Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Given (finite dimensional?) vector spaces $V$ and $W$, we can define the transpose of a linear map $f:V \to W$ by the obvious map $W^* \to V^*$. Can we do a similar thing for the determinant? Can we define the determinant of an operator without making reference to a matrix, or bases? I feel we should be able to do this, given that the value of the determinant is independent of the basis chosen.

share|improve this question
1  
Are you aware of the alternating product of vector spaces? en.wikipedia.org/wiki/Exterior_algebra#The_exterior_power –  user27126 Jan 18 '13 at 4:41

2 Answers 2

up vote 3 down vote accepted

The determinant of a linear map $f:V\to W$ between finite dimensional vector spaces over a field $K$ can only be defined if $V=W \; !$
And be very careful: we really have to require that our vector spaces be equal and not merely isomorphic.

The intrinsic definition is then as follows :
If $V$ has dimension $n$ then the top exterior power $\wedge^n V$ has dimension one so that the induced linear mapping $\wedge^n f:\wedge^n V\to \wedge^n V$ is necessarily a homothety $\omega\mapsto d \cdot\omega$ i.e. the multiplication by some scalar $d\in K$ in the base field.
And that scalar is the required determinant: $\text {det} f\stackrel {\text {def}}{=}d$.

This approch (pioneered by Bourbaki around 1950, I think) is undoubtedly very elegant, but it requires a knowledge of multilinear algebra not taught in elementary undergraduate courses.

share|improve this answer
    
What problem does it have if $V \cong W$? Surely we can still take the top exterior power on both sides? –  user27126 Jan 18 '13 at 5:08
    
Dear @Sanchez: yes, you can take the top exterior product of both sides but the resulting linear map is not a homothety. The problem is that a linear map between two different vector spaces of dimension one is not a homothety and it is absolutely impossible to describe it by a single scalar unless you choose a base in each or the vector spaces. This is an elementary but subtle point –  Georges Elencwajg Jan 18 '13 at 5:14
    
Ah, I see your point. Thanks! –  user27126 Jan 18 '13 at 5:16
    
Yes, I see that the determinant can only be defined when $V=W$. I should have been more careful. –  nigelvr Jan 18 '13 at 13:57
    
Dear nigelvr, don't be too harsh on yourself: this is a rather delicate point, perhaps not sufficiently emphasized in books. –  Georges Elencwajg Jan 18 '13 at 14:50

Elaborating on Sanchez's answer, it's useful to consider the action of a linear operator on objects created by wedge products--which represent oriented planes, volumes, and so on, just as a single vector represents an oriented line.

We generally define the action of linear operators across the wedge product as follows:

$$\underline T(a \wedge b) \equiv \underline T(a) \wedge \underline T(b)$$

The operator on a wedge product is the wedge product of the operator on the individual vectors.

Now, consider the highest-dimensional object that can be formed by wedges. In an $N$-dimensional space, this is a wedge product of $N$ linearly independent vectors. This object itself forms a 1d vector space---all objects of this kind are scalar multiples of each other. For this reason, this object is often called the pseudoscalar. We'll call it $i_N$.

Now, what is $\underline T(i_N)$? First, linear operators that can be extended across wedges preserve the grade of their arguments--vectors go to vectors, planes to go planes, and the pseudoscalar can only go to (some multiple of) itself.

In other words,

$$\underline T(i_N) = \alpha i_N$$

We call the number $\alpha$ the determinant, and it describes how a unit volume in the space is dilated or shrunk by the linear operator. There is no need to resort to a matrix representation to do this, nor must we choose a specific basis to know this is so.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.