Finally, we tum to the last question of how to generate a new basis when the
current basic feasible solution is not optimal. A new criterion for finding the location of
a variable within the transportation problem needs to be defined. Because ofthe structure
of the tableau displayed in Figure (2-3), the sum ofthe basic variables in each row and
each column must remain the same at each step. Using a loop that contains the entering
variable to obtain a new basis will allow for the feasibility of the primal problem to be
kept throughout the changes of the basis. A 8-100p will assist in changing the basis.