Tangent Maps of Isometries

Part 22
Tangent Map of an Isometry
Printed version of the lecture Differential Geometry on 23. November 2009
Tommy R. Jensen, Department of Mathematics, KNU
22.1
Overview
Contents
1
Tangent Maps
1
2
Two Frames Determine a Unique Isometry
3
3
Conclusion
4
22.2
1
Tangent Maps
Tangent maps
Definition 7.4 (Chapter 1)
Let F : Rn → Rm be a mapping.
Let p be a point in Rn and v be a tangent vector to Rn at p.
The initial velocity vector β 0 (0) of the curve
β (t) = F(p + tv)
is denoted by F ∗ (v). It is a tangent vector to Rm at β (0) = F(p).
The function F∗ from tangent vectors to Rn to tangent vectors to Rm is called the tangent map of F.
Proposition 7.5 (Chapter 1)
Let F = ( f1 , f2 , . . . , fm ) : Rn → Rm be a mapping.
If v is a tangent vector to Rn at p, then
F ∗ (v) = (v[ f1 ], v[ f2 ], . . . , v[ fm ])F(p) .
22.3
1
The tangent map of an isometry
Theorem 2.1
Let F be an isometry of R3 with orthogonal part C.
Let v p be any tangent vector to R3 at some point p.
Then
F ∗ (v p ) = C(v)F(p) .
Proof of Theorem 2.1
Write F = T (C), where T is the translation by a, so
F(p) = T (C(p)) = a +C(p).
Then using that C is linear:
β (t) = F(p + tv) = a +C(p + tv) = a +C(p) + tC(v) = F(p) + tC(v).
This implies F ∗ (v p ) = β 0 (t)|t=0 = C(v)β (0) = C(v)F(p) .
22.4
Calculating Euclidean coordinates of the tangent map
Theorem 2.1 can be used to calculate F∗
Let F = T (C), where C is given by the orthogonal matrix


c11 c12 c13
 c21 c22 c23  .
c31 c32 c33
For any tangent vector v p we can write
v p = ∑ viUi (p).
Then using Theorem 2.1 we have
F ∗ (v p ) = F ∗ (∑ viUi (p)) = ∑ ci j v jUi (F(p)).
i, j
22.5
Tangent maps of isometries preserve dot products
Corollary 2.2
Let F be an isometry and v p and w p be tangent vectors to R3 at the same point p.
Then
F ∗ (v p ) • F ∗ (w p ) = v p • w p .
Proof of Corollary 2.2
Let C be the orthogonal part of F.
Then using Theorem 2.1, and that C preserves dot products:
F ∗ (v p ) • F ∗ (w p )
=
C(v)F(p) •C(w)F(p)
=
C(v) •C(w)
=
v•w
=
vp • wp.
22.6
2
Tangent maps of isometries preserve frames
Frames get mapped to frames
Let F be an isometry of R3 and let e1 , e2 , e3 be a frame at some point p.
Then
ei • e j = δi j for all i, j = 1, 2, 3.
From Corollary 2.2 follows that also
F ∗ (ei ) • F ∗ (e j ) = ei • e j = δi j for all i, j = 1, 2, 3.
Therefore F ∗ (e1 ), F ∗ (e2 ), F ∗ (e3 ) form a frame at point F(p).
22.7
2
Two Frames Determine a Unique Isometry
Two frames determine a unique isometry
Theorem 2.3
Let e1 , e2 , e3 be a frame at a point p, and let f1 , f2 , f3 be a frame at a point q, both in R3 .
Then there exists a unique isometry F of R3 such that
F ∗ (ei ) = fi for i = 1, 2, 3.
Proof of Theorem 2.3
There are two parts to the proof:
• Existence.
• Uniqueness.
22.8
Existence proof
Let eˆ1 , eˆ2 , eˆ3 be the vector parts of e1 , e2 , e3 . These are points in R3 .
Let fˆ1 , fˆ2 , fˆ3 be the vector parts of f1 , f2 , f3 .
Let C be the mapping
C : R3 → R3 , ∑ ai eˆi 7→ ∑ ai f̂i .
It is clear that C preserves dot products, so C is orthogonal.
Let T be the translation by q −C(p), and let F = T (C).
Then F(p) = T (C(p)) = C(p) + (q −C(p)) = q.
Using Theorem 2.1 to calculate F ∗ (ei ) gives:
F ∗ (ei ) = C(eˆi )F(p) = ( f̂i )q = fi for i = 1, 2, 3.
It follows that F is an isometry with a tangent map that maps the frame e1 , e2 , e3 to the frame f1 , f2 , f3 .
22.9
3
Uniqueness proof
Assume that F = T (C) has a tangent map that maps the frame e1 , e2 , e3 to the frame f1 , f2 , f3 .
Then it follows from Theorem 2.1 that C is a linear map that maps eˆi to f̂i for each i = 1, 2, 3.
Then C(∑ ai eˆi ) = ∑ aiC(eˆi ) = ∑ ai f̂i implies C = C.
It also follows from Theorem 2.1 that F maps p to q :
q = F(p) = T (C(p)) = T (C(p)) = T (T −1 (q)) ⇒ T = T.
So F = T (C) = F, which proves the Theorem.
22.10
How to calculate the isometry of Theorem 2.3
Assume that the attitude matrix of the frame e1 , e2 , e3 at the point p is A = (ai j ), and the attitude matrix
of the frame f1 , f2 , f3 at the point q is B = (bi j ).
This means that ei = (ai1 , ai2 , ai3 ) p , and fi = (bi1 , bi2 , bi3 )q , for i = 1, 2, 3.
Therefore A(eˆi ) = (e1 • ei , e2 • ei , e3 • ei ) = ui for i = 1, 2, 3.
It follows that
t
BA(eˆi ) = t B(ui ) = (bi1 , bi2 , bi3 ) = f̂i for i = 1, 2, 3.
If F = T (C) is the isometry which maps one frame to the other, then we know that C(eˆi ) = f̂i for
i = 1, 2, 3, and that C is uniquely determined.
So we have C = t BA.
It remains to calculate T as the translation by q −C(p).
22.11
3
Conclusion
The End
END OF THE LECTURE!
(Never give up. . . )
22.12
4
Next time:
Orientation
22.13
5