Abstract
A bounded linear programming problem with feasible solutions may be cast as a discrete L_1 curve-fitting problem of the same size. This may be usefully exploited in solving dense LP problems: On the average, a recent L_1 algorithm solves the equivalent L_1 curve fit far fewer required problem steps, and taking far less time, in than that which would be by the one-phase Simplex method applied to the original The relative advantage increases with problem size and the comparison is even more favorable against the two-phase Simplex method. Finally, the Klee-Minty problems on which the Simplex method is of exponential complexity seem to be "easy" problems as equivalent L_1 curve fits.