Você está na página 1de 27

INVARIANT METHODS AND THE

EUCLIDEAN GROUP
José Marı́a Rico Martı́nez
Mechanical and Aeronautical Engineering Department
University of California, Davis. Davis, CA. 95616
Bahram Ravani
Mechanical and Aeronautical Engineering Department
University of California, Davis. Davis, CA. 95616

1 Abstract
The Euclidean group has always played a critical role in the study of kinematics.
In fact, even though in many studies it is not mentioned, the object of study of
most mechanisms and robotics research is precisely the Euclidean group. Thus,
the Euclidean group appears to be too well researched for any additional work.
Nevertheless, in this work, coordinate free methods are employed to the fullest
extend to study the Euclidean group. The results are surprising because the
beauty of the approach and the computational power of the results.

2 Introduction
3 Physical Space as a Euclidean Space
For purposes of study of classical mechanics, physical space can be modeled as
a three-dimensional Euclidean space.
Definition 1. A Euclidean space is a set of points E coupled with a mapping
δ from the Cartesian product E × E into 3 , a three-dimensional vector space
endowed with a positive definite quadratic form, defined by

δ : E × E → 3 δ(M, N ) = M → N, (1)
see Figure 1, with the additional properties
1. For every N ∈ E, the restriction mapping

δ N : E → 3 δN (M ) = δ(M, N ) = M → N, (2)
is a bijection and

1
Figure 1: Mapping δ from E × E to 3 .

2. For all L, M, N ∈ E, the triangle axiom holds, see Figure 2; i.e.

(L → M ) + (M → N ) = (L → N ) (3)

Figure 2: The Triangle Axiom of Mapping δ from E × E to 3 .

This property can be expressed in terms of the mapping δ by

δ(L, M ) + δ(M, N ) = δ(L, N ) (4)

The elements of 3 are called vectors. In particular, the vector M → N is


called the difference vector of point M with respect to point N. This notation
is specially useful since it emphasizes that the elements of 3 are, in this case,
dependent upon the points of E.
In fact, the existence of a mapping δ : E × E → 3 fulfilling definition 1
ensures the existence of a family of similar mappings

2
δ μ : E × E → 3 δμ (M, N ) = μδ(M, N ) for each μ ∈ , μ = 0, (5)

all of which also satisfy the axioms. The specific value of μ depends on an
arbitrary election of a unit of length which cannot be intrinsically previously
selected.
It is important to recognize that the set of points E is at the outset com-
pletely void of algebraic structure. Hence, there is no meaningful way of adding,
subtracting or multiplying the points of E. However, the rich algebraic struc-
ture already existing in the quadratic vector space 3 can be employed to induce
several algebraic structures in E.

4 Free Vector Algebraic Structures in E × E


The first of the algebraic structures, induced by 3 into E × E, to be studied is
that of the free vectors. Before proceeding to the introduction of this structure,
it is necessary to prove that E × E supports an equivalence relationship.
Definition 2. The Cartesian product, E×E of the Euclidean space supports
the following relationship

(M, N ) ≡ (P, Q) ⇔ δ(M, N ) = δ(P, Q), (6)

where M, N, P, Q ∈ E.
Proposition 3. The relationship given in definition 2 is an equivalence
relationship.
Proof: It is sufficient to prove that the relationship is reflexive, symmetric
and transitive.

1. It is reflexive. Trivially δ(M, N ) = δ(M, N ). Thus (M, N ) ≡ (M, N ).


2. It is symmetric. Let (M, N ) ≡ (P, Q). Then δ(M, N ) = δ(P, Q), and thus
δ(P, Q) = δ(M, N ). Therefore (P, Q) ≡ (M, N ).
3. It is transitive. Let (M, N ) ≡ (P, Q), and (P, Q) ≡ (R, S). Then
δ(M, N ) = δ(P, Q) and δ(P, Q) = δ(R, S). Thus δ(M, N ) = δ(R, S)
and therefore (M, N ) ≡ (R, S).

Further since δ : E × E → 3 is surjective, it is possible to associate with


every v ∈ 3 an equivalence class, see Figure 3, denoted by v and defined by

v = {(M, N ) ∈ E × E|δ(M, N ) = v }.

In the field of classical mechanics v is usually referred as a free vector.


Definition 4. The set of all free vectors will be denoted by 3 , and defined
by

3
Figure 3: Elements of an Equivalence Class on the Set of Free Vectors.

3 
 = v. (7)
v ∈3


It should be noted that an equivalence relationship on E × E sets up a


partition of E × E, where each one of the subsets are the equivalence classes,
see Figure 4.

Figure 4: Partition of E × E in Terms of the Equivalence Classes.


3
It will now be demonstrated that  has the structure of a real orthogonal
vector space. Hence, it can be regarded as a real orthogonal space of free vectors.
3
For that purpose, the mapping δ̄ :  → 3 will be defined.
Definition 5. The mapping δ̄ induced by the mapping δ from the real
orthogonal space of free vectors into the the three-dimensional vector space, is
defined by
3
δ̄ :  → 3 δ̄(v) = v . (8)

4
Proposition 6. The induced mapping δ̄ is well defined and bijective.
Proof: It suffices to show that δ̄ is well defined, injective and surjective.

1. δ̄ is well defined. Let (M, N ), (P, Q) ∈ v; therefore δ(M, N ) = v = δ(P, Q),


and δ̄(M, N ) = v = δ̄(P, Q).
3
2. δ̄ is injective. Let v, w ∈  such that δ̄(v) = v = w
 = δ̄(w). It follows,
from the partition properties, that v ⊆ w and w ⊆ v. Hence v = w.
3. δ̄ is surjective. Since δ : E × E → 3 is surjective, ∀v ∈ 3 there is a
3
(M, N ) ∈ E × E such that δ(M, N ) = v . Further, let v ∈  such that
(M, N ) ∈ v. Thus δ̄(v) = v , and δ̄ is surjective.

This bijective mapping provides a way for translating the algebraic structure
3
of 3 into an algebraic structure of  .
3
Proposition 7. The set  together with the operations of addition and
scalar multiplication defined by
  3
v + w = δ̄ −1 δ̄(v) + δ̄(w) ∀v, w ∈  , (9)

and
  3
λ(v) = δ̄ −1 λδ̄(v) ∀λ ∈ , ∀v ∈  . (10)
form a real vector space isomorphic to 3 .
3
Proof: First, it will be proved that  is a vector space.
3
1.  is closed under addition. Let v, w ∈ 3 . Then there exits v , w
 ∈ 3
such that, for M, N, P, Q ∈ E, satisfy that (M, N ) ∈ v ⇔ δ(M, N ) = v ,
and (P, Q) ∈ w ⇔ δ(P, Q) = w.  Since 3 is a vector space (v + w)
 ∈ 3 ,
3 3
and since δ̄ :  → 3 is surjective, there is a (v + w) ∈ 
2. Addition is associative. ∀v, w, x ∈ 3
     
(v + w) + x = δ̄ −1 δ̄(v + w) + δ̄(x) = δ̄ −1 δ̄ δ̄ −1 (δ̄(v) + δ̄(w) + δ̄(x)
     
= δ̄ −1 δ̄(v) + δ̄(w) + δ̄(x) + δ̄ −1 δ̄(v) + δ̄(w) + δ̄(x)
  
= δ̄ −1 δ̄(v) + δ̄ δ̄ −1 (δ̄(w) + δ̄(x))
  
= δ̄ −1 δ̄(v) + δ̄(w) + δ̄(x) = v + (w + x).

