文本lingo tutorial教学菜鸟资源.pptxVIP

Improving Solver Performance on Tough Optimization Models by Removing Nonlinearities ? LINDO Systems 2013 Keywords: Linearization, Piecewise linear, Log transformation, Scaling, Second order cone, Conic program, Bivariate piecewise linear, Global optimization;;Reducing Solution Time of My Optimization Model;Q:What is the best way to formulate a nonlinear model? A: Don’t. Formulate it as a Linear Program (perhaps larger with integer variables). Linear programs are an order of magnitude easier to solve than general nonlinear programs of comparable size. Let’s pursue this, . . .;Reducing Solution Time of My Optimization Model;? 1) Substitute out fixed variables, 2) Disregard extraneous variables and associated nonlinearities, 3) Exploit common expressions, 4) Linearization of piecewise linear functions such as ABS(x), and MAX( x, y), 5) Aggregation of interchangeable variables, 6) Use the Prayer Algorithm, i.e., speculatively disregard complicating constraints and pray they are satisfied. 7) Variable scaling and change of variable. 8) Transform convex polynomial programs to SOC/Conic. 9) Transform Geometric programs to (almost) linear programs. 10) Transformation to piecewise linear programs. 11) Disjunctive/Scenario formulations for integer programs. 12) Exploit Semi-Definite Program constraints, e.g., to choose a correlation matrix. 13) Global optimization. ?;Linearization: Eliminate Fixed Variables;MODEL: ! Markowitz Value-at-Risk Portfolio Model(PORTVAR); SETS: STOCKS: X, RET; COVMAT(STOCKS, STOCKS): VARIANCE; ENDSETS DATA: STOCKS = ATT GMC USX; !Covariance matrix and expected returns; VARIANCE = ; RET = 1.0890833 1.213667 1.234583 ; STARTW = 1.0; ! How much