1 Introductions and main results
The notion of curvature is one of the central concepts of differential geometry; it also plays a key role in physics. The magnitude of a force required to move an object at constant speed along a curved path is, according to Newton’s laws, a constant multiple of the curvature of the trajectory. The motion of a body in a gravitational field is determined, according to Einstein, by the curvature of space-time. All sorts of shapes, from soap bubbles to red blood cells, seem to be determined by various curvatures.
A geometric inequality describes the relation among the invariants of geometric object in space. The isoperimetric inequality is one of the most ancient and beautiful geometric inequality. It states: For a closed curve γ in the Euclidean plane \(\mathbb{R}^{2}\), its length L and area A enclosed by γ satisfies
where equality holds if and only if γ is a circle. It follows that the circle is the only curve of constant length L enclosing a maximum area. One can find more stronger isoperimetric inequalities in [1‐15].
$$\begin{aligned} L^{2}-4\pi A\geq 0, \end{aligned}$$
(1.1)
Anzeige
There are many well-known inequalities or equalities about curvature, such as the Ros theorem, the Fenchel theorem, and the Willmore theorem. There are also many interesting results involving integrals of curvature for a plane curve, one can refer to [16‐19].
In [18], Pan and Zhang established a reverse isoperimetric inequality for the plane curves under some assumptions on curvature. If γ is a closed strictly convex curve in the Euclidean plane \(\mathbb{R}^{2}\) with length L and enclosing an area A, then
where \(\tilde{A}_{1}\) denotes the oriented area of the domain enclosed by the locus of curvature centers of γ, and the equality holds if and only if γ is a circle.
$$\begin{aligned} L^{2}-4\pi A\leq 4\pi \vert \tilde{A}_{1} \vert , \end{aligned}$$
(1.2)
In [20], Pan and Xu posed the following question: Does there exist a best constant C such that
with equality if and only if γ is a circle?
$$\begin{aligned} L^{2}-4\pi A\leq C \vert \tilde{A}_{1} \vert , \end{aligned}$$
In this paper, we give a new method to prove the reverse isoperimetric inequality (1.2). Meanwhile, we give an affirmative answer to Pan and Xu’s question.
Anzeige
Theorem 1
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve with length
L
in the Euclidean plane
\(\mathbb{R}^{2}\). Let
A
and
\(\tilde{A}_{1}\)
denote the area of the domain enclosed by
γ
and the oriented area of the domain enclosed by the locus of curvature centers of
γ, respectively. Then, for any
\(\varepsilon >0\),
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} L^{2}-4\pi A\leq (1+\varepsilon )\pi \vert \tilde{A}_{1} \vert , \end{aligned}$$
Let D be a domain bounded by γ, and the width \(w(\theta )\) of D in direction \(u(\theta )=(\cos \theta , \sin \theta )\) is defined to be the distance between two tangents to γ be perpendicular to \(u(\theta )\). It is clearly that
$$\begin{aligned} w(\theta )=p(\theta )+p(\theta +\pi ). \end{aligned}$$
(1.3)
If the point has a distance of \(w(\theta )/2\) from γ along the unit inward normal vector, we call it as the width center of γ. Let ζ denote the locus of width centers of γ, then \(\zeta (\theta )\) can be given by
where \(N(\theta )= (-\cos \theta , -\sin \theta )\) is the unit inward normal vector along γ.
$$\begin{aligned} \zeta (\theta ) =\gamma (\theta )+\frac{w(\theta )}{2}N(\theta ) , \end{aligned}$$
(1.4)
In Sect. 4, we research the domain enclosed by the locus of width centers instead of the domain enclosed by the locus of curvature centers of closed strictly convex curve γ, and establish the following reverse isoperimetric inequality.
