• Keine Ergebnisse gefunden

Curves and Surfaces in Euclidean Space

N/A
N/A
Protected

Academic year: 2021

Aktie "Curves and Surfaces in Euclidean Space"

Copied!
29
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Lecture 7 A

Curves and Surfaces in Euclidean Space

p p p

CurvesinIR3 andIRd,Frenetframes SurfacesinIR3,Gaussframes Firstandsecondfundamentalform Curvatureofsurfaces

c 2003–2008 Martin Welk, 2015 Martin Schmidt, 2019 Marcelo Cárdenas

(2)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Curves inIR3

A

Reminder about Curves and Curvature in 2D

p In 2D, a regular curve is characterised up to Euclidean transformations by the curvatureκ(as function of the curve parameter)

p At each curve point, there are tangent and normal unit vectors~tand~n, such that in arc-length parametrisation

cs=~t , css=κ~n

c css

n

cs t = x

Curve with tangent and normal vectors and first two derivatives at a point, from Lecture 2

(3)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Curves inIR3

A

Curvature Parameters of a Curve in 3D

p Consider regular curve in 3D Euclidean space,c:I→IR3, in arc-length parametrisation

p At each curve pointc(s), there are

a a unit tangent vector ~t(s), cs =~t(s)

a a unit normal vector ~n(s), css =κ(s)~n(s)

a a unitbinormalvector~b(s), ~b(s) =~t(s)×~n(s) (unique ifcis2-regular, i.e.css6= 0)

p In contrast to the 2D case,κ(s)can always be chosen nonnegative

p ~t(s),~n(s),~b(s)form an orthonormal system, theFrenet frameofc

n b

t c s( )

c

CurvecinIR3with Frenet frame atc(s)

(4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Curves inIR3 (3)

A

Curvature Parameters of a Curve in 3D, cont.

p the derivative of~bis perpendicular to both~tand~b, i.e

~bs(s) =−τ(s)~n(s) with a functionτ(s), thetorsionofc

p Curvatureκ(s)and torsionτ(s)determine the curvec(s)up to rotations and translations

p Frenet-Serret equations(or Frenet equations):

d ds

~t

~n

~b

=

0 κ 0

−κ 0 τ

0 −τ 0

~t

~ n

~b

p The torsion can also be defined by

csss(s) =−κ(s)2~t(s) +κs(s)~n(s) +κ(s)τ(s)~b(s)

p The torsion vanishes identically if and only if the curve is contained in a plane

(5)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Curves inIRd

A

Frenet Frames in Higher Dimensions

p Consider a curvecinIRdin arc-length parametrisation

p For each curve pointc(s), there is an orthonormal basis(~e1, . . . , ~ed)such that cs(s) =~e1, css(s)∈Span (~e1, ~e2),

cs(k)(s)∈Span (~e1, . . . , ~ek), k≤d

p (~e1, . . . , ~ed)is theFrenet frameofc.

p Frenet equationsinddimensions:

d ds

~e1

~e2

...

~ed

=

0 κ1 0 . . . 0 0

−κ1 0 κ2 0 0

0 −κ2 0 0 0

... . .. ...

0 0 0 0 κd−1

0 0 0 . . . −κd−1 0

~e1

~e2

...

~ ed

p Curvature functionsκi(s),i= 1, . . . , d−1(nonnegative fori≤d−2) determinecup to rotations and translations

(6)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Surfaces in IRd

A

Surfaces

p Surface inIRd: Differentiable functionσ:D→IRd,D⊂IR2 connected domain

p Graph (image) of a surfaceσ: Set of points inIRdgiven by {σ(u, v)|(u, v)∈D}

Remark: Similarly as for curves, surfaces with identical graphs but different parametrisations are considered different.

Curves on a Surface

p σ:D→IR3 surface

p I⊂IRinterval, mappingI3p7→(u(p), v(p))∈D

p Byc:I→σ(D)⊂IR3, p7→σ(u(p), v(p)), a curve onσis given

(7)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Surfaces in IRd

A

Related Definitions

p Regular surface: Surfaceσ: (u, v)7→IRdis regular if the (Jacobi) matrix

Dσ:=

 σu1 σ1v

... ... σud σdv

has rank2everywhere inD.

p k-regularity can be defined analogously using higher order derivatives. We will always assume that surfaces are sufficiently many times differentiable.

p Tangent planeT(u,v)σtoσatσ(u, v): image of Dσ(u, v) : T(u,v)IR2→Tσ(u,v)IRd

(8)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Surfaces in IRd

A

Reparametrisation

p Transforms a surface into another one with the same graph

p σ:D→IRd surface

p D˜ ⊂IR2 connected domain

p ϕ: ˜D→Ddifferentiable mapping with rank Dϕ= 2everywhere

p σ˜:=σ◦ϕ: ˜D→IRdreparametrised surface

p orientation-preservingifdet Dϕ >0

(9)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

Gauss Frame for Surfaces in 3D

A

Gauss Frame for Surfaces in 3D

p Restrict now to surfacesσ:D→IR3 in 3D Euclidean space

p Gauss frame(analog of Frenet frame for surfaces) consists of the three unit vectors

~t1(u) := σu

uk , ~t2(u) := σv

vk, ~n:= σu×σv

u×σvk

p First two vectors~t1 and~t2 of the frame lie in tangential direction,~n perpendicular to the surface

p In general,~t1 and~t2 are not orthogonal

p ~t1 and~t2depend on parametrisation

p Normal vector~ndoes not change under orientation-preserving reparametrisation, is reverted by orientation-changing reparametrisation

t2 t1 n

(10)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

First Fundamental Form for Surfaces in 3D

A

First Fundamental Form

p Consider regular surfaceσ:D→IR3

p Use boldface lettersu,v, . . .for points and vectors inIR2

p Symmetric bilinear form

Iu(v,w) :=hDσ(u)v,Dσ(u)wi, v,w∈TuD

p p

iscalledfirstfundamentalformofσatu RegularityimpliesIu(w,w)6=0, for nonzero w

Incoordinates,IuisdescribedbyamatrixwhichwewillalsodenotebyIu

(u= (u, v)):

Iu=I(u,v)= E F

F G

,

E=hσu, σui, F =hσu, σvi, G=hσv, σvi

p MapDσ(u) : TuIR2→Tuσ⊂Tσ(u)IR3allows to transferIu also into a bilinear form onTuσ

(11)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

First Fundamental Form for Surfaces in 3D

A

Transformation Properties of the First Fundamental Form

p The first fundamental form isinvariant under Euclidean transformations (including reflections) ofIR3:

Forψ:x7→Ax+b,A∈O(3,IR),b∈IR3, and˜σ:=ψ◦σone has

˜Iu(v,w) =Iu(v,w)

where˜Iu is first fundamental form of˜σ

p The first fundamental formtransforms under reparametrisationsas follows:

Letσ˜:=σ◦ϕ,ϕ: ˜D→D; then

˜Iu(v,w) =Iϕ(u)(Dϕ(v),Dϕ(w)) where˜Iu is first fundamental form of˜σ

(12)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

First Fundamental Form for Surfaces in 3D

A

MeasurementsUsingtheFirstFundamentalForm

p Considersurfaceσ:D → IR3 andacurvec:[0,P] → D)onσ.Then

I(u(p),v(p))(cp(p), cp(p)) =cTpI(u(p),v(p))cp=E u2p+ 2F upvp+G vp2 Lengthofc:

L[c] =

P

Z

0

kcp(p)kdp=

P

Z

0

q

E u2p+ 2F upvp+G vp2 dp

p Anglebetween vectors in a point (here forσu, σv) cos <) (σu, σv) = hσu, σvi

uk kσvk= F

√EG

(13)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M I

First Fundamental Form for Surfaces in 3D

A

Measurements Using the First Fundamental Form, cont.

p Areaof surfaceσ(D):

A[σ] = Z Z

D

q

detI(u,v)dudv= Z Z

D

pEG−F2 dudv

p The bilinear form onT(u,v)σdefined by the first fundamental form is a Riemannian metric on the surface (graph). It is obtained by restricting the Euclidean metric ofIR3 to the surface.

(14)

1 2 3 4 5 6 7 8 9 10 11 12 13

M I

Outlook and References

A

References

p G. Sapiro: Geometric Partial Differential Equations and Image Analysis.Cambridge University Press 2001

p W. Haack: Differential-Geometrie, Teil I.Wolfenbütteler Verlagsanstalt, Wolfenbüttel 1948 (in German)

14

(15)

c c ss

n

c s

t =

x

(16)

n

b

t c s ( )

c

(17)

t 2

t 1

n

(18)
(19)
(20)

n t

projection

κ n c

(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
(29)

Referenzen

ÄHNLICHE DOKUMENTE

In this paper, we find travelling wave solutions of some nonlinear partial differential equations (PDEs) by using the extended tanh-function method1. Some illustrative equations

In this section, we introduce the basic tools necessary for the study of rough PDEs of the form (2.22), namely, the rough Gronwall Lemma and an a priori estimate on the remainder

• In Section 7.3 we introduce the moduli space of quadratic differentials with double poles, and prove that the number of integral points is computed by discrete topological

Looking for both single-valued and set-valued solutions t o systems of first-order partial differential inclusions is then the topic of this paper.. We present it

Using Post’s concept of generalized norm resolvent convergence on varying Hilbert spaces we prove that solutions to the Burgers equation can be approximated in a certain weak sense

In the papers Papageorgiou (1991), Papageorgiou (1993) and Halidias and Pa- pageorgiou (2002), the existence of a solution for optimal control problems of nonlinear and

Some will be with respect to the norm of the phase space, for bounded sets of data, which provide a good approximation of the initial flow in an obvious way - the non-squeezing

In the case of hyperbolic scalar conservation laws, Debussche and Vovelle [9] defined a notion of generalized kinetic solution and obtained a comparison result showing that