nts and where k denotes & vector of unknown Lagrange multipliers or correlates, 
  
A 
ion of direct solution would be obtained by the inversion of the System (21). That 
Xpress . 18, however, hardly practical because in addition to the unknown parameter 
Such ^ Corrections » A's, we have for an M-ray problem 2m additional unknown k-values, 
vality & fact which increases the number of unknown parameters beyond the means of 
ena practical computations. 
image Next, a method which is commonly used in partitioning, a system of linear 
arious equations shall be presented which will be useful for our Problem, Assuming a 
as well, system of linear equations: 
rge Ax = 0 
rom the (22) 
oe co- | or consequently 
to the xs A. 
all the | (23) 
tion. 3 Such à system can be partitioned in any arbitrary manner leading to the follow- 
ondary ing sub-matrices and Sub-vectors; 
lly b = 2 
v AaX1 | * Ag Xo * fs (24) 
se the 
pendent According to (25) we may write with the notation of (24) 
intro- “| : 
In X | - A, A A717 Ajo. x2 (25) 
1e unit 
&nd therefore | 
| = A Oc 
into u* Av Qi-AT App x; uel. 
sion (26) 
ain 
- Introducing (26) into the lower portion of the system (24) we obtain: 
a - -| : 
if the Azı Ar 0,-A2 AV Ap xp + Axa = à, (27) 
oh IV). | -I / = -l 
ia (Ao-Aa A, Ajo) Xo = 02-A2, A] 0 
(21) 
21