spobooks 5597602.0002.001 in
    10-6 LINEAR EQUATIONS 735 K. We can use these vectors as column vectors to form an n by k matrix D: D = l " l i= ", i= 1,...,k Now the vectors of the linear variety L are all vectors of the form x c+t11 + t22,+-... + tkk where tI,...., tk are arbitrary scalars or, in terms of coordinates, are all (x1..... x) such that x1 = c1 + d11t1 + d12t2 +... + dlktk (10-63) xn ~ = c+ dnitl + dn2t2 + " + d ktk Thus x = c + Dt (10-64) where t is an arbitrary column vector in V. All vectors in (10-64) are written vertically. E(quations (10-63) (or (10-64)) are termed a parametric representation of the linear variety L in terms of parameters t1,..., tk. (When k = 1 and n = 2, we obtain the parametric equations of a line, as in Section 1-15.) When c = 0, L K and the linear variety is a subspace of V". Thus x = Dt is the parametric representation of a k-dimensional subspace K in terms of parameters t1,...., tk. When k = n, t is an arbitrary vector of V? and the equation x= Dt is the same as x= t161 + -... + t, the representation of x in terms of a basis 61...., n" For general k, the equation x = Dt can be regarded as a linear mapping D of \ 1 into Vn whose range is K. Because c,..., k are a basis for K, the representation of x as )t is unique; that is, D is a one-to-one mapping as in No. 5 above, with D of maximum rank, here equal to k). We remark finally that, given a subspace Z of V1 of dimension k, there are systems of n - k linear homogeneous equations in n unknowns whose solutions are the vectors in Z. Furthermore, if L is a k-dimensional linear variety of V, not a subspace of V,, then L is the set of solutions of n - k linear equations: a:x1 + a + -... ainxn = b, i 1,2...., n- k, where the matrix (a;) has rank n - k. These assertions follow from Theorems 1 and 2 in Section 10-4. Results Formulated Without Vectors. Because of their importance for many applications, we restate some of the results for the case m = n in other language: THEOREM 3. Let there be given the n equations in n unknowns (1 1X ti x... + a n n -- x, bl..... (nl, 1 " x +- "' - annn x, bn (10-65) with co esonding homogeneous equations a11x1 + --... + alxn = 0,... anx1 +... + annxn = 0 (10-66)
    Top of page Top of page