Documente Academic
Documente Profesional
Documente Cultură
) is linearly dependent on the remaining g’s in (1). Sets which are not linearly dependent are called linearly independent. We require the following lemmas: (i) There exists a g linearly dependent on u, v, ..., and if and only if u, v,...,y, % ... are linearly ongst themeclves. Thia follows from the (i) Any m of the g’s are linearly dependent if m>n (Aitken, 28 3). (iii) At least t10 of the X's in (1) are not zero, For if all the k's except k) are zero, (2) reduces to Kg? = 0 (i = 1,....0)-HOMOGENEOUS COORDINATES—POINTS AT INFINITY 43 But we have excluded the possibility g? =... = gM = 0. Hence k(5) = 0, and wo have a contradiction. (iv) Two geometrical objects g’, g” are identical if and only if g’, g” are linearly dependent, For, if k’g’+k’g” = 0, then, by (iii), &', k” # 0, 80 gil? =... = g'/9%, and conversely. Bearing in mind this lemma, we may speak of linear dependence of sete of coordinates in terms of the corresponding geometrical objects. 22. Point-coordinates : Space é Let P(z, y, 2) be any point of @. We shall now label P by four numbers 2, 23, #3, 2, such that ce afm, yeas, t= fy. ~ (1) If, in the definition in 20, @ is the aggregate of points of & and 2%, .., % can take all values, then we see that %y» Zqy %y, Zq are homogeneous coordinates of P in @, provided we exclude the possibility x, = 0. Suppose now wo define a collection & of objects as those having homogeneous coordinates 2, 2, %, 2%, where these range unrestrictedly over all real values (except only the set 0, 0, 0,0). We shall, as is possible from what has just been said, identify those objects of & for which x, # 0 with the points of &. We shall then choose to call all the objects of & the “points” of €. When it is necessary to make the distinction, we shall call points having x, # 0 “ordinary” points of &, and those having x, = 0 “special” points. The latter are not points of @; so we have hitherto no geometrical definitions concerning them and we are consequently now free to impose any we please (provided they are self-consistent). Their usefulness in respect to & will become evident in 27. Reversion to cartesian coordinates. If =x is any ordinary point P, then, kx being the same point for all & and x,, being not zero, we can take k = 1/z, and so obtain for P, from (1), the particular set of coordinates (x, y, z, 1). We can thus make the 2,, 2, x-coordinates equal to the