not really: The quaternion algebra is a THREE dimensional Lie algebra over a field K generated f.i. by your i,j,k with anticommutativ multiplication and Jacobi identity (+/-1 is not in the algebra). The quaternion group is SU(2) , the group of unitary 2x2 matrices with determinant 1 (sometimes: U(1)). If ai+bj+ck is a general element of the now REAL Lie algebra then exp[I(ai+bj+ck)], with I² = -1, is an element of the quaternion group in the neighborhood of the identity.
o.k., sorry, I was concentrating on Lie algebras and groups.
However, 1st, define a composition algebra C over a field K, nonassociative and a map called the norm N(X) of C, X element of C: 3 nonstandard properties (!), but see below for an example.
Next, define pairs by the Cayley-Dixon formulas, (X,Y)(Z,W) = … = (A,B), and N’((X,Y)) = N(X) – aN(Y), and call this algebra the quadratic algebra C, a is element of K
Next, C is called generalized quaternion algebra and the C are called generalized Cayley algebra.
Further, R are “the” quaternion” numbers, and the algebra of 2x2 matrices over K is a split quaternion algebra, the norm is the determinant.
Your quaternion group is a discrete subset of this quaternion algebra, just take the standard Pauli spin matrices and the unit matrix.
I will be grateful if you have some references in what you gave in your answer.
Dear Wiwat, yes it is a skew field (division ring), I think Tandeusz used the historical term for skew fields (which considered differs from fields by commutativity )
Thank you dear Janab so much, this is an other substantial relation between them. in addition to the Quaternion group is a discrete subset of this quaternion algebra as showed Anton in his answer.
quaternion algebra H is not a group algebra for the classical quaternion group Q_8 since it is NOT 8-dimensional (Q_8 contains 8 elements). There is only a representation of Q_8 in H.
quaternion algebra H is not the group algebra for the classical quaternion group Q_8 since it is NOT 8-dimensional (Q_8 contains 8 elements). There is only a representation of Q_8 in H.
Yes, I searched about that point in time. I found the notion: "Algebra of a finite group, in French ( l'algèbre d'un groupe fini ), this one requires that the dimension of the algebra is the order of the group. The second notion is the group algebra of a group as said Janab, it meant taking all the linear combinations of the elements of the group through a field, just short information doesn't talk about the dimension. My question: are both notions the same?
Now, I think if we take the second notion, we get an algebra a 4- dimensional algebra, because the genrators of Q8 are the basis of H.
For he first notion, I think we have 8-dimensional, because we work on the real field then on the complex field, so the algebra of a group of order n becomes 2n-dimensional.
in general, if you have an algebra A over a field F and some group G of invertible elements in A, you encounter two DIFFERENT but closely related objects.
1. The group algebra FG consisting of formal finite linear combinations of g in G with coefficients in F. Always dim FG= card G. Usually, FG\ne A.
2. The algebra A itself as a natural FG-module via inclusion G \subset A. It is equivalent to say that either G is represented in A, or that G acts on A by (left) shifts a->ga, since there is a multiplication in A.
For quaternion algebra H=A and quaternion group Q=G we have respectively
1. The 8-dimensional (over the field R of real numbers) group algebra RQ for which the basic elements '1' and '-1' in Q are linearly independent.
2. The 4-dimensional (over the field R of real numbers) quaternion algebra H for which the quaternions 1 and -1 are linearly dependent: 1+(-1)=0.
Please pay attention to the same difference for cases A=R and G={-1;1} or A= Mat(n,R), G=GL(n,R).
in general, if you have an algebra A over a field F and some group G of invertible elements in A, you encounter two DIFFERENT but closely related objects.
1. The group algebra FG consisting of formal finite linear combinations of g in G with coefficients in F. Always dim FG= card G. Usually, FG\ne A.
2. The algebra A itself as a natural FG-module via inclusion G \subset A. It is equivalent to say that either G is represented in A, or that G acts on A by (left) shifts a->ga, since there is a multiplication in A.
For quaternion algebra H=A and quaternion group Q=G we have respectively
1. The 8-dimensional (over the field R of real numbers) group algebra RQ for which the basic elements '1' and '-1' in Q are linearly independent.
2. The 4-dimensional (over the field R of real numbers) quaternion algebra H for which the quaternions 1 and -1 are linearly dependent: 1+(-1)=0.
Please pay attention to the same difference for cases A=R and G={-1;1} or A= Mat(n,R), G=GL(n,R).
in general, if you have an algebra A over a field F and some group G of invertible elements in A, you encounter two DIFFERENT but closely related objects.
1. The group algebra FG consisting of formal finite linear combinations of g in G with coefficients in F. Always dim FG= card G. Usually, FG\ne A.
2. The algebra A itself as a natural FG-module via inclusion G \subset A. It is equivalent to say that either G is represented in A, or that G acts on A by (left) shifts a->ga, since there is a multiplication in A.
For quaternion algebra H=A and quaternion group Q=G we have respectively
1. The 8-dimensional (over the field R of real numbers) group algebra RQ for which the basic elements '1' and '-1' in Q are linearly independent.
2. The 4-dimensional (over the field R of real numbers) quaternion algebra H for which the quaternions 1 and -1 are linearly dependent: 1+(-1)=0.
Please pay attention to the same difference for cases A=R and G={-1;1} or A= Mat(n,R), G=GL(n,R).
I think that the relations [1]+[-1]=0, [i]+[-i]=0, [j]+[-j]=0, [k]+[-k]=0 are needed, brackets standing for copies of elements of Q in RQ (in order to not confound them with quaternions).
If you prefer the relations [i]^2=-[1], ... should be written since the relation, say, [i]^2=[-1] holds in G or RG and "kills" nothing.
in RQ they are independent but in H algebra they are not independent so dimH = 4 and dim RQ = 8 which is reason im wrong not killing by group relations
As you said Quaternion group is a set with 8 elements {+1, _1, i,-i, j, -j, k, -k}, where the operations defined as above. It is non abelian
But Quaternion algebra is a special case of a general algebra.
Let F be a field of characteristic not equal to 2. A be a 4 dimensional (right)vector space over F with basis {1, i, j, k} Let a, b be two non zero elements of F. Now we define multiplication in A, so that is is bilinear and associative as below.
i2 = a, j2= b, ij = -ji = k, k2 =ab, ik = -ki = ja, jk = -kj = -ib. The element a of F is identified with 1a of A and similar identification for other elements of F Then A is a (right) algebra over F.
Now take F to be the field of real numbers and a=1 and b=-1. The algebra now we get is Quaternion Algebra. Its dimension is 4
As you said Quaternion group is a set with 8 elements {+1, _1, i,-i, j, -j, k, -k}, where the operations defined as above. It is non abelian
But Quaternion algebra is a special case of a general algebra.
Let F be a field of characteristic not equal to 2. A be a 4 dimensional (right)vector space over F with basis {1, i, j, k} Let a, b be two non zero elements of F. Now we define multiplication in A, so that is is bilinear and associative as below.
i2 = a, j2= b, ij = -ji = k, k2 =ab, ik = -ki = ja, jk = -kj = -ib. The element a of F is identified with 1a of A and similar identification for other elements of F Then A is a (right) algebra over F.
Now take F to be the field of real numbers and a=1 and b=-1. The algebra now we get is Quaternion Algebra. Its dimension is 4