Theorem 2
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve with length
L
in the Euclidean plane
\(\mathbb{R}^{2}\). Let
A
denote the area of the domain enclosed by
γ. Then, for any
\(\varepsilon >0\),
The equality holds if and only if
γ
is a circle. Here
\(\tilde{A}_{1}\)
denotes the oriented area of the domain enclosed by the locus of curvature centers of
γ
and
\(\tilde{A}_{2}\)
denotes the oriented area of the domain enclosed by
ζ.
$$\begin{aligned} L^{2}-4\pi A\leq \frac{3+\varepsilon }{2}\pi \bigl( \vert \tilde{A}_{1} \vert - \vert \tilde{A}_{2} \vert \bigr). \end{aligned}$$
2 Preliminaries
In this section, we recall some facts about Minkowski’s support function and Fourier series. Without loss of generality, suppose γ is a smooth regular positively oriented and closed strictly convex curve in the Euclidean plane. Take a point O inside γ as the origin of our frame. Let p be the oriented perpendicular distance from O to the tangent line at a point on γ, and θ the oriented angle from the positive \(x_{1}\)-axis to this perpendicular ray. Clearly, p is a single-valued periodic function of θ with period 2π and γ can be parameterized in terms of θ and \(p(\theta )\) as follows:
$$\begin{aligned} \gamma (\theta )= \bigl(x_{1}(\theta ), x_{2}(\theta ) \bigr)= \bigl(p( \theta )\cos \theta - p'( \theta )\sin \theta , p(\theta )\sin \theta + p'(\theta )\cos \theta \bigr). \end{aligned}$$
(2.1)
The couple \((\theta , p(\theta ))\) is usually called the polar tangential coordinate on γ, and \(p(\theta )\) is the support function of γ.
Then the curvature κ of γ can be calculated,
The radius of the curvature ρ of γ is given by
$$\begin{aligned} \kappa (\theta )=\frac{d\theta }{ds}= \frac{1}{p(\theta )+p''(\theta )}>0. \end{aligned}$$
$$\begin{aligned} \rho (\theta )=\frac{ds}{d\theta }=p(\theta )+p''( \theta ). \end{aligned}$$
Since the support function of a given convex curve γ is always continuous, bounded and 2π-periodic, it has the following Fourier series:
where
for \(n\geq 1\).
$$\begin{aligned} p(\theta )=\frac{1}{2}a_{0}+\sum _{n=1}^{\infty }(a_{n}\cos n\theta + b _{n}\sin n\theta ), \end{aligned}$$
(2.2)
$$\begin{aligned}& a_{0} = \frac{1}{\pi } \int _{0}^{2\pi }p(\theta )\,d\theta , \\& a_{n} = \frac{1}{\pi } \int _{0}^{2\pi }p(\theta )\cos n\theta \,d\theta , \\& b_{n} = \frac{1}{\pi } \int _{0}^{2\pi }p(\theta )\sin n\theta \,d\theta , \end{aligned}$$
Then we have
$$\begin{aligned}& p'(\theta ) = \sum_{n=1}^{\infty }(-na_{n} \sin n\theta +n b_{n}\cos n \theta ), \end{aligned}$$
(2.3)
$$\begin{aligned}& p''(\theta ) = \sum _{n=1}^{\infty } \bigl[-n^{2}(a_{n}\cos n\theta + b_{n} \sin n\theta ) \bigr]. \end{aligned}$$
(2.4)
Let L be the length of γ, and A be the area of the domain enclosed by γ, then
$$\begin{aligned}& L= \int _{\gamma }\,ds = \int _{0}^{2\pi } \bigl(p(\theta )+p''( \theta ) \bigr)\,d\theta = \int _{0}^{2\pi }p(\theta )\,d\theta , \end{aligned}$$
(2.5)
$$\begin{aligned}& A=\frac{1}{2} \int _{\gamma }p(\theta )\,ds =\frac{1}{2} \int _{0}^{2 \pi } \bigl(p^{2}(\theta )-p^{\prime 2}(\theta ) \bigr)\,d\theta , \end{aligned}$$
(2.6)
$$\begin{aligned}& \int _{\gamma }\frac{1}{\kappa }\,ds= \int _{0}^{2\pi }\rho ^{2}\,d\theta = \int _{0}^{2\pi } \bigl(p(\theta )+p''( \theta ) \bigr)^{2}\,d\theta . \end{aligned}$$
(2.7)
By (2.2), (2.3), (2.4) and the Parseval equality, the above inequalities (2.5), (2.6), (2.7) can be, respectively, rewritten as
$$\begin{aligned}& L=\pi a_{0}, \end{aligned}$$
(2.8)
$$\begin{aligned}& A=\frac{1}{4}\pi a^{2}_{0} + \frac{1}{2}\pi \sum_{n=1}^{\infty } \bigl(1-n ^{2} \bigr) \bigl(a_{n}^{2}+b_{n}^{2} \bigr), \end{aligned}$$
(2.9)
$$\begin{aligned}& \int _{\gamma }\frac{1}{\kappa }\,ds=\frac{1}{2}\pi a^{2}_{0} + \pi \sum_{n=1}^{\infty } \bigl(1-n^{2} \bigr)^{2} \bigl(a^{2}_{n} + b^{2}_{n} \bigr). \end{aligned}$$
(2.10)
3 Formulas of curvature
In this section, we will give a new proof for the inequalities about curvature which obtained in [19] and [18]. Meanwhile, we obtain some new reverse isoperimetric inequalities for convex plane curves.