3. Addition is commutative. ∀v, w ∈ 3


   
v + w = δ̄ −1 δ̄(v) + δ̄(w) = δ̄ −1 δ̄(w) + δ̄(v) = w + v

3 3
4. Existence of an additive identity. ∀v ∈  , ∃0 ∈  such that

v + 0 = (v + 0) = v

5
It is interesting to note that the equivalence class of E × E, corresponding
to 0, is given by
0 = {(M, N ) ∈ E × E|M = N }.
This result can be demonstrated by firstly considering (M, M ) ∈ E × E,
where M is arbitrary. From the triangle axiom

δ(M, M ) + δ(M, M ) = δ(M, M )

Therefore δ(M, M ) = 0 and (M, M ) ∈ 0. Secondly consider (M, N ) ∈


/ 0, for if (M, N ) ∈ 0, then both
E × E with M = N . Therefore (M, N ) ∈

δN (M ) = δ(M, N ) = 0, δN (N ) = δ(N, N ) = 0,

a contradiction to the injectivity of δN .


3 3
5. Existence of an additive inverse. ∀v ∈  , ∃(−v) ∈  such that

v + (−v) = [v + (−v)] = 0

It will now be demonstrated that if

v = {(M, N ) ∈ E × E|δ(M, N ) = v },

then
(−v) = {(N, M ) ∈ E × E|(M, N ) ∈ v }.
Employing the triangle axiom

δ(M, N ) + δ(N, M ) = δ(M, M )

Since (M, M ) ∈ 0 and assuming (M, N ) ∈ v, then

δ(N, M ) = −δ(M, N ) = −v .

and thus (N, M ) ∈ (−v).


3 3
6.  is closed under scalar multiplication. ∀v ∈  , ∃v ∈ 3 such that
for M, N ∈ E, (M, N ) ∈ v ⇔ δ(M, N ) = v . Further, since 3 is a
3
vector space, ∀λ ∈ , ∀v ∈ 3 , λv ∈ 3 . Moreover since δ̄ :  → 3 is
3
surjective,(λv) ∈  such that for P, Q ∈ E, (P, Q) ∈ (λv) ⇔ δ(P, Q) =
λv .
3
7. Scalar multiplication satisfies ∀λ, μ ∈  and v ∈ 
        
λ(μv) = δ̄ −1 λδ̄(μv) = δ̄ −1 λδ̄ δ̄ −1 μδ̄(v) = δ̄ −1 λ μδ̄(v)
 
= δ̄ −1 (λμ) δ̄(v) = (λμ) v.

6
3
8. Scalar multiplication and addition satisfy ∀λ, μ ∈  and v, w ∈ 
   
(λ + μ)v = δ̄ −1 (λ + μ)δ̄(v) = δ̄ −1 λδ̄(v) + μδ̄(v)
 
= δ̄ −1 δ̄(λv) + δ̄(μv) = λv + μv.

and
    
λ(v + w) = δ̄ −1 λδ̄ (v + w) = δ̄ −1 λ δ̄(v) + δ̄(w)
 
= δ̄ −1 δ̄(λv) + δ̄(λw) = λv + λw.

3
Furthermore, the mapping δ̄ :  → 3 is also linear. In fact, equations (9)
and (10), Proposition 7, can be written as
3
δ̄ [v + w] = δ̄(v) + δ̄(w) ∀ v, w ∈  ,

and
3
δ̄ [λ(v)] = λδ̄(v) ∀λ ∈ , ∀ v ∈  .
Thus δ̄ is additive and homogeneous and, therefore, it is a linear isomorphism;
3
then 3 and  are isomorphic as vector spaces.
3
Finally, it is also possible to introduce an orthogonal structure on  by
defining the symmetric bilinear form
3 3 3
 × → (v, w) = (v , w)
 ∀ v, w ∈  .

It is straightforward to show the form is well defined and indeed symmetric.


3
Further, the mapping δ̄ :  → 3 preserves the form, for trivially

(δ̄(v), δ̄(w)) = (v , w)


 (11)
3
and therefore 3 and  are isomorphic as real orthogonal spaces.
3
It is important to recognize that in the process of providing  with an
orthogonal vector space structure it is not necessary to make any arbitrary
choice of origin.

5 Bound Vector Space Algebraic Structure in E.


The Euclidean space E is endowed with a second algebraic structure induced
by means of the mapping

δ O : E → 3 δO (M ) = δ(M, O) = M → O (12)
By definition 1 of mapping δ, the mapping δO is bijective; thus it can be used
to induce a vector space, and ultimately, an orthogonal space structure in E.
Here the point O ∈ E is an arbitrary reference point or origin which, under the
mapping δO , is set equal to the zero vector.

7
In particular, it is possible to define the operations of addition and scalar
multiplication of points of E by

M + N = δO −1 (δO (M ) + δO (N )) ∀M, N ∈ E, (13)


and

λM = δO −1 (λδO (M )) ∀M ∈ E, ∀λ ∈ . (14)
It will be now be demonstrated that the points of E together with these
operations have the structure of a real vector space.

1. E is closed under addition. ∀M, N ∈ E, δO (M ), δO (N ) ∈ 3 . Since 3 is


a vector space, δO (M )+δO (N ) ∈ 3 and thus δO −1 (δO (M )+δO (N )) ∈ E.
2. Addition is commutative. ∀M, N ∈ E

M + N = δO −1 (δO (M ) + δO (N )) = δO −1 (δO (N ) + δO (M )) = N + M.

3. Addition is associative. ∀L, M, N ∈ E

(L + M ) + N = δO −1 [δO (L + M ) + δO (N )]
= δO −1 {δO [δO −1 (δO (L) + δO (M ))] + δO (N )}
= δO −1 [(δO (L) + δO (M )) + δO (N )]
= δO −1 [δO (L) + (δO (M ) + δO (N ))]
= δO −1 {δO (L) + δO [δO −1 (δO (M ) + δO (N ))]}
= δO −1 [δO (L) + δO (M + N )] = L + (M + N ).

4. Existence of an additive identity. From step 4 in the proof of proposition


7
δO (O) = δ(O, O) = 0.
Therefore, ∀M ∈ E

M + O = δO −1 (δO (M ) + δO (O)) = δO −1 (δO (M )) = M.

Thus, it has been shown that point O acts, in this structure, as the additive
identity.
5. Existence of an additive inverse. Let M ∈ E be such that δO (M ) = v ∈
3 ; then ∃ − M ∈ E such that −M = δO −1 (−v ), hence

M + (−M ) = δO −1 [δO (M ) + δO (−M )] = δO −1 [v + (−v )] = δO −1 (0) = O.

6. E is closed under scalar multiplication. ∀M ∈ E, δO (M ) ∈ 3 . Since 3 is


a real vector space, ∀λ ∈ , λδO (M ) ∈ 3 , and thus δO −1 [λδO (M )] ∈ E.

8
7. The scalar multiplication satisfies ∀λ, μ ∈  and ∀M ∈ E
λ(μM ) = δO −1 (λδO (μM )) = δO −1 {λδO [δO −1 (μδO (M ))]}
= δO −1 [(λμ)δO (M )] = (λμ)M.

