From: Dave Rusin
Date: Thu, 22 Oct 1998 21:02:47 -0500 (CDT)
To: edgar@math.ohio-state.edu
Subject: Re: linear algebra
Newsgroups: sci.math.research
In article you write:
>(1) Let A be an algebra over a field F of dimension n. If B
> is a subalgebra, must the dimension of B be a divisor of n?
Appropriate for the class: take M to be a 3x3 matrix whose minimal polynomial
is X^3. Then let A be the subalgebra of M_3(F) generated by I and M,
and B the subalgebra generated by I and M^2. Their dimensions are
3 and 2 respectively.
Or let A = F^3 and B = F^2 + (0).
On the other hand, if A and B are simple algebras over F then
yes, [B:F] | [A:F].
dave
==============================================================================
From: Dave Rusin
Date: Thu, 22 Oct 1998 21:23:49 -0500 (CDT)
To: edgar@math.ohio-state.edu
Subject: linear algebra: correction.
I wrote in haste.
>>(1) Let A be an algebra over a field F of dimension n. If B
>> is a subalgebra, must the dimension of B be a divisor of n?
[...]
>Or let A = F^3 and B = F^2 + (0).
That's not really right. Usually the definition of "algebra" includes
having a unit element, so that F is contained in A and B; in that
case we usually assume subalgebras also share the unit, which this
example doesn't. But I wanted to give you an example which shows
semi-simplicity is insufficient. Take A = M_3(F) and B = M_2(F) + M_1(F);
5 doesn't divide 9.
So the most general result you are likely to get still requires
A and B to be simple, I suppose.
Of course I am assuming you mean "associative algebra".
dave