US Army Materiel Command AMCP 706-197 — AMCP 706-197 Reliability Prediction Handbook — US Army Engineering Design Guide PDF

Downloaded from https://www.everyspec.com AMCP 706-197 xaX, 6 g, 0 5-X 1 X3 .0 9 2 0 *i/ Xb Allowable values for the variables in a problem may -0 be limited or constrained. The area within four 0 L * " i boundary curves is called the constra~nt set. , 2 3 4 x FIGURE 12.3. C j& When the minimum value of the objective function is F 1inside the constraint set, the constraint does not af- fect the solution. Here the point f(x) = 0 is the de- sired minimum value. X, FIGURE 12-5.

Nonlinear Programming Problem With Objective Function Inside the Constraint Set 4 f(l - O fill"0.67 - /2 As an example of a nonlinear problem in -frx) 4 which local optima occur, consider an objec- 4 tive function with two minima, both of which fall within the constraint set so that there are two local m;.iima. Contours of such a func- tion are like those shown in Fig. 12-6. The chief nonlinearity in a programming 2 - problem often appears in the constraints rath- 13 er than in the objective function. The con- straint set will then have curved boundaries.

A 2.5 +.x, - problem with nonlinear constraints can very easily have local optima, even if the objective function has only one unconstrained mini- mum. This is demonstrated in Fig. 12-7, X 1 where there is a nonlinear objective function Values of the nonlinear objective function, which is with a nonlinear constraint set that gives local to be minimized, are shown as concentric circles, optima at the two points a and b. No point of the constraint set in the immediate vicinity of The constrained minimum is one of these lines,. ihrpityed salrvleo ~) either point yields a smaller value of () FIGURE 124. Nonlinear Programming Problem With Constrained Minimum 4 12-8