8. The scalar multiplication and addition satisfy ∀λ, μ ∈ , and M, N ∈ E


(λ + μ)M = δO −1 [(λ + μ)δO (M )] = δO −1 [λδO (M ) + μδO (M )]
= δO −1 [δO (λM ) + δO (μM )] = λM + μM.
and
λ(M + N ) = δO −1 [λδO (M + N )] = δO −1 {λ[δO (M ) + δO (N )]}
= δO −1 [λδO (M ) + λδO (N )]} = λM + λN.
where the last equality, in each of two previous sequences of equalities,
holds due to the bijectivity of δO .
Rewriting equations 13 and 14 in the form
δO (M + N ) = δO (M ) + δO (N ) ∀M, N ∈ E, (15)

and
δO (λM ) = λδO (M ) ∀M ∈ E, ∀λ ∈ . (16)
3
it is evident that δO : E →  is a linear isomorphism. Additionally, by
providing E with an orthogonal structure, induced from 3 , according to
E × E → 3 (M, N ) = (δO (M ), δO (N )) ∀M, N ∈ E,
the linear isomorphism δO becomes an orthogonal isomorphism, for it trivially
preserves the form.
A prominent consideration is that some of the results obtained within the
realm of this algebraic structure are only valid for the specific selection of the
origin O. Hence, they are not Euclidean geometry properties.

6 Euclidean Mappings
In this section, an analysis of mappings of the Euclidean space which preserve
3
the orthogonal structure of  is presented. Such mappings ψ : E → E must
3
preserve the quadratic form of the associated orthogonal space,  , i.e.
(δ(M, N ), δ(M, N )) = (δ(ψM, ψN ), δ(ψM, ψN )) ∀M, N ∈ E (17)
It can be shown (Porteous [1981]) that this condition is equivalent to the preser-
vation of the associated symmetric bilinear form or inner product
(δ(M, N ), δ(P, Q)) = (δ(ψM, ψN ), (δ(ψP, ψQ)) M, N, P, Q ∈ E. (18)

9
Mappings with this property are called Euclidean mappings, and it is straight-
forward to show that this definition is independent of the particular value of μ
chosen in the mapping δμ : E × E → 3 . Furthermore, any Euclidean mapping
3
induces a mapping of  into itself according with the rule
3 3
ψ: → ψ(v) = δ(ψM, ψN ), where (M, N ) ∈ v.

Since the definition of ψ involves a choice, it will now be shown that it is indeed
well defined.
Let (M, N ), (P, Q) ∈ v, and assume δ(ψM, ψN ) = v 1 , and δ(ψP, ψQ) = v 2 ,
where v 2 can be orthogonally decomposed (Kaplansky, [1969]) as v 2 = λv 1 + u
with u ∈ [v 1 ]⊥ , the orthogonal complement of [v 1 ], it suffices to show that
v1 = v2 .

1. v 1 = 0. Then

0 = (0, 0) = (v 1 , v 1 ) = (δ(ψM, ψN ), δ(M, N )) = (δ(M, N ), δ(M, N )).

Thus δ(M, N ) = 0, and (M, N ) ∈ 0. Hence (P, Q) ∈ 0 then δ(P, Q) = 0,


and

0 = (δ(P, Q), δ(P, Q)) = (δ(ψP, ψQ), δ(ψP, ψQ)) = (v 2 , v 2 )

Thus v 1 = 0 = v 2 .
2. v 1 = 0. Then

(v 1 , v 1 ) = (δ(ψM, ψN ), δ(ψM, ψN ) = (δ(M, N ), δ(M, N ))


= (δ(M, N ), δ(P, Q)) = (δ(ψM, ψN ), δ(ψP, ψQ)) = (v 1 , v 2 )
= (v 1 , λv 1 + u) = λ(v 1 , v 1 ) + (v 1 , u) = λ(v 1 , v 1 ).

Therefore λ = 1 and thus v 2 = v 1 + u. Consider again the same relation-


ship

(v 1 , v 1 ) = (δ(ψM, ψN ), δ(ψM, ψN )) = (δ(M, N ), δ(M, N ))


= (δ(P, Q), δ(P, Q))(δ(ψP, ψQ), δ(ψP, ψM ))
= (v 2 , v 2 ) = (v 1 + u, v 1 + u) = (v 1 , v 1 ) + (u, u).

3
Therefore (u, u) = 0. Since  is positive definite, then u = 0 and v 2 = v 1 .

Furthermore, it can be proved that the induced mapping is independent of


the scale of length singled out by the mapping δ : E × E → 3 .
It is also evident that if ψ(v) = w, then the image of any v ∈ v, under the
induced mapping on 3 , will be given by w.

10
The following proposition provide more insight into the nature of the induced
linear transformation.
3
Proposition 1. The mapping ψ preserves the inner product of  .
3
Proof: Let v, w ∈  with v = δ(M, N ), w = δ(P, Q) for some M, N, P, Q ∈
E. Then

(ψv, ψw) = (δ(ψM, ψN ), δ(ψP, ψQ)) = (δ(M, N ), δ(P, Q)) = (v, w). (19)

This result is, in fact, a logical consequence of the particular way in which the
3
induced orthogonal structure of  as well as the induced mapping ψ have been
defined.
3 3
It will now be demonstrated that the induced mapping ψ :  →  is
orthogonal. Since it has already been shown that ψ preserves the inner product,
it suffices to show that ψ is a linear mapping. The proof is accomplished by
extending and slightly modifying a result reported in Angeles [1982].
Proposition 2. Every mapping of a positive definite (or negative definite)
real orthogonal space into itself which preserves the inner product is linear.
Proof: Let X be a positive definite (or negative definite) real orthogonal
space, ψ a mapping which preserves the inner product, and v , w  ∈ X, define

e = ψ(v + w)
 − ψ(v ) − ψ(w)


Then

(e, e) = (ψ(v + w)


 − ψ(v ) − ψ(w),
 ψ(v + w)
 − ψ(v ) − ψ(w))

= (ψ(v + w),
 ψ(v + w))
 + (ψ(v ), ψ(v )) + (ψ(w),
 ψ(w))

−2(ψ(v + w),
 ψ(v )) − 2(ψ(v + w),
 ψ(w)) + 2(ψ(v ), ψ(w))

= (v + w,  v + w)
 + (v , v ) + (w,
 w)
 − 2(v + w,  v ) − 2(v + w, w)
 + 2(v , w)

= (v , v ) + (w,
 w)
 + 2(v , w) + (v , v ) + (w,
 w)
 − 2(v , v ) − 2(w,
 v )
−2(v , w)  − 2(w, w)
 + 2(v , w)  = 0.