Lemma 1
([19])
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane
\(\mathbb{R}^{2}\)
with the length
L
and enclosing a domain
D
of area
A. Let
κ
be the curvature of
γ, then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} \frac{L^{2}}{2\pi }\leq \int _{\gamma }\frac{1}{\kappa }\,ds, \end{aligned}$$
Proof
By (2.8) and (2.10) we can easily get
where equality holds if and only if \(a_{n}=b_{n}=0\) for \(n>1\), from (2.2), this shows \(p(\theta )=a_{0}/2+a_{1}\cos \theta +b_{1} \sin \theta \), which together with (2.1) implies that γ is a circle with center \((a_{1}, b_{1})\) and radius of \(a_{0}/2\). □
$$\begin{aligned} L^{2} =& \pi ^{2} a^{2}_{0} \\ \leq & 2\pi \Biggl(\frac{1}{2}\pi a^{2}_{0} + \pi \sum_{n=1}^{\infty } \bigl(1-n^{2} \bigr)^{2} \bigl(a^{2}_{n} +b^{2}_{n} \bigr) \Biggr) \\ =&2\pi \int _{\gamma }\frac{1}{\kappa }\,ds, \end{aligned}$$
Lemma 2
([19])
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane
\(\mathbb{R}^{2}\)
enclosing a domain
D
of area
A. Let
κ
be the curvature of
γ, then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} 2A\leq \int _{\gamma }\frac{1}{\kappa }\,ds, \end{aligned}$$
Proof
By (2.10) and (2.9) we have
where equality holds if and only if \(a_{n}=b_{n}=0\) for \(n>1\). By the same argument as in the proof of Lemma 1, we conclude that γ is a circle. □
$$\begin{aligned} \int _{\gamma }\frac{1}{\kappa }\,ds =&\frac{1}{2}\pi a^{2}_{0} +\pi \sum_{n=1}^{\infty } \bigl(1-n^{2} \bigr)^{2} \bigl(a^{2}_{n} + b^{2}_{n} \bigr) \\ =&\frac{1}{2}\pi a^{2}_{0}+\pi \sum _{n=1}^{\infty } \bigl(1-n^{2} \bigr) \bigl(a^{2} _{n} + b^{2}_{n} \bigr)+\pi \sum_{n=1}^{\infty }n^{2} \bigl(n^{2}-1 \bigr) \bigl(a^{2}_{n} + b ^{2}_{n} \bigr) \\ \geq &\frac{1}{2}\pi a^{2}_{0} + \pi \sum _{n=1}^{\infty } \bigl(1-n^{2} \bigr) \bigl(a ^{2}_{n} +b^{2}_{n} \bigr) \\ =&2A, \end{aligned}$$
In [18], Pan and Zhang investigated the properties of the locus of curvature centers of a closed strictly convex plane curve γ which is given by (2.1). Let β denote the locus of curvature centers of γ. Then \(\beta (\theta )=(\bar{x}_{1}( \theta ), \bar{x}_{2}(\theta ))\) can be given by
where \(N(\theta )=(-\cos \theta , -\sin \theta )\) is the unit inward normal vector along γ.
