Differential geometry

Japanese: 微分幾何学 - びぶんきかがく(英語表記)differential geometry
Differential geometry

In the classical sense, differential geometry is a branch of mathematics that studies the properties of curves and surfaces in the plane and in space by applying differential calculus. In modern terms, it refers to the theory of manifolds with some structure, such as Riemannian metrics, complex structures, and connections. The classical definition of differential geometry was established by K. F. Gauss.

There are two types of concepts and properties of curves, surfaces, and manifolds in general: those that are determined only by information within a small enough neighborhood of each point, and those that are determined in relation to the whole. The former are called local (or small-area) and the latter global. The curvature of a curve or surface is a typical example of a local concept. In contrast, the width of a closed convex curve, the relationship between the length of a closed curve and the area it encloses, the number of holes in a surface, and the number of closed geodesics on a closed surface are global concepts (or properties). While local properties are primarily studied using differential calculus, the study of global properties requires the mobilization of knowledge from other fields, including topology and Lie group theory.

[Koichi Ogiue]

Geometry of Plane Curves

If a plane curve is expressed as x=x(s) with the arc length s as a variable, then e 1 =x′(s) is a unit tangent vector. If the vector obtained by rotating e 1 90 degrees in the positive direction is e 2 , then e 1 and e 2 form an orthonormal system at each point on the curve. In this case, e 1 and e 2 are expressed by the differential equation

satisfies. κ is called the curvature of the curve. If κ ≡ 0 (identically zero at every point on the curve), it is a straight line, and if κ ≡ 1/r (constant), it is a circle of radius r. The local properties of a plane curve are completely determined by its curvature (Fundamental Theorem of Plane Curves). Well-known global properties of plane curves include the four-vertex theorem, isoperimetric inequalities, and curves of constant width.

[Koichi Ogiue]

Geometry of Space Curves

A space curve is expressed as x = x(s) with the arc length s as a variable, e 1 = x′(s), e 2 = x″(s)/‖x″(s)‖, e 3 = e 1 × e 2
Then, e 1 , e 2 , and e 3 form an orthonormal system at each point on the curve. e 1 is called the unit tangent vector, e 2 is called the unit principal normal vector, and e 3 is called the unit binormal vector. In this case, e 1 , e 2 , and e 3 are differential equations (Frenet-Seret formula)

where κ=‖x″(s)‖,
τ = (1/κ 2 )(x′×x″)・x
where κ is the curvature and τ is the torsion. Curvature represents the degree to which a curve deviates from its tangent, and torsion represents the degree to which the curve deviates from the plane spanned by e1 and e2 (called the osculating plane). The local properties of a space curve are completely determined by its curvature and torsion (Fundamental Theorem of Space Curves). A well-known example of a global property of space curves is the inequality for the total curvature ∫κ ds , proposed by Werner Fenchel (1905-1988) and John Willard Milnor (1931- ).

[Koichi Ogiue]

Various properties of surfaces

A surface is completely determined by the first and second fundamental forms. The first fundamental form is a quadratic form (i.e. a second-order symmetric matrix) that gives a scale of length on the surface and is an example of a Riemannian metric. The second fundamental form is a quadratic form that describes the curvature of the surface as viewed from the outside.

At a point P on a surface S, the maximum and minimum values ​​of the curvature (as a plane curve) at P of the intersection line between S and a plane containing the normal to S (when various planes are used) are called the principal curvatures of S at P, and the product of the principal curvatures is called the Gaussian curvature ( Figure A ). Gaussian curvature is a quantity that faithfully represents the "curvature" of a surface. Depending on whether the Gaussian curvature at P is positive, negative, or zero, P is called an elliptical point, a hyperbolic point, or a parabolic point ( Figure B ). A mapping that maps a unit normal vector at P to a point P on a surface S is called a Gaussian mapping ( Figure C ). The Gaussian mapping is a mapping from S to a unit sphere, and the Gaussian curvature is equal to the "magnification rate of the area" due to the Gaussian mapping. In other words, if the area enclosed by a closed curve around a point P on S is A, and the area enclosed by the closed curve on the unit sphere corresponding to the Gaussian map is A * , the limit of A * /A when the closed curve is converged to P is equal to the Gaussian curvature at P. A surface with a constant Gaussian curvature is called a surface of constant curvature. Planes, cylinders, and spheres are surfaces of constant curvature. Closed surfaces with constant curvature are limited to spheres (Liebmann's theorem). The arithmetic mean of the principal curvatures is called the mean curvature. A surface whose mean curvature is zero everywhere is called a minimal surface. A minimal surface has the property that it has the smallest area locally among surfaces that have a given closed curve as its boundary. A surface created by stretching a soap film around a wire ring is a minimal surface. A surface of revolution with a catenary as its generating line (called a catenary surface) is a minimal surface, but conversely, the only surface of revolution that is a minimal surface is a catenary surface.

Two surfaces whose first fundamental forms at corresponding points are the same can be superimposed on each other without stretching (in this case, the two surfaces are said to be isometric). Two surfaces whose first and second fundamental forms at corresponding points are the same can be superimposed by spatial motion (in this case, the two surfaces are said to be congruent). Even if two surfaces have the same length, they are not necessarily congruent. For example, in Figure D , (1) and (2) are globally isometric but not congruent (locally), and (1) and (3) and (2) and (3) are locally isometric but not congruent. The most important result among the local properties of surfaces is Gauss's fundamental theorem, which states that "Gaussian curvature is determined only by the first fundamental form." This shows that two surfaces that are isometric have the same Gaussian curvature at corresponding points. However, the converse does not hold. For example, if we consider a surface consisting of cylinders of different lengths with hemispherical lids as shown in Figure E , even though there can be a one-to-one correspondence between the two such that the Gaussian curvatures at corresponding points are the same, the two cannot be of equal length. A surface that is locally isometric to a plane is called a developable surface. For a surface to be developable, a necessary and sufficient condition is that its Gaussian curvature is zero everywhere. A developable surface is the envelope surface of a one-parameter plane family, and can be either a plane, a cylinder, a cone, or a tangent surface.

One of the famous global properties of surfaces is the theorem of Stefan Cohn-Vossen (1902-1936), which states that "two closed convex surfaces of equal length are congruent." This proves that a sphere in particular cannot be distorted. However, if a small disk is removed from a sphere, the remainder can be distorted. The most wonderful theorem among global results concerning surfaces is the Gauss-Bonnet theorem. In other words, there is a relationship between the Gaussian curvature K and the Euler characteristic χ(S) of a closed surface S, whose front and back can be determined,

The Euler characteristic χ(S) for a triangulation of a surface S is χ(S)=(number of vertices)-(number of edges)+(number of faces).
The Gauss–Bonnet theorem is a very elegant theorem that gives the relationship between the differential geometric quantity (curvature) and the topological quantity (Euler characteristic) of a surface.

The properties of the first fundamental form of a surface were generalized by B. Riemann, and have continued to exist to this day as Riemannian geometry. Research into the properties of the second fundamental form has grown into the geometry of submanifolds. Today, differential geometry has made remarkable progress in connection with other fields such as topology, Lie group theory, algebraic geometry, the theory of functions of many variables, and the theory of partial differential equations.

[Koichi Ogiue]

[Reference] | Riemannian geometry
Differential Geometry (Gaussian Curvature) [Figure A]
©Shogakukan ">

Differential Geometry (Gaussian Curvature) [Figure A]

Differential Geometry (Elliptic Points, Hyperbolic Points, Parabolic Points) [Figure B]
©Shogakukan ">

Differential Geometry (Elliptic Points, Hyperbolic Points, Parabolic Points) […

Differential Geometry (Gaussian Mapping) (Figure C)
©Shogakukan ">

Differential Geometry (Gaussian Mapping) (Figure C)

Differential Geometry (Isometry and Congruence) [Figure D]
©Shogakukan ">

Differential Geometry (Isometry and Congruence) [Figure D]

Differential geometry (surfaces made of cylinders of different lengths with hemispherical lids) [Figure E]
©Shogakukan ">

Differential geometry (cylinders of different lengths with hemispherical covers)


Source: Shogakukan Encyclopedia Nipponica About Encyclopedia Nipponica Information | Legend

Japanese:

微分幾何学は、古典的な意味では、平面および空間における曲線や曲面などの性質を微分学を応用して研究する数学の一分野である。現代的には、リーマン計量、複素構造、接続などのようになんらかの構造が与えられた多様体の理論を意味する。古典的な意味の微分幾何学を確立したのはK・F・ガウスである。

 曲線や曲面、あるいは一般に多様体に関する概念や性質には、各点の十分小さい近傍(きんぼう)内の情報だけで定まるものと、全体に関連して定まるものとがある。前者を局所的(または小域的)といい、後者を大域的という。曲線や曲面の曲率は局所的概念の典型である。これに対して、凸閉曲線の幅、閉曲線の長さとその囲む面積との関係、曲面の穴の数、閉曲面上に存在する閉測地線の数などは、大域的な概念(または性質)である。局所的な性質は、微分学を主要な手段として研究されるのに対して、大域的な性質の研究には、位相幾何学やリー群論などをはじめとする他の諸分野の知識を動員する必要がある。

[荻上紘一]

平面曲線の幾何学

平面曲線を弧長sを変数としてx=x(s)と表せば、e1=x′(s)は単位接ベクトルである。e1を正の向きに90度回転して得られるベクトルをe2とすれば、e1、e2は曲線上の各点で正規直交系をなす。このときe1、e2は微分方程式

を満たす。κを曲線の曲率という。κ≡0(曲線上の各点で恒等的にゼロ)ならば直線であり、κ≡1/r(一定)ならば半径rの円である。平面曲線の局所的な性質は曲率によって完全に決まる(平面曲線の基本定理)。平面曲線に関する大域的な性質としては、四頂点定理、等周不等式、定幅曲線などが有名である。

[荻上紘一]

空間曲線の幾何学

空間曲線を弧長sを変数としてx=x(s)と表し
 e1=x′(s),e2=x″(s)/‖x″(s)‖,e3=e1×e2
とすれば、e1、e2、e3は曲線上の各点で正規直交系をなす。e1を単位接ベクトル、e2を単位主法線ベクトル、e3を単位従法線ベクトルという。このときe1、e2、e3は微分方程式(フレネ‐セレーの公式)

を満たす。ただし
  κ=‖x″(s)‖,
  τ=(1/κ2)(x′×x″)・x
であり、κを曲率、τを捩率(れいりつ)という。曲率は曲線がその接線から離れる度合いを表し、捩率は曲線がe1、e2で張られる平面(接触平面という)から離れる度合いを表す。空間曲線の局所的な性質は曲率と捩率により完全に決まる(空間曲線の基本定理)。空間曲線に関する大域的な性質としてはフェンヒェルWerner Fenchel(1905―1988)やミルナーJohn Willard Milnor(1931― )らによる全曲率∫κdsに関する不等式が有名である。

[荻上紘一]

曲面のいろいろな性質

曲面は第一基本形式と第二基本形式によって完全に決まる。第一基本形式は曲面上で長さを測る尺度を与える二次形式(すなわち二次の対称行列)で、リーマン計量の例である。第二基本形式は外からみた曲面の曲がりぐあいを表す二次形式である。

 曲面S上の点PにおいてSの法線を含む平面とSとの交線のPにおける(平面曲線としての)曲率の(平面をいろいろ変えてみたときの)最大値と最小値をSのPにおける主曲率といい、主曲率の積をガウス曲率という(図A)。ガウス曲率は曲面の「曲がりぐあい」を忠実に表す量である。Pにおけるガウス曲率が正、負、ゼロに応じてPを楕円(だえん)点、双曲点、放物点という(図B)。曲面Sの点Pに対してPにおける単位法ベクトルを対応させる写像をガウス写像という(図C)。ガウス写像はSから単位球面への写像であり、ガウス曲率はガウス写像による「面積の拡大率」に等しい。すなわちSの点Pの周りの閉曲線の囲む面積をA、ガウス写像で対応する単位球面上の閉曲線の囲む面積をA*として、閉曲線をPに収束させたときのA*/Aの極限がPにおけるガウス曲率に等しい。ガウス曲率が一定であるような曲面を定曲率曲面という。平面、柱面、球面などは定曲率曲面である。定曲率閉曲面は球面に限る(リープマンの定理)。主曲率の相加平均を平均曲率という。平均曲率が至る所ゼロであるような曲面を極小曲面という。極小曲面は、局所的には与えられた閉曲線を境界にもつ曲面のなかで面積が最小であるという性質をもつ。針金の輪にせっけん膜を張ってできる曲面は極小曲面である。懸垂(けんすい)線を母線とする回転面(懸垂面という)は極小曲面であるが、逆に極小曲面となる回転面は懸垂面に限る。

 対応する点における第一基本形式が一致するような二つの曲面は、互いに伸縮なく重ね合わせることができる(このとき二つの曲面は互いに等長であるといわれる)。対応する点における第一基本形式と第二基本形式がともに一致するような二つの曲面は、空間の運動によって重ね合わせることができる(このとき二つの曲面は合同であるといわれる)。二つの曲面は、等長であっても合同であるとは限らない。たとえば、図Dにおいて(1)と(2)は大域的に等長であるが(局所的にも)合同ではなく、(1)と(3)および(2)と(3)は、局所的に等長であるが合同ではない。曲面の局所的性質のなかでもっとも重要な結果は、「ガウス曲率が第一基本形式だけで決まる」というガウスの基本定理である。これにより互いに等長な二つの曲面は、対応する点において同じガウス曲率をもつことがわかる。しかし、この逆は成り立たない。たとえば、図Eのように長さの異なる円柱に半球の蓋(ふた)をつけた曲面を考えれば、両者の間に対応点におけるガウス曲率が一致するような1対1の対応がつけられるにもかかわらず、両者は等長ではありえない。平面と局所的に等長な曲面をとくに可展面という。曲面が可展面であるためには、そのガウス曲率が至る所ゼロであることが必要十分条件である。可展面は一径数平面族の包絡面(ほうらくめん)であり、平面、柱面、錐(すい)面、接線曲面のどれかである。

 曲面の大域的な性質のなかで有名なものとして、「互いに等長な二つの凸閉曲面は合同である」というコーン・フォッセンStefan Cohn-Vossen(1902―1936)の定理がある。このことからとくに球面は歪曲(わいきょく)不可能であることがわかる。しかし、球面から小円板を取り除いた残りは歪曲可能である。曲面に関する大域的な結果のなかでもっともすばらしい定理は、ガウス‐ボンネの定理である。すなわち、表裏が決められる閉曲面Sのガウス曲率Kとオイラー標数χ(S)の間には

なる関係が成り立つ。オイラー標数χ(S)は、曲面Sの三角形分割に対して
  χ(S)=(頂点の数)-(辺の数)+(面の数)
で定義される。ガウス‐ボンネの定理は、曲面の微分幾何学的な量(曲率)と位相幾何学的な量(オイラー標数)との間の関係を与える非常に美しい定理である。

 曲面の第一基本形式に関する性質はB・リーマンによって一般化されて、リーマン幾何学として今日に至っている。また、第二基本形式に関する性質の研究は部分多様体の幾何学に成長した。今日、微分幾何学は位相幾何学、リー群論、代数幾何学、多変数関数論、偏微分方程式論など、他の分野との関連のもとに著しい発展をしている。

[荻上紘一]

[参照項目] | リーマン幾何学
微分幾何学(ガウス曲率)〔図A〕
©Shogakukan">

微分幾何学(ガウス曲率)〔図A〕

微分幾何学(楕円点、双曲点、放物点)〔図B〕
©Shogakukan">

微分幾何学(楕円点、双曲点、放物点)〔…

微分幾何学(ガウス写像)〔図C〕
©Shogakukan">

微分幾何学(ガウス写像)〔図C〕

微分幾何学(等長と合同)〔図D〕
©Shogakukan">

微分幾何学(等長と合同)〔図D〕

微分幾何学(長さの異なる円柱に半球の蓋をつけた曲面)〔図E〕
©Shogakukan">

微分幾何学(長さの異なる円柱に半球の蓋…


出典 小学館 日本大百科全書(ニッポニカ)日本大百科全書(ニッポニカ)について 情報 | 凡例

<<:  Differential form

>>:  Differentiable manifold

Recommend

Koinumaru no ho - Koinumaru no ho

A ho in Issai County, Harima Province. It was also...

Requinto (English spelling) [Spain]

A small guitar that is usually tuned a fifth highe...

motor scooter

...In English, the word scooter originally meant ...

Wholesale price - Oroshi soba

...Even if transactions are conducted individuall...

Completed payment list - Completed payment list

In the Edo period, this document was issued when t...

Nakanoshima School - Nakanoshima

This is an artistic school of Yamada-ryu koto musi...

Globe amaranth (Gomphrena globosa)

An annual plant of the Amaranthaceae family, nativ...

United States - America

...The USA itself did not officially become a nat...

Public Service - Regret

〘noun〙 ("ku" is the Go-on pronunciation ...

Roraima

In 1757, the western half of the state was separa...

Abandoned Dynasty - Haicho

The emperor does not oversee the affairs of the I...

Cooperative corps - Kyodotai

…During that time, he put his ideas into practice...

Clothing - Emon

In art, it refers to the wrinkles in clothing depi...

Picture book of the Minister of the Interior's journey to China

An illustrated scroll from the early Kamakura peri...

Mizuki

A district in the west of Dazaifu City, central Fu...