Since X is a definite orthogonal space e = 0, and

ψ(v + w)
 = ψ(v ) + ψ(w),


then ψ is additive. Further, let v ∈ X, v = 0, and λ ∈ ; then ψ(λv ) can be


expressed in the form
ψ(λv ) = αψ(v ) + w,
 ∈ [ψ(v )]⊥ . Consider now the inner product
where α ∈ , and w

λ(v , v ) = (v , λv ) = (ψ(v ), ψ(λv )) = (ψ(v ), αψ(v ) + w)



= α(ψ(v ), ψ(v )) + (ψ(v ), w) = α(ψ(v ), ψ(v )) = α(v , v )

11
Thus λ = α , and

λ2 (v , v ) = (λv , λv ) = (λψ(v ) + w,


 λψ(v ) + w)

= λ2 (ψ(v ), ψ(v )) + 2λ(ψ(v ), w)
 + (w,  = λ2 (v , v ) + 0 + (w,
 w)  w)


Therefore (w,
 w) = 0. Since X is a definite orthogonal space, then w  = 0;
thus ψ is homogeneous and linear.
Corollary 3. Every mapping of a positive definite ( or negative definite )
real orthogonal space into itself which preserves the inner product is orthogonal.
In particular, these two last results apply to mappings which preserve the
3
inner product in 3 , or in  .
After having shown that the induced Euclidean mappings are linear trans-
formations, it appears natural to inquiry whether the mappings are injective
and/or surjective. The answer to these questions can be obtained from a result
reported by Porteous [1981].
Proposition 4. Every orthogonal mapping of a non-degenerate orthogonal
space into itself is injective.
Proof: Let X be a non-degenerate orthogonal space, v , w  ∈ X, and ψ an
orthogonal map satisfying ψ(v ) = ψ(w);
 then ψ(v − w)  = 0, and
 = ψ(v ) − ψ(w)
for every u ∈ X
0 = (0, ψ(u)) = (ψ(v − w),
 ψ(u)) = (v − w,
 u)

Since X is non-degenerate,

 = 0
v − w or v = w


Corollary 5. Every orthogonal mapping of a non-degenerate,


finite-dimensional orthogonal space into itself is a bijection.
Once more, these results apply in particular to the orthogonal mappings
3
induced in  by Euclidean mappings.
Summarizing the results, it has been shown that the induced mappings of
Euclidean transformations are orthogonal automorphisms of the real orthogonal
3 3
space  . It is well known that the orthogonal automorphisms of  form a
group, called the orthogonal group and denoted by O(3). However, there are
two classes of orthogonal mappings. Those which preserve the orientation of
3
 form a subgroup of the orthogonal group, denoted by SO(3), and called
the special orthogonal group. Those which do not preserve the orientation of
3
 do not form a subgroup. Bottema and Roth [1979] show that the induced
3
orthogonal mappings must preserve the orientation of  ; hence they must be
elements of SO(3). Even though the definition of Euclidean mapping is broad
3
enough to include mappings which do not preserve the orientation of  , those
will not be considered anymore.

12
7 Properties of the Euclidean Mappings
In the previous section, Euclidean mappings and their induced transformations
were defined, but the lack of structure in the Euclidean space E hampered any
effort to directly find their characteristics. After searching the properties of the
induced mappings, it is possible to pursue the investigation of the Euclidean
ones.
3 3
Proposition 1. Let ψ : E → E be a Euclidean map, and ψ :  →  the
induced (orthogonal) map; then for any N ∈ E

ψ = (δψ(N ) )−1 ψδN . (20)


Proof: Let M ∈ E be arbitrary; then

((δψ(N ) )−1 ψδN )(M ) = (δψ(N ) )−1 ψ δN (M )


= (δψ(N ) )−1 ψ δ(M, N ) = (δψ(N ) )−1 (ψ(δ(M, N )))
= (δψ(N ) )−1 (δ(ψM, ψN ))
= (δψ(N ) )−1 (δψ(N ) )(ψ(M )) = ψ(M ).

Corollary 2. Every Euclidean mapping is bijective.


Proof: By definition δψ(N ) and (δψ(N ) )−1 are bijections and, by corollary
6.5, ψ is also a bijection, since ψ is a composition of bijective mappings, ψ is
also bijective.
Proposition 3. The composition of Euclidean mappings is also a Euclidean
mapping. Furthermore, the induced mapping is the composition of the corre-
sponding induced mappings.
Proof: Let ψ1 , ψ2 be Euclidean mappings, and ψ1 ψ2 be their associated
mappings. Let M, N ∈ E be arbitrary. Then

(δ(ψ2 ψ1 M, ψ2 ψ1 N ), δ(ψ2 ψ1 M, ψ2 ψ1 N )) = (δ(ψ1 M, ψ1 N ), δ(ψ1 M, ψ1 N ))


= (δ(M, N ), δ(M, N )).
3 3
Thus ψ2 ψ1 is also Euclidean. Assume now φ :  →  is the induced mapping
of ψ2 ψ1 , and δ(M, N ) = v be arbitrary. Then

φ(v) = δ(ψ2 ψ1 M, ψ2 ψ1 N ) = ψ2 (δ(ψ1 M, ψ1 N )) = ψ2 ψ1 (δ(M, N )) = ψ2 ψ1 v


(21)
Therefore
φ = ψ2 ψ1
Proposition 4. The inverse of a Euclidean mapping is also Euclidean,
and its induced mapping is the inverse of the induced mapping of the original
Euclidean mapping.

13
Proof: Let ψ be a Euclidean mapping, and ψ its induced mapping; then

(δ(ψM, ψN ), δ(ψM, ψN )) = (δ(M, N ), δ(M, N ))

Renaming M  = ψM and N  = ψN ; then M = ψ −1 M  , and N = ψ −1 N are,


respectively, the unique images of M  and N  under ψ −1 . Therefore

(δ(M  , N  ), δ(M  , N  )) = (δ(ψ −1 M  , ψ −1 N  ), δ(ψ −1 M  , ψ −1 N  )),

∀M  , N  ∈ E; hence ψ −1 is also Euclidean.


3 3
Assume now φ :  →  is the induced mapping of ψ −1 and δ(M  , N  ) = v ;
then

v = I 3 (v) = δ(M  , N  ) = δ(ψ1 ψ1 −1 M  , ψ1 ψ1 −1 N  )


= ψ1 (δ(ψ − 1−1 M  , ψ1 −1 N  ) = ψ1 φ(δ(M  , N  )) = ψ1 φ(v).

Thus I 3 = ψ1 φ and

φ = (ψ1 )−1 I 3 = (ψ1 )−1 . (22)


Proposition 5. The set of Euclidean mappings together with the com-
position operation form a group, called the Euclidean group, and denoted by
E(3).
Proof: By proposition 3, the composition of two Euclidean mappings is also
a Euclidean mapping; thus the set is closed under the operation.
The composition of Euclidean mappings being a special case of the compo-
sition of arbitrary mappings, which is associative, is also associative.
Consider the mapping ι : E → E such that ι(M ) = M, ∀M ∈ E. It is
straightforward to prove that ι is a Euclidean mapping, and it behaves as the
identity element of the group.
By corollary 2, every Euclidean mapping is invertible, and, by proposition
4, its inverse is also a Euclidean mapping.
Now that it has been established that Euclidean mappings constitute a
group, proposition 3 provides a proof for the following statement.
Corollary 6. The mapping Ψ : E(3) → SO(3), which assigns to every
Euclidean mapping its induced orthogonal mapping is a group homomorphism.
It is a well known fact that the composition of Euclidean mappings is not
commutative (Bottema and Roth [1979]). Therefore, the Euclidean group is not
abelian. Now, it appears natural to inquire about the possible existence and
properties of subgroups of the Euclidean group. It will be shown, in the next
two sections, that there are two important classes of subgroups of the Euclidean
group namely translations, and rotations about a fixed point.

14
8 Translations
A translation is defined as a mapping τ : E → E with the property that there
exists a v ∈ 3 such that

δ(τ M, M ) = v ∀M ∈ E (23)

More precisely, τ is called a translation of E by the vector v .


Proposition 1. A translation is a Euclidean map.
Proof: Let M, N ∈ E be arbitrary. Then, by the triangle axiom,

δ(τ M, τ N ) = δ(τ M, M )+δ(M, N )+δ(N, τ N ) = δ(τ M, M )+δ(M, N )−δ(τ N, N )

by definition δ(τ M, M ) = δ(τ N, N ). Therefore

δ(τ M, τ N ) = δ(M, N )

Hence the preservation of the quadratic form is immediate, for

(δ(τ M, τ N ), δ(τ M, τ N )) = (δ(M, N ), δ(M, N )) ∀M, N ∈ E.