$$\begin{aligned} \beta (\theta ) =&\gamma (\theta )+\rho (\theta )N(\theta ) \\ =& \bigl(-p'(\theta )\sin \theta -p''(\theta )\cos \theta , p'( \theta ) \cos \theta -p''(\theta )\sin \theta \bigr) , \end{aligned}$$
(3.1)
We calculate the oriented area, denoted by \(\tilde{A}_{1}\), of β by Green’s formula. From (3.1), we get
and thus \(\tilde{A}_{1}\) is given by
$$\begin{aligned} \bar{x}_{1}\,d\bar{x}_{2} - \bar{x}_{2}\,d \bar{x}_{1}=p'(\theta ) \bigl(p'( \theta )+p'''(\theta ) \bigr)\,d\theta , \end{aligned}$$
$$\begin{aligned} \tilde{A}_{1} =&\frac{1}{2} \int _{\gamma }\bar{x}_{1}\,d\bar{x}_{2} - \bar{x}_{2}\,d\bar{x}_{1} \\ =&\frac{1}{2} \int _{0}^{2\pi }p'(\theta ) \bigl(p'(\theta )+p'''(\theta ) \bigr)\,d \theta \\ =& \frac{1}{2} \int _{0}^{2\pi } \bigl(p^{\prime 2}(\theta )-p^{\prime\prime 2}(\theta ) \bigr)\,d \theta . \end{aligned}$$
By (2.3), (2.4) and the Parseval equality, we have
where equality holds if and only if \(a_{n}=b_{n}=0\) for \(n> 1\), namely γ is a circle.
$$\begin{aligned} \tilde{A_{1}}=\frac{1}{2}\pi \sum _{n=1}^{\infty }n^{2} \bigl(1-n^{2} \bigr) \bigl(a^{2} _{n}+b^{2}_{n} \bigr) \leq 0, \end{aligned}$$
(3.2)
Via (2.9), (3.2) and (2.10) we have
$$\begin{aligned} 2 \bigl(A+ \vert \tilde{A}_{1} \vert \bigr) =&\frac{1}{2} \pi a^{2}_{0} + \pi \sum_{n=1}^{ \infty } \bigl(n^{2}-1 \bigr)^{2} \bigl(a^{2}_{n} + b^{2}_{n} \bigr) \\ =& \int _{\gamma }\frac{1}{\kappa }\,ds . \end{aligned}$$
(3.3)
Then we can prove the following inequalities.
Lemma 3
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane
\(\mathbb{R}^{2}\)
with the length
L
and enclosing a domain
D
of area
A. Let
κ
be the curvature of
γ, then
each equality holds if and only if
γ
is a circle.
$$\begin{aligned}& L^{2}-4\pi A \leq 2\pi \biggl( \int _{\gamma }\frac{1}{\kappa }\,ds-\frac{L ^{2}}{2\pi } \biggr), \end{aligned}$$
(3.4)
$$\begin{aligned}& L^{2}-4\pi A \leq 2\pi \biggl( \int _{\gamma }\frac{1}{\kappa }\,ds-2A \biggr), \end{aligned}$$
(3.5)
Proof
Via (2.8), (2.9) and (2.10), we have
Hence, we get (3.4), and equality holds if and only if \(a_{n}=b_{n}=0\) for \(n> 1\), namely γ is a circle. (3.5) is an immediate consequence according to (3.4) and the isoperimetric inequality (1.1). □
$$\begin{aligned} 2\pi \int _{\gamma }\frac{1}{\kappa }\,ds-L^{2} =&2\pi ^{2}\sum_{n=1} ^{\infty } \bigl(1-n^{2} \bigr)^{2} \bigl(a_{n}^{2}+b_{n}^{2} \bigr) \\ \geq &2\pi ^{2}\sum_{n=1}^{\infty } \bigl(n^{2}-1 \bigr) \bigl(a_{n}^{2}+b_{n}^{2} \bigr) \\ =&L^{2}-4\pi A. \end{aligned}$$
The inequality (3.4) is new. Substituting (3.3) into (3.5), we obtain the following reverse isoperimetric inequality as shown in [18].
Corollary 1
([18])
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve with length
L
in the Euclidean plane, κ
be the curvature of
γ, A
the area enclosed by
γ
and
\(\tilde{A}_{1}\)
the oriented area enclosed by
β. Then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} L^{2}-4\pi A\leq 4\pi \vert \tilde{A}_{1} \vert , \end{aligned}$$
Corollary 2
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve with length
L
in the Euclidean plane, κ
be the curvature of
γ, A
the area enclosed by
γ
and
\(\tilde{A}_{1}\)
the oriented area enclosed by
β. Then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} L^{2}-4\pi A\leq 2\pi \vert \tilde{A}_{1} \vert , \end{aligned}$$
We are now in the position to prove Theorem 1.
