou Y P. General Formula -mixed width-integral of convex bodies and related inequalities [J]. Journal of Nonlinear Sciences and Applications, 2017, 10(8): 4372-4380. [CrossRef] [Google Scholar]
  7. Li X Y, Zhao C J. On the Formula -mixed affine surface area [J]. Math Inequal Appl, 2014, 17: 443-450. [MathSciNet] [Google Scholar]
  8. Lutwak E, Yang D, Zhang G Y. Formula -affine isoperimetric inequalities [J].Journal of Differential Geom, 2000, 56(1): 111-132. [MathSciNet] [Google Scholar]
  9. Wang W D, Ma T Y. Asymmetric Formula -difference bodies [J]. Proc Amer Math Soc, 2014, 142(7): 2517-2527. [CrossRef] [MathSciNet] [Google Scholar]
  10. Wang W D, Feng Y B. A general Formula -version of Petty's affine projection inequality [J]. Taiwan J Math, 2013, 17(2): 517-528. [Google Scholar]
  11. Wang W D, Wan X Y. Shephard type problems for general Formula -projection bodies [J]. Taiwan J Math, 2012, 16(5): 1749-1762. [Google Scholar]
  12. Leng G S, Zhao C J, He B W, et al. Inequalities for polars of mixed projection bodies [J]. Sci China Ser A, 2004, 47(2): 175-186(Ch). [CrossRef] [Google Scholar]
  13. Lutwak E. Dual mixed volumes [J]. Pacific J Math, 1975, 58(2): 531-538. [CrossRef] [MathSciNet] [Google Scholar]
  14. Lutwak E. Inequalities for mixed projection bodies [J]. Trans Amer Math Soc, 1993, 339(2): 901-916. [CrossRef] [MathSciNet] [Google Scholar]
  15. Gardner R J. Geometric Tomography [M]. 2nd Edition. Cambridge: Encyclopedia Math Appl Cambridge University Press, 2006. [CrossRef] [Google Scholar]
  16. Schneider R. Convex Bodies: The Brunn-Minkowski Theory [M]. 2nd Edition. Cambridge: Encyclopedia Math Appl Cambridge University Press, 2014. [Google Scholar]
  17. Lutwak E. The Brunn-Minkowski-Firey theory I: Mixed volumes and the Minkowski problem [J]. Differential Geom, 1993, 38(1): 131-150. [MathSciNet] [Google Scholar]
  18. Beckenbach E F, Bellman R. Inequalities [M]. 2nd Edition. Berlin: Springer-Verlag , 1965. [CrossRef] [Google Scholar]
  19. Losonczi L, Páles Z S. Inequalities for indefinite forms [J]. Journal of Math Anal Appl, 1997, 205(1): 148-156. [CrossRef] [MathSciNet] [Google Scholar]
  20. Lv S J. Dual Brunn-Minkowski inequality for volume differences [J]. Geom Dedicata, 2010, 145(1): 169-180. [CrossRef] [MathSciNet] [Google Scholar]
  21. Hardy G H, Littlewood J E, Pölya G. Inequalities [M]. New York: Cambridge University Press, 1934. [Google Scholar]
  22. Bechenbach E F, Bellman R. Inequalities [M]. Berlin: Springer-Verlag , 1961. [CrossRef] [Google Scholar]

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.