The following result provides a useful characterization of a translation.


Proposition 2. A Euclidean mapping ψ is a translation if and only if
3
ψ = I 3 , where I 3 is the identity mapping in  .
Proof: Assume ψ is a translation; then by equation (2)

ψ(δ(M, N )) = δ(ψ(M ), ψ(N )) = δ(M, N ).


3
Therefore ψ = I . Assume there are two points M, N ∈ E with
δ(ψM, M ) = δ(ψN, N ). Then

ψ (δ(M, N )) = δ(ψM, ψN ) = δ(ψM, M ) + δ(M, N ) + δ(N, ψN )


= δ(M, N ) + [δ(ψM, M ) − δ(ψN, N )]

Since δ(ψM, M ) = δ(ψN, N ), then

ψ(δ(M, N )) = δ(M, N ), and, ψ = I 3 .

Proposition 3. The subset of all translation is closed under the composition


operation. Furthermore, composition of translations is commutative.
Proof: Let τ1 , τ2 be two translations, with v1 , v2 ∈ 3 their associated
vectors, and M ∈ E be arbitrary. Then using the triangle axiom

δ(τ2 τ1 M, M ) = δ(τ2 τ1 M, τ1 M ) + δ(τ1 M, M ) = v2 + v1 .

Thus τ2 τ1 is a translation, and the associated vector is the sum of the associated
vectors. Moreover

δ(τ2 τ1 M, M ) = δ(τ2 τ1 M, τ2 M ) + δ(τ2 M, M ) = v2 + v1 .

15
Since addition in 3 is commutative, v1 + v2 = v2 + v1 , and

δ(τ2 τ1 M, τ1 τ2 M ) = δ(τ2 τ1 M, M ) + δ(M, τ1 τ2 M )


= δ(τ2 τ1 M, M ) − δ(τ1 τ2 M, M )
= (v2 + v1 ) − (v1 + v2 ) = 0.

Thus

τ2 τ1 M = τ1 τ2 M ∀M ∈ E

Proposition 4. Let τ : E → E be the translation of E by a vector v ; then


τ −1 : E → E is the translation of E by the vector −v .
Proof: Let M ∈ E be arbitrary, with τ M = N , and τ −1 N = τ −1 τ M = P ,
for some P ∈ E. Applying the triangle equality

δ(P, M ) = δ(P, N ) + δ(N, M ) = δ(τ −1 N, N ) + δ(τ M, M ) = −v + v = 0.

Thus P = M , or τ −1 τ M = M , hence τ −1 τ = ι, and this result coupled with


the commutativity of the composition proves the assertion. Here, ι denotes the
identity Euclidean mapping.
Proposition 5. The set of all translations forms a normal abelian subgroup,
denoted by T , of the Euclidean group.
Proof: By proposition 3, the set is closed under composition, and the com-
position is commutative. By proposition 4, the set is closed under the operation
of taking inverses. Finally, let τ and ψ be an arbitrary translation and Euclidean
mapping respectively; consider the mapping φ = ψτ ψ −1 ; then

φ = ψτ ψ −1 = ψ τ (ψ)−1 = ψ I 3 (ψ)−1 = I 3 .

Thus by proposition 2, φ is a translation.


Further, with the usual definition of a scalar multiple of a mapping, this
group can be made isomorphic to the vector space 3 .

9 Rotations.
A Euclidean mapping ρ : E → E is called a rotation if there exists a point
P ∈ E such that ρP = P . P is then called a fixed point of the rotation.
Proposition 1. The set of rotations having a common fixed point P , de-
noted by ΩP , forms a subgroup of the Euclidean group.
Proof: Let ρ1 , ρ2 ∈ ΩP ; then

ρ2 ρ1 P = ρ2 (ρ1 P ) = ρ2 P = P.

and ρ2 ρ1 ∈ ΩP . Trivially, the identity mapping, ι : E → E, belongs to ΩP for


ιP = P . Let ρ ∈ ΩP ; then

P = ιP = (ρ−1 ρ)P = ρ−1 (ρP ) = ρ−1 P.

16
Thus, ρ−1 ∈ ΩP and the set is closed under the operation of taking inverses.
It is noteworthy to recognize that for any point P ∈ E there is a subgroup
ΩP of the rotations that leave P fixed. Sometimes the fixed point will be used
as a suffix of a rotation as a way of specifically stating the invariance of that
point.
The following result characterizes the equivalence of two rotations.
Proposition 2. Two rotations ρ1 and ρ2 are equal if, and only if, they have
a common fixed point and the same induced orthogonal mapping.
Proof: Let P be the common fixed point, and consider an arbitrary M ∈ E;
then

δ(ρ1 M, ρ2 M ) = δ(ρ1 M, ρ1 P ) + δ(ρ1 P, ρ2 P ) + δ(ρ2 P, ρ2 M )


= ρ1 δ(M, P ) + δ(P, P ) + ρ2 δ(P, M )
= ρ1 [δ(M, P ) + δ(P, M )] + 0
= ρ1 [δ(M, M )] = ρ1 (0) = 0.

Thus, ρ1 (M ) = ρ2 (M ), ∀M ∈ E, and ρ1 = ρ2 .
Assume ρ1 and ρ2 do not have a common fixed point, then if P is any fixed
point of ρ1 , ρ1 P = P ; however by assumption ρ2 P = P . Hence ρ1 P = ρ2 P and
ρ1 = ρ2
Finally, assume ρ1 and ρ2 have P as a common fixed point but ρ1 = ρ2 then
3
∃v ∈  such that ρ1 (v) = ρ2 (v). Let M ∈ E be such that δP (M ) = v ; then
applying the triangle equality

δ(ρ1 M, ρ2 M ) = δ(ρ1 M, P ) + δ(P, ρ2 M ) = δ(ρ1 M, ρ1 P ) − δ(ρ2 M, ρ2 P )


= ρ1 δ(M, P ) − ρ2 δ(M, P ) = ρ1 (v) − ρ2 (v) = 0.

Thus ρ1 M = ρ2 M and ρ1 = ρ2 .
If a rotation ρ : E → E is also a translation, then the vector associated with
the translation is given – employing the fixed point P of ρ – by

δ(ρP, P ) = δ(P, P ) = 0

It follows that for every M ∈ E

δ(ρM, M ) = 0; thus ρM = M

Therefore ρ is the identity mapping. Conversely, the identity mapping is the


unique
Euclidean mapping which is both a rotation and a translation; hence
ΩP T = {ι}.
Given a non-identity rotation ρ : E → E, then it can be shown that the
3
induced mapping ρ is a non-identity proper orthogonal transformation of  .
Further, from the theory of proper orthogonal mappings (Herstein [1975]), 1 is

17
the unique real eigenvalue of any non-identity proper orthogonal mapping of
3
 ; Euler’s theorem will now be proved using these results.
Proposition 3 (Euler’s Theorem). Every rotation ρ : E → E has asso-
ciated a pointwise fixed affine line, which is called the rotation axis of ρ.
3
Proof: Let P ∈ E be any fixed point of ρ, and v ∈  be an eigenvector
associated with the eigenvalue 1; consider Q ∈ E given by

δ(Q, P ) = λv , for some λ ∈ 

Then applying the triangle equality

δ(ρQ, Q) = δ(ρQ, ρP ) + δ(ρP, P ) + δ(P, Q)


= ρ[δ(Q, P )] + δ(P, P ) + δ(P, Q)
= ρ(λv ) + 0 − δ(Q, P ) = λv − λv = 0.

