Why is determinant not defined separately when working with both vectors and scalars?

164 Views Asked by At

consider the determinant $$ \begin{vmatrix} i & j & k \\ 1 & 0 & 2 \\ 0 & 2 & 5 \\ \end{vmatrix} $$ that comes when we calculate the cross product $(i+0j+2k)\times(0i+2j+5k)$.But the matrix contains both scalars and vectors,it is not an ordinary matrix $M_{n,n}(\mathbb F)$ where $\mathbb F$ is a field.Becuase here the second and third rows are real numbers but first one are vectors.Then should we define the meaning of this determinant separately?Because if we calculate it,we have,

$i$$ \begin{vmatrix} 0 & 2 \\ 2 & 5 \\ \end{vmatrix} $$-j$$ \begin{vmatrix} 1 & 2 \\ 0 & 5 \\ \end{vmatrix} $$+k$$ \begin{vmatrix} 1 & 0 \\ 0 & 2 \\ \end{vmatrix} $.Here each multiplication stands for scalar multiplication with a vector not multiplication between scalars.So if we do not define the meaning of determinants for such a matrix which is not HOMOGENEOUS in terms of its entries i.e. all elements not coming from a particular field,then how will it make sense.So how should we define determinant in this case?

1

There are 1 best solutions below

8
On BEST ANSWER

Generally, we don't define determinants in this way. This method for computing the cross product something known as a mnemonic: a pattern used for remembering something complex. It's not a coincidence that it works, of course, but at the same time, it's not a real determinant.

If you wanted to define a more general non-homogeneous determinant, then you'll have to look at the reasons why it's well-defined. Between scalars $\Bbb{R}$ and vectors in $\Bbb{R}^3$, we have a multiplication operation $$a \cdot (x_1, x_2, x_3) = (ax_1, ax_2, ax_3).$$ The scalar multiplication operation is inherently asymmetric, so we can also symmetrise the operation by defining $v \cdot a = a \cdot v$ for all $v \in \Bbb{R}^3$ and $a \in \Bbb{R}$, which forces a kind of commutativity.

This scalar multiplication operation acts sort of associatively with scalar multiplication on $\Bbb{R}$, specifically \begin{align*} b \cdot (a \cdot v) &= (ab) \cdot v \\ b \cdot (v \cdot a) &= (b \cdot v) \cdot a \\ v \cdot (ba) &= (v \cdot b) \cdot a . \end{align*} We also have distributivity laws \begin{align*} a \cdot(u + v) &= (a \cdot u) + (a \cdot v) \\ (a + b) \cdot v &= (a \cdot v) + (b \cdot v). \end{align*} All in all, this builds up a partial ring structure on $\Bbb{R}^3 \cup \Bbb{R}$ where not every element of the set can add or multiply to each other, but the associativity, commutativity, and distributivity all hold. Plus we have some partial additive identities (i.e. $0 \in \Bbb{R}$ and $(0, 0, 0) \in \Bbb{R}^3$ each act as identities for the two parts of the set $\Bbb{R} \cup \Bbb{R}^3$), and associated additive inverses.

It's not a ring, because we don't get closure of addition and multiplication. We cannot multiply two vectors, and we cannot add a scalar to a vector. But, provided we avoid performing these operations, the structure is very much ring-like.

Thus, we can define the determinant of a matrix with elements from $\Bbb{R}^3 \cup \Bbb{R}$, so long as we ensure that no vectors product with vectors, or add to scalars. The given determinant is fine, evidently, but these two determinants are not:

$$\begin{vmatrix} i & j & 1 \\ 1 & 2 & 1 \\ -1 & 0 & 3 \end{vmatrix}, \quad \begin{vmatrix} i & 2 & 0 \\ 3 & j & -1 \\ 0 & 2 & k \end{vmatrix}.$$