Constrained Linear Least Squares Problems 449

- ( ; : ; . )

• ( ' ; )

and

T\7i - i 7_/^Мп7г.т7Л\/б1

/ - I 0_ :1:_0

= iI-VV^-qq^)g = h-q^gq

= h----ejhq

( 7

Let ^-^l гМ- This completes the proof.

7 . Adding a Row to A

Now we add a new constraint

aySß to the constraints of problem (LS). Let

Л - 4Л . b.-h. ,.J^^ ae)

a / ' '^ \ßj' ^ \1

If ax<j^, then the tableau Ш remains feasible. Otherwise we have to compute Ш according to:

( 7 . 1 ) Algorithm. Adding a row to A. If ax^ß, consider the linear programming problem

min a у

Aypb (27)