Thus ρQ = Q. Furthermore, the pointwise fixed affine line is given by

Ra = {Q ∈ E|δ(Q, P ) = λv , for some λ ∈ }

Proposition 4. The only invariant points under a rotation ρ are those along
its rotation axis.
Proof. Let M ∈ E be a point not along the rotation axis of ρ ∈ ΩP ; i.e.
M ∈ / Ra , then δ(M, P ) = λv + w,  ∈ [v ]⊥ . Assume, by contradiction,
 where w
that M is invariant under ρ, then

0 = δ(ρM, M ) = δ(ρM, ρP ) + δ(ρP, M ) = ρ(δ(M, P )) + δ(P, M )


= ρ(λv + w)
 − (λv + w)
 = λv + ρw
 − λv − w.


Thus ρ(w)
 = w. Thus {v, w}
 forms a basis for the eigenspace associated with the
eigenvalue 1. A contradiction to the properties of a rotation. Geometrically, this
result would imply that any rotation has an associated pointwise fixed plane.
Proposition 5. Consider the subgroup of rotations around a fixed point P ,
ΩP . Denote RP,v , the subset of ΩP that has the affine line

Ra = {Q ∈ E|δ(Q, P ) = λv , for some λ ∈ }

as its rotation axis; i.e. as a pointwise fixed line. Then RP,v < ΩP .
Proof: Let ρ1 , ρ2 ∈ RP,v . Since ρ1 (P ) = ρ2 (P ) = P , it is clear that
RP,v ⊂ ΩP . Further, consider any Q ∈ Ra

δ(ρ2 ρ1 Q, Q) = δ(ρ2 ρ1 Q, ρ1 Q) + δ(ρ1 Q, Q) = δ(ρ2 (ρ1 Q), (ρ1 Q)) + δ(Q, Q)


= δ(ρ2 Q, Q) + 0 = 0 + 0 = 0.

Thus ρ2 ρ1 Q = Q. Thus Ra is also a pointwise fixed affine line; i.e. a rotation


axis of ρ2 ρ1 hence ρ2 ρ1 ∈ RP,v .

18
By assumption δ(ρ1 Q, Q) = 0, ∀Q ∈ Ra . consider

0 = ρ1 −1 (0) = ρ1 −1 [δ(ρ1 Q, Q)] = δ(ρ1 −1 ρ1 Q, ρ1 −1 Q)


= δ(Q, ρ1 −1 Q) = −δ(ρ1 −1 Q, Q)

Thus δ(ρ1 −1 Q, Q) = 0 and ρ1 −1 Q = Q. Thus ρ1 −1 ∈ RP,v . Hence, the set is


closed under composition and under taking inverses, and it is a subgroup.
Proposition 6. Consider the subgroup of rotations around a fixed point P ,
ΩP . Let ρ1 , ρ2 ∈ ΩP , so that v1 , v2 are eigenvectors of the orthogonal induced
mappings ρ1 , ρ2 , respectively. Furthermore, assume that {v1 , v2 } are linearly
independent. The only invariant point of E under ρ1 and ρ2 is P .
Proof: That P is an invariant point under ρ1 and ρ2 is obvious since ρ1
and ρ2 ∈ ΩP . Assume that M ∈ E, with M = P is invariant under ρ1 and ρ2 .
Then,

δ(ρ1 M, M ) = 0 ⇒ δ(M, P ) = λ1v1 , with λ1 = 0.

and

δ(ρ2 M, M ) = 0 ⇒ δ(M, P ) = λ2v2 , with λ2 = 0.

Then
0 = δ(M, M ) = δ(M, P ) + δ(P, M ) = δ(M, P ) − δ(P, M ) = λ1v1 + λ2v2 .

A contradiction of the linear independence of {v1 , v2 }.


Proposition 7. The only class of proper subgroups of ΩP are RP,v , where
v ∈ 3 .

10 Decomposition of Euclidean Mappings


After establishing that translations and rotations around a fixed point are sub-
groups of the Euclidean group, it appears natural to ask whether it is possible
to decompose an arbitrary Euclidean mapping in terms of a translation and a
rotation. The answer is in the affirmative.
Proposition 1. Every Euclidean mapping can be written as the composition
of a translation and a rotation whose fixed point is arbitrarily chosen. Moreover,
the decomposition is unique up to the selected fixed point of the rotation.
Proof: Let ψ : E → E be an arbitrary Euclidean map, and P ∈ E be an
arbitrary point. Denote v = δ(ψP, P ) and define

ρP : E → E ρP = (τv )−1 ψ = τ−v ψ.

Then

δ(ρP P, P ) = δ(ρP P, ψP ) + δ(ψP, P ) = δ(τ−v ψP, ψP ) + v = −v + v = 0.

19
Thus ρP = P , and ρP is indeed a rotation leaving the point P fixed. Hence

ψ = τv ρP . (24)
There are two special cases of this proof. If δ(ψP, P ) = 0, then ψP = P , and ψ
is a rotation. Therefore trivially
ψ = ιψ
If there is a v ∈ 3 such that δ(ψM, M ) = v ∀M ∈ E, then ψ is a translation
and trivially
ψ = ψι
Now assume that ψ has two decompositions with the same fixed point P

ψ = τv ρP and ψ = τw (ρP )∗


Since v = δ(ψP, P ) = w,  then τv = τw ; therefore τv ρP = ψ = τv (ρP )∗ , and
(τv ) (τv ρP ) = (τv ) (τv (ρP )∗ ); thus
−1 −1

ρP = (ρP )∗
This result completes the characterization of the Euclidean group, as given
by the following proposition
Proposition 2 The Euclidean group, E(3), is the semidirect product of the
subgroup of translations, T , times the subgroup of rotations ,ΩP , around an
arbitrary point P ∈ E. Namely

E(3) = T ⊗ ΩP
Proof: By proposition 5, section 8, translations form a normal subgroup of
the Euclidean group; i.e. T E(3). Further, given an arbitrary point P ∈ E,
it follows, see section 9, that ΩP ∩ T = {ι}. Finally, proposition 1, has shown
that for an arbitrary ψ ∈ E(3), ψ = τv ρP , where τv ∈ T and ρP ∈ ΩP . Hence,
T ΩP = E(3). Thus, the result follows, see Rotman [1973].
Proposition 3. Every Euclidean mapping can be written as the composition
of a rotation whose fixed point is arbitrarily chosen, and a translation. Moreover,
the decomposition is unique up to the selected fixed point of the rotation.
Proof: Let ψ : E → E be an arbitrary Euclidean map, and P ∈ E be an
arbitrary point. Define

ρ(ψP ) = ψτw −1 = ρ(ψP ) τ−w .


with w
 = δ(ψP, P ). Then consider

δ(ρ(ψP ) ψP, ψP ) = δ(ρ(ψP ) ψP, ρ(ψP ) P ) + δ(ρ(ψP ) , ψP )


= ρ(ψP ) δ(ψP, P ) + δ(ρ(ψP ) P, ρ(ψP ) τw P )
= ρ(ψP ) w
 + ρ(ψP ) [δ(P, τw P )] = ρ(ψP ) w
 − ρ(ψP ) [δ(P, τw P )]
= ρ(ψP ) w  = 0
 − ρ(ψP ) w

