AP 
The feasibility of this method of partitioning depends on the effort necessary 
to invert the matrix. AP^!AT, Because this matrix is even for the most 
general case of our problem, a Sequence of fully separated symmetrically 
arranged (2 x 2) square sub-matrices, it is possible to accumulate the normal 
equation system: C Stepwise, as explained in {21 page 22. Thus we obtain: 
2 $ [BT(AP" AT !8]i^- z (BTtar=lary"", ], Nc 
whereby m, the number of AP-IAT submatrices, equals the number of rays 
present in the specific problem. As already mentioned at the beginning of this | 
paragraph, in case only the residuals of the plate measurements are present the | 
Aj matrices are unit matrices and therefore the (AP^! AT]-!- term in (21) 
reduces to P, In such a case the system (31) resembles a system Of normal 
equations associated with observational equations for independent indirect 
measurements. The final normal equation gystem in such a case can be accumu- 
lated stepwise according to formula (32) by considering in each Computational 
Step & single observational equation. 
After the vector of the A corrections of the unknown parameters is 
computed, we obtain with the first group of equations in formula (21) , the 
k-values. 
: 7H AT )-! = 
k= (AP A J'(BA-1) (33) 
and the residuals V and V by: 
(34) 
The V-values are then computed with formulas (16), 
05 
E