ebook img

Second-Order Cone Program (SOCP) Detection and - Ampl PDF

46 Pages·2010·0.45 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Second-Order Cone Program (SOCP) Detection and - Ampl

SOCP Detectionand Transforma- tion Second-Order Cone Program (SOCP) Detection Erickson, Fourer and Transformation Algorithms for Introduction Optimization Software Background Detection Transformation Conclusion Jared Erickson [email protected] Robert Fourer [email protected] NorthwesternUniversity INFORMSAnnualMeeting,Austin,TX November 8, 2010 SOCP Detectionand Transforma- tion 1 Introduction Erickson, Fourer Introduction 2 Background Background Detection Transformation 3 Detection Conclusion 4 Transformation 5 Conclusion Second-Order Cone Programs (SOCPs) SOCP Detectionand Transforma- tion Can be written as a quadratic program Erickson, Fourer Not positive semi-definite Introduction Convex Background Efficiently solvable with interior-point methods Detection Transformation SOCP general form: Conclusion minimize fTx subject to (cid:107)A x +b (cid:107)2 ≤ (cTx +d )2 ∀ i i i i i cTx +d ≥ 0 ∀ i i i Introduction SOCP Detectionand Transforma- Previous situation: tion Erickson, SOCPs can be written in numerous equivalent forms Fourer The form a modeler wants to use may not be the form a Introduction solver accepts Background Detection Converting the problem for a particular interior-point Transformation solver is tedious and error-prone Conclusion Ideal situation: Write in modeler’s form in a general modeling language Automatically transform to a standard quadratic formulation Transform as necessary for each SOCP solver Motivating Example SOCP Detectionand Transforma- (cid:113) (cid:113) tion minimize (x +2)2+(y +1)2+ (x +y)2 Erickson, Fourer Ampl model: Introduction Background var x; Detection var y