20
Thus ρ(ψP ) ψP = ψP , and ρ(ψP ) is, indeed a rotation.
Moreover by proposition 5.3
ρP = I 3 ρP = τ v ρP = ψ = 1364ρ(ψP ) τ w = ρ(ψP ) I 3 = ρ(ψP )
Thus,
ρP = ρ(ψP ) .
Similarly, it is a routine task to verify that if a Euclidean mapping is decom-
posed as
ψ = τv ρP or ψ = ρ(ψP ) τw
Then, the inverse mapping can be decomposed as
ψ −1 = τ−w (ρ(ψP ) )−1 , or ψ −1 = (ρP )−1 τ−v .
Although the selection of different points, as fixed points of the rotation,
leads to distinct decompositions, it will be proved, analogously to the proof
equation 3, that the induced orthogonal mapping remains invariant.
Proposition 3. The induced orthogonal mapping of an arbitrary Euclidean
mapping is independent of the fixed point chosen to accomplish the decomposi-
tion.
Proof: Let ψ : E → E be an arbitrary Euclidean mapping with two distinct
decompositions
ψ = τv ρP and ψ = τw ρQ .
Then, by proposition 5.3,
ρP = I 3 ρP = τ v ρP = ψ = τ w ρQ = I 3 ρQ = ρQ
This result coupled with the invariance of the induced orthogonal mapping
under changes of the scale of length ensures that the induced orthogonal map-
ping is a true Euclidean property of any Euclidean mapping.
On the other hand, the selection of different points, as fixed points of the
rotation, also leads to distinct values of the translation vector; however, those
vectors still have interesting invariant characteristics whose investigation leads
up to some of the results credited to Rodrigues (Gray [1980]), and Chasles (Bot-
tema and Roth [1979]), and ultimately to the screw representation of Euclidean
mappings.
Proposition 4. Let ψ be a Euclidean mapping decomposed as ψ = τv ρP ,
and ψ = τw ρQ , where P, Q are two arbitrary distinct points and let u be an
eigenvector related to the eigenvalue 1 of the common induced orthogonal map-
ping. The components of the translation vector along the eigenvector are equal.
Proof: It is obvious that (v , u) = (w,  u) = 0. Thus it suffices
 u) ⇔ (v − w,

to show that (v − w)
 ∈ [u] .
By definition v = δ(ψP, P ) and w  = δ(ψQ, Q) . Hence by invoking the
triangle equality
v − w
 = δ(ψP, P ) − δ(ψQ, Q)
= δ(ψP, ψQ) + δ(ψQ, Q) + δ(Q, P ) − δ(ψQ, Q)
= ψ[δ(P, Q)] − δ(P, Q)

21
Expressing δ(P, Q) = λu + y with y ∈ [u]⊥ , then

v − w
 = ψ(λu + y ) − (λu + y ) = λψu + ψy − λu − y

Since by assumption ψu = u, then

v − w
 = λu + ψy − λu − y = ψy − y .

Finally, since ψ is orthogonal, then

(v − w,
 u) = (ψy − y, u) = (ψy , u) − (y , u)
= (ψy , ψu) − (y , u) = (y , u) − (y , u) = 0.

 ∈ [u]⊥ , which establishes the result.


Thus (v − w)
The component of the translation vector, along the eigenvector u, which by
the previous result is independent of the fixed point chosen during the decom-
position. It does however depend on the scale of length chosen in the mapping
δ : E × E → 3 , and thus, it is not a Euclidean property of the mapping. In
view of this result, the question arises whether there is a point S such that
ψ = τr ρS with r = λu, for some λ ∈ , and if it exists then where is it located.
Proposition 5. If a point S, satisfying the conditions shown above, exists
then all the points on the affine line

{Pψ |δ(P, S) = μu for some μ ∈ } (25)


satisfy the same conditions. Additionally, all the points belonging to this line
are displaced equally by the Euclidean mapping ψ.
Proof: Consider an arbitrary point P belonging to the affine line given by
equation 5. Then

δ(ψP, P ) = δ(ψP, ψS) + δ(ψS, S) + δ(S, P )


= ψ[δ(P, S)] + λu − δ(P, S)
= ψ(μu) + λu − μu = μu + λu − μu = λu.

Proposition 6. Let ψ be a Euclidean mapping; then for any decomposition


ψ = τw ρQ the restriction mapping I 3 − ρQ : [u]⊥ → [u]⊥ is a bijective linear
mapping.
Proof: Let y ∈ [u]⊥ ; then

((I 3 − ρQ )y , u) = (y − ρQ y, u) = (y , u) − (ρQ y, u)
= (y , u) − (ρQ y, ρQ u) = 0 − (y , u) = 0.

Thus the transformation indeed maps [u]⊥ into [u]⊥ .


Assume that v ∈ Ker(I 3 − ρQ ); then

0 = (I 3 − ρQ )y = y − ρQ y ⇔ ρQ y = y .

22
If y = 0, then the result follows for the function is injective and consequently
bijective. Assume y = 0; then y is an eigenvector of ρQ associated with the
eigenvalue 1; hence  y = μu for some μ ∈  a contradiction to y ∈ [u]⊥ .
Proposition 7. Let ψ : E × E be an arbitrary Euclidean mapping; then a
unique affine line can be found such that for all the points S that belong to the
line
ψ = τr ρS with r = λu.
where u is an eigenvector associated with the eigenvalue 1 of the induced or-
thogonal mapping.
Proof: By proposition 5, it suffices to find a point in the affine line. Let Q
be an arbitrary point. Then
ψ = τw ρQ .
Let u be an eigenvector associated with the eigenvalue 1 of ρQ . Then w
 = μ1 u+y

with μ1 ∈ , and y ∈ [u] .
Let S be another arbitrary point. Then δ(S, Q) = μ2 u + x with μ2 ∈  and
x ∈ [u]⊥ . Applying the triangle equality

δ(ψS, S) = δ(τw ρQ S, S) = δ(τw ρQ S, ρQ S) + δ(ρQ S, S)


= w
 + δ(ρQ S, ρQ Q) + δ(Q, S) = w  + ρQ (δ(S, Q)) − δ(S, Q)
=  + ρQ (μ2 u + x) − (μ2 u + x)
w
= μ1 u + y + μ2 u + ρQ (x) − μ2 u − x
= μ1 u + y − (I 3 − ρQ )x

By proposition 6, the mapping (I 3 − ρQ ) : [u]⊥ → [u]⊥ is bijective. Hence a


solution of
(I 3 − ρQ )x = y
always exists, and it is unique. Hence, a point S can be found such that

r = δ(ψS, S) = μ1 u.

This affine line is commonly referred as the screw axis of the Euclidean
mapping. It is certainly a Euclidean invariant of the Euclidean mapping.

11 Composition of Euclidean Mappings


Now that the decomposition and properties of an arbitrary Euclidean mapping
have been examined, it seems natural to inquire how the composition of two
arbitrary Euclidean mappings can be related to this decomposition. This anal-
ysis provides the necessary foundation for a further result associated with the
structure of the Euclidean group.
Proposition 1. Let ψ1 = τtρO and ψ2 = τv (ρO )∗ be two arbitrary Euclidean
mappings decomposed using a common point O. Then

23
ψ2 ψ1 = τv τ(ρO )∗t(ρO )∗ ρO (26)
Proof: Consider ψ2 ψ1 = (τv (ρO )∗ )(τtρO ) = τv ((ρO )∗ τt)ρO . Then (ρO )∗ τt
is a Euclidean mapping transforming point O into (ρO )∗ M where δ(M, O) = t.
According with proposition 8.1
(ρO )∗ τt = τw (ρO )+
where
 = δ((ρO )∗ M, O) = ((ρO )∗ M, (ρO )∗ O) = (ρO )∗ δ(M, O) = (ρO )∗ t.
w

and
(ρO )∗ = (ρO )∗ I 3 = (ρO )∗ τt = τw (ρO )+ = I 3 (ρO )+ = (ρO )+
Then, by proposition 7.2
(ρO )∗ = (ρO )+ , and (ρO )∗ τt = τ(ρO )∗t(ρO )∗

Thus finally

ψ2 ψ1 = τv τ(ρO )∗t(ρO )∗ ρO

Proposition 2. The quotient group E(3)/T is isomorphic to ΩP , where P


is an arbitrary point of E.
Proof: Consider the mapping
ΦP : E(3) → ΩP ΦP ψ = ρP where ψ = τtρP .
It is easy to note that ΦP assigns to an arbitrary Euclidean mapping the rotation
part in its decomposition with respect to the point P . It will now be shown
that ΦP is a group homomorphism onto ΩP with kernel T , then resorting to the
so-called isomorphisms theorems (Herstein [1975] pp.59) the result will follow.
Let ψ1 , ψ2 ∈ E(3), where ψ1 = τtρP and ψ2 = τv (ρP )∗ . Then

ΦP (ψ2 ψ1 ) = ΦP (τv τ(ρP )∗t(ρP )∗ ρP ) = (ρP )∗ ρP = ΦP (ψ2 )ΦP (ψ1 )

Let ρP ∈ ΩP be arbitrary; then trivially ρP ∈ E(3) and

ΦP (ρP ) = ΦP (ιρP ) = ρP
Thus the mapping is surjective. Finally

ψ ∈ KerΦP ⇔ ΦP (ψ) = ι ⇔ ψ = τt ι for some t ∈ 3


⇔ ψ = τt for some t ∈ 3 ⇔ ψ ∈ T.
This result indicates that E(3) is an extension of the normal subgroup of
translations, T , by ΩP , the subgroup of rotations around an arbitrary fixed point
P ∈ E. The extension is just an alternative characterization of a semidirect
product (Rose [1978] pp. 208-210, and Rotman [1973]).

24
12 Equivalence Conditions of Euclidean Map-
pings
The aim of this section is to find necessary and sufficient conditions for a pair
of Euclidean mappings to be equal. At the outset, the question seems trivial.
Euclidean mappings are functions defined on the physical space, then it follows
that

ψ1 = ψ2 ⇔ ψ1 (P ) = ψ2 (P ) ∀P ∈ E.

However, suppose that ψ1 and ψ2 are given as their decomposition using a pair
of distinct points P and Q of E; namely,

ψ1 = τv ρP and ψ1 = τw ρP

where v = δ(ψ1 (P ), P ) and w


 = δ(ψ2 (P ), P ). Then, the question naturally
arises, what are the necessary and sufficient conditions, expressed in terms of
these decompositions, for ψ1 and ψ2 to be equal? The answer to that question
constitutes the following proposition.
Proposition 1 Let ψ1 and ψ2 be a pair of Euclidean mappings such that

ψ1 = τv ρP and ψ2 = τw ρQ ,

where P and Q are arbitrary points of E, and

v = δ(ψ1 (P ), P ) w
 = δ(ψ2 (P ), P ) r = δ(P, Q).

Then ψ1 = ψ2 = ψ if, and only if,

1. ψ 1 = ρP = ρQ = ψ 2 = ψ.

 = (ψ − I 3 )r.
2. v − w

Proof: Assume that ψ1 = ψ2 = ψ, then

ψ = ψ 1 = τv ρP = τv ρP = ρP = ρQ = τw ρQ = τw ρQ = ψ 2 ,

and the first part of the proof is finished. Consider now

w
 = δ(ψQ, Q) = δ(ψQ, ψP ) + δ(ψP, P ) + δ(P, Q) = ψ(δ(Q, P )) + v + r
= ψ(−r) + r + v = (−ψ + I 3 )r + v ,

therefore

v − w
 = (ψ − I 3 )r.

In the opposite direction, assume that both conditions are satisfied, and let
M ∈ E be arbitrary. Then

25
δ(ψ1 M, M ) = δ(τv ρP M, M ) = δ(τv ρP M, ρP M ) + δ(ρP M, ρP P ) + δ(ρP P, M )
= v + ρP δ(M, P ) + δ(P, M ) = v + ρP δ(M, P ) − δ(M, P )
= v + (ρP − I 3 )δ(M, P ),

similarly

δ(ψ2 M, M ) = δ(τw ρQ M, M ) = δ(τw ρQ M, ρQ M ) + δ(ρQ M, ρQ Q) + δ(ρQ Q, M )


=  + ρQ δ(M, Q) + δ(Q, M ) = w
w  + ρQ δ(M, Q) − δ(M, Q)
= w
 + (ρQ − I 3 )δ(M, Q).

Then

δ(ψ1 M, ψ2 M ) = δ(ψ1 M, M ) + δ(M, ψ2 M ) = δ(ψ1 M, M ) − δ(ψ2 M, M )


= v + (ρP − I 3 )δ(M, P ) − w
 − (ρQ − I 3 )δ(M, Q).

However, from the first condition

ψ = ρP = ρQ ,

then

δ(ψ1 M, ψ2 M ) = v + (ψ − I 3 )δ(M, P ) − w
 − (ψ − I 3 )δ(M, Q)
= v − w
 + (ψ − I 3 ){δ(M, P ) − δ(M, Q)}
 + (ψ − I 3 ){δ(M, P ) + δ(Q, M )}
= v − w
= v − w
 + (ψ − I 3 ){δ(Q, P )} = v − w
 + (ψ − I 3 )(−r)

 − (ψ − I 3 )r = 0,
= v − w

where this last result follows from the second condition. Finally, since

δ(ψ1 M, ψ2 M ) = 0, ∀M ∈ E,

then
ψ1 M = ψ2 M ∀M ∈ E.
Thus, ψ1 = ψ2 .
It is interesting to note that with a coordinate free approach, the necessary
and sufficient conditions for a pair of Euclidean mappings are relatively simple.

26
References
[1] Angeles, J. 1982, Spatial Kinematic Chains: Analysis, Synthesis and Opti-
mization, New York : Springer Verlag.
[2] Bottema, O. and Roth B. 1979, Theoretical Kinematics, New York : North
Holland Pub. Co.
[3] Gray, J.J. 1980, “Olinde Rodrigues’ Paper of 1840 on Transformation
Groups”, Archive for History of Exact Sciences, Vol. 21 pp. 375-385.
[4] Herstein, I.N. [1975], Topics in Algebra, 2nd. ed. Lexington Mass : Xerox
College Pub.
[5] Kaplansky I. [1969], Linear Algebra and Geometry: A second Course, Boston
: Allyn and Bacon.
[6] Porteous, I.R. [1981], Topological Geometry, Cambridge : Cambridge Uni-
versity Press.
[7] Rose, J.S. [1978] A Course on Group Theory, Cambridge : Cambridge Uni-
versity Press.
[8] Rotman, J.J. [1973] The Theory of Groups, an Introduction, 2nd. ed., Boston
: Allyn and Bacon.

27

Você também pode gostar