also how did they know to create two separate vectors for s and t? Is it that a single variable goes for a single combination, meaning if x1 and x3 can are versions of one another then you can describe them both in terms of S, but x2 is not so it gets it's own variable t, and therefore it's own vector? But if t=0 then the second vector should be all zeros not 0,1,0 no?
Now (x1 x2 x3) looks like (s t -s).
We need to express (s t -s) as a linear combination of linearly independent vectors.
(s t -s) = (s 0 -s) + (0 t 0) = s(1 0 -1) + t(0 1 0).
Here (1 0 -1) and (0 1 0) are linearly independent and from the above expression it it clear that any vector of the eigen space( which is of the form (s t -s)) can be expressed as a linear combination of these two vectors, in other words these two linearly independent vectors span the eigen space. So, (1 0 -1) and (0 1 0) form a basis for the eigen space.
Note: c(1 0 -1), where c is a scalar, is a scalar multiplication of (1 0 -1) which means any vector of the form c(1 0 -1) lies on the same straight line passing through (0 0 0) as (1 0 -1) does and they form a subspace of R3.