Proof of Theorem 1
Because of \(\varepsilon >0\), so \(n^{2}(1+\varepsilon )/4>1\) for \(n\geq 2\). Together with this fact and (2.8), (2.9) and (3.2), we have
equality holds if and only if \(a_{n}=b_{n}=0\) for \(n> 1\), namely γ is a circle. □
$$\begin{aligned} L^{2}-4\pi A =& 2\pi ^{2}\sum_{n=1}^{\infty } \bigl(n^{2}-1 \bigr) \bigl(a_{n}^{2}+b_{n} ^{2} \bigr) \\ \leq & 2\pi ^{2}\sum_{n=1}^{\infty } \frac{1+\varepsilon }{4}n^{2} \bigl(n ^{2}-1 \bigr) \bigl(a_{n}^{2}+b_{n}^{2} \bigr) \\ =& \frac{1+\varepsilon }{2} \pi ^{2}\sum_{n=1}^{\infty }n^{2} \bigl(n^{2}-1 \bigr) \bigl(a _{n}^{2}+b_{n}^{2} \bigr) \\ =&(1+\varepsilon )\pi \vert \tilde{A}_{1} \vert , \end{aligned}$$
Remark 1
The proof of Theorem 1 shows that the coefficient \((1+\varepsilon )\pi \) of \(|\tilde{A}_{1}|\) is the best constant for the equality condition. In fact: suppose there exists a constant \(\lambda <(1+\varepsilon )\pi \), such that
with the equality holds if and only if γ is a circle. Since ε is arbitrary, \(\lambda \leq \pi \). If \(\lambda =\pi \), the equality holds in Theorem 1 if and only if \(a_{n}=b_{n}=0\) for \(n>2\). That means there exists a γ that is not circle when the equality holds.
$$\begin{aligned} L^{2}-4\pi A\leq \lambda \vert \tilde{A}_{1} \vert , \end{aligned}$$
4 Stronger reverse isoperimetric inequality
From the definition of the width (1.3), (1.4) and (2.1), \(\zeta (\theta )=(\tilde{x}_{1}(\theta ), \tilde{x}_{2}( \theta ))\) can be given by
$$ \textstyle\begin{cases} \tilde{x}_{1}(\theta ) = \frac{p(\theta )-p(\theta +\pi )}{2}\cos \theta -p'(\theta )\sin \theta , \\ \tilde{x}_{2}(\theta )=\frac{p( \theta )-p(\theta +\pi )}{2}\sin \theta +p'(\theta )\cos \theta . \end{cases} $$
(4.1)
Lemma 4
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane, D
be a domain bounded by
γ, w
be the width function of
D, A
be the area enclosed by
γ
and
\(\tilde{A}_{2}\)
be the oriented area enclosed by
ζ. Then
\(\tilde{A}_{2}>0\), \(\tilde{A}_{2}=0\)
or
\(\tilde{A}_{2}<0\).
Proof
Using Green’s formula and (4.1), we have
and thus \(\tilde{A}_{2}\) is given by
$$\begin{aligned} &\tilde{x}_{1}\,d\tilde{x}_{2} - \tilde{x}_{2} \,d\tilde{x}_{1} \\ &\quad = \frac{1}{2} \biggl(p(\theta )p''(\theta )+p^{\prime 2}(\theta )+ p'( \theta )p'(\theta +\pi )- p''(\theta )p(\theta +\pi )\\ &\qquad {}+\frac{1}{2} \bigl(p( \theta )- p(\theta +\pi ) \bigr)^{2} \biggr)\,d\theta , \end{aligned}$$
$$\begin{aligned} \tilde{A}_{2} =&\frac{1}{2} \int _{\gamma }\tilde{x}_{1}\,d\tilde{x}_{2} - \tilde{x}_{2}\,d\tilde{x}_{1} \\ =&\frac{1}{4} \int _{0}^{2\pi } \biggl[p(\theta )p''( \theta )+p^{\prime 2}(\theta )+ p'(\theta )p'(\theta +\pi ) \\ &{} -p''(\theta )p(\theta +\pi )+\frac{1}{2}(p( \theta )- p(\theta + \pi ) \biggr]^{2} \,d\theta \\ =&\frac{1}{4} \int _{0}^{2\pi } \biggl(2p'(\theta )p'(\theta +\pi )+ \frac{1}{2} \bigl[p(\theta )-p(\theta +\pi ) \bigr]^{2} \biggr)\,d\theta . \end{aligned}$$
By (2.2) we get
$$\begin{aligned}& p(\theta +\pi ) = \frac{1}{2}a_{0}+\sum _{n=1}^{\infty }(-1)^{n}(a_{n} \cos n\theta + b_{n}\sin n\theta ), \end{aligned}$$
(4.2)
$$\begin{aligned}& p'(\theta +\pi ) = \sum _{n=1}^{\infty }(-1)^{n}(-na_{n}\sin n \theta +n b_{n}\cos n\theta ). \end{aligned}$$
(4.3)
Via the Parseval equality, we have
where
$$\begin{aligned} \tilde{A_{2}} =& \frac{\pi }{2}\sum_{n=1}^{\infty }(-1)^{n}n^{2} \bigl(a ^{2}_{n}+b^{2}_{n} \bigr)+ \frac{\pi }{8} \sum_{n=1}^{\infty } \bigl(1-(-1)^{n} \bigr)^{2} \bigl(a ^{2}_{n}+b^{2}_{n} \bigr) \\ =&\frac{\pi }{2}\sum_{k=1}^{\infty } \bigl((2k)^{2} \bigl(a^{2}_{2k}+b^{2} _{2k} \bigr)+ \bigl[1-(2k-1)^{2} \bigr] \bigl(a^{2}_{2k-1}+b^{2}_{2k-1} \bigr) \bigr) \\ =&B_{1} + B_{2}, \end{aligned}$$
(4.4)
$$\begin{aligned}& B_{1}=\frac{\pi }{2}\sum _{k=1}^{\infty }(2k)^{2} \bigl(a^{2}_{2k}+b^{2}_{2k} \bigr) \geq 0, \end{aligned}$$
(4.5)
$$\begin{aligned}& B_{2}=\frac{\pi }{2}\sum _{k=1}^{\infty } \bigl(1-(2k-1)^{2} \bigr) \bigl(a ^{2}_{2k-1}+b^{2}_{2k-1} \bigr)\leq 0. \end{aligned}$$
(4.6)
(I) If \(a_{0}=20\), \(a_{2}=b_{2}=1\) and \(a_{n}=b_{n}=0\) (\(n\geq 1\) and \(n\neq 2\)). Then \(p(\theta )=10 + \cos 2\theta + \sin 2 \theta \) is Minkowski’s support function of a convex domain D, since \(p(\theta )+p''(\theta )=10-3(\cos 2\theta + \sin 2\theta )>0\). From (4.4), \(\tilde{A}_{2}=2\pi >0\).
(II) If \(a_{0}=40\), \(a_{2}=\sqrt{2}\), \(a_{3}=1\), \(a_{n}=0\) (\(n\geq 1\) and \(n\neq 2, 3\)) and \(b_{n}=0\) (\(n\geq 1\)). Then \(p(\theta )=20 + \sqrt{2}\cos 2\theta + \cos 3\theta \) is Minkowski’s support function of a convex domain D, since \(p(\theta )+p''(\theta )=20-3 \sqrt{2}\cos 2\theta -8 \cos 3\theta >0\). From (4.4), \(\tilde{A} _{2}=0\).
(III) If \(a_{0}=20\), \(a_{3}=1\), \(a_{n}=0\) (\(n\geq 1\) and \(n\neq 3\)) and \(b_{n}=0\) (\(n\geq 1\)). Then \(p(\theta )=10 + \cos 3\theta \) is Minkowski’s support function of a convex domain D, since \(p(\theta )+p''(\theta )=10-8\cos 3\theta >0\). From (4.4), \(\tilde{A}_{2}=-4 \pi <0\). □
Next, we consider two special cases of γ and get the following lemma.
Lemma 5
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane, D
be a domain bounded by
γ, w
be the width function of
D, A
be the area enclosed by
γ
and
\(\tilde{A}_{2}\)
be the oriented area enclosed by
ζ.
(i)
If
D
is symmetrical about the origin, then
\(\tilde{A}_{2} \geq 0\)
and equality can occur if and only if
D
is a disc.
(ii)
If
D
has constant width, then
\(\tilde{A}_{2}\leq 0\)
and equality can occur if and only if
D
is a disc.
Proof
(i) If D is symmetrical about the origin, i.e., \(p(\theta )=p(\theta +\pi )\). By (2.2) and (4.2) we have \(a_{2k-1}=b_{2k-1}=0\), for \(k\geq 1\), from (4.4), this yields \(\tilde{A}_{2}=B_{1}\geq 0\), the equality holds if and only if D is a disc.
(ii) When D is a convex domain with constant width, \(\omega (\theta )=p(\theta )+p(\pi +\theta )=\) constant, so that \(\omega '(\theta )=p'( \theta )+p'(\pi +\theta )=0\). By (2.3) and (4.3), we have
We can easily get \(a_{2k}=b_{2k}=0\), for \(k\geq 1\).
$$\begin{aligned} 0 =&\omega '(\theta )=p'(\theta )+p'(\pi +\theta ) \\ =&\sum_{n=1}^{\infty } \bigl(1+(-1)^{n} \bigr) (-na_{n}\sin n\theta +nb_{n}\cos n \theta ) \\ =&\sum_{k=1}^{\infty }2(-2ka_{2k}\sin 2k\theta +2kb_{2k}\cos 2k \theta ). \end{aligned}$$
It is obvious that the sign of \(\tilde{A}_{2}\) is more complex than \(\tilde{A}_{1}\), so we consider the absolute value of \(\tilde{A}_{2}\) throughout this paper, and obtain a reverse isoperimetric inequality (Theorem 2).
Proof of Theorem 2
(I) If \(\tilde{A}_{2}\geq 0\). Let \(f_{1}(x)=\frac{x^{2}(x^{2}-2)}{x^{2}-1}\), \(x\geq 2\), then \(f_{1}'(x)=\frac{x ^{3}(3x^{2}-4)+4x}{(x^{2}-1)^{2}}>0\) for \(x\geq 2\), which yields \(f_{1}(x)\) to be a monotonically increasing function. So for any real \(\varepsilon >0\), \((2k)^{2}[(2k)^{2}-2]>\frac{8}{3+\varepsilon }[(2k)^{2}-1]\) for \(k\geq 1\), and \([(2k-1)^{2}+1]\geq 10>\frac{8}{3+\varepsilon }\) for \(k>1\).
By (3.2), (4.4), (2.8) and (2.9), we have
where equality holds if and only if \(a_{n}=b_{n}=0\) for \(n>1\), namely γ is a circle.
$$\begin{aligned} 4\pi \bigl( \vert \tilde{A}_{1} \vert - \vert \tilde{A}_{2} \vert \bigr) =&2\pi ^{2}\sum _{n=1}^{\infty }n ^{2} \bigl(n^{2}-1 \bigr) \bigl(a_{n}^{2}+b_{n}^{2} \bigr) -2\pi ^{2}\sum_{k=1}^{\infty }(2k)^{2} \bigl(a _{2k}^{2}+b_{2k}^{2} \bigr) \\ &{}+2\pi ^{2}\sum_{k=1}^{\infty } \bigl[(2k-1)^{2}-1 \bigr] \bigl(a_{2k-1}^{2}+b_{2k-1} ^{2} \bigr) \\ =&2\pi ^{2}\sum_{k=1}^{\infty }(2k)^{2} \bigl[(2k)^{2}-2 \bigr] \bigl(a_{2k}^{2}+b_{2k} ^{2} \bigr) \\ &{}+2\pi ^{2}\sum_{k=1}^{\infty } \bigl[(2k-1)^{2}+1 \bigr] \bigl[(2k-1)^{2}-1 \bigr] \bigl(a_{2k-1} ^{2}+b_{2k-1}^{2} \bigr) \\ \geq &2\pi ^{2}\sum_{n=1}^{\infty } \frac{8}{3+\varepsilon } \bigl(n^{2}-1 \bigr) \bigl(a _{n}^{2}+b_{n}^{2} \bigr) \\ =&\frac{8}{3+\varepsilon } \bigl(L^{2}-4\pi A \bigr), \end{aligned}$$
(II) If \(\tilde{A}_{2}<0\). Let \(f_{2}(x)=\frac{x^{4}}{x^{2}-1}\), \(x \geq 2 \), then \(f'_{2}(x)=\frac{2x^{3}(x^{2}-2)}{(x^{2}-1)^{2}}>0\) for \(x\geq 2\), which yields \(f_{2}(x)\) to be a monotonically increasing function. So for any real \(\varepsilon >0\), \((2k)^{4}>\frac{16}{3+ \varepsilon }[(2k)^{2}-1]>\frac{8}{3+\varepsilon }[(2k)^{2}-1]\) for \(k\geq 1\), and \((2k-1)^{2}-1\geq 8>\frac{8}{3+\varepsilon }\) for \(k>1\).
By (3.2), (4.4), (2.8) and (2.9) we have
where equality holds if and only if \(a_{n}=b_{n}=0\) for \(n>1\), namely γ is a circle.
$$\begin{aligned} 4\pi \bigl( \vert \tilde{A}_{1} \vert - \vert \tilde{A}_{2} \vert \bigr) =&2\pi ^{2}\sum _{n=1}^{\infty }n ^{2} \bigl(n^{2}-1 \bigr) \bigl(a_{n}^{2}+b_{n}^{2} \bigr) +2\pi ^{2}\sum_{k=1}^{\infty }(2k)^{2} \bigl(a _{2k}^{2}+b_{2k}^{2} \bigr) \\ &{}+2\pi ^{2}\sum_{k=1}^{\infty } \bigl[1-(2k-1)^{2} \bigr] \bigl(a_{2k-1}^{2}+b_{2k-1} ^{2} \bigr) \\ =&2\pi ^{2}\sum_{k=1}^{\infty }(2k)^{4} \bigl(a_{2k}^{2}+b_{2k}^{2} \bigr) \\ &{}+2\pi ^{2}\sum_{k=1}^{\infty } \bigl[(2k-1)^{2}-1 \bigr] \bigl[(2k-1)^{2}-1 \bigr] \bigl(a_{2k-1} ^{2}+b_{2k-1}^{2} \bigr) \\ \geq &2\pi ^{2}\sum_{n=1}^{\infty } \frac{8}{3+\varepsilon } \bigl(n^{2}-1 \bigr) \bigl(a _{n}^{2}+b_{n}^{2} \bigr) \\ =&\frac{8}{3+\varepsilon } \bigl(L^{2}-4\pi A \bigr), \end{aligned}$$
Therefore we obtain Theorem 2 and complete the proof. □
Letting \(\varepsilon =5\) in Theorem 2, we get a stronger lower bound of the integral of radius of curvature by using \(\tilde{A}_{2}\).
Theorem 3
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve with length
L
in the Euclidean plane, κ
be the curvature of
γ, w
be the width function of
D, A
be the area enclosed by
γ
and
\(\tilde{A}_{2}\)
be the oriented area enclosed by
ζ. Then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} \frac{L^{2}}{2\pi }+2 \vert \tilde{A}_{2} \vert \leq \int _{\gamma }\frac{1}{ \kappa }\,ds, \end{aligned}$$
Proof
Let \(\varepsilon =5\) in Theorem 2, by (3.3), we have
i.e.,
where equality holds if and only if γ is a circle. □
$$\begin{aligned} L^{2}+4\pi \vert \tilde{A}_{2} \vert \leq 4\pi \bigl( \vert \tilde{A}_{1} \vert +A \bigr)=2\pi \int _{\gamma }\frac{1}{\kappa }\,ds, \end{aligned}$$
$$\begin{aligned} \frac{L^{2}}{2\pi }+2 \vert \tilde{A}_{2} \vert \leq \int _{\gamma }\frac{1}{ \kappa }\,ds, \end{aligned}$$
Corollary 3
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane, w
be the width function of
D, A
be the area enclosed by
γ
and
\(\tilde{A}_{2}\)
be the oriented area enclosed by
ζ. Then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} 2 \bigl(A+ \vert \tilde{A}_{2} \vert \bigr)\leq \int _{\gamma }\frac{1}{\kappa }\,ds, \end{aligned}$$
By the isoperimetric inequality (1.1) and (3.3) we get an upper bound for the integral of radius of curvature.
Corollary 4
Let
γ
be a
\(C^{2}\)
closed and strictly convex curve in the Euclidean plane, κ
be the curvature of
γ, A
be the area enclosed by
γ
and
\(\tilde{A}_{1}\)
be the oriented area enclosed by
β. Then
where equality holds if and only if
γ
is a circle.
$$\begin{aligned} \int _{\gamma }\frac{1}{\kappa }\,ds\leq \frac{L^{2}}{2\pi } +2 \vert \tilde{A}_{1} \vert . \end{aligned}$$
Acknowledgements
The author would like to thank two anonymous referees for many helpful comments and suggestions that directly lead to the improvement of the original manuscript.
Availability of data and materials
Not applicable.
Competing interests
The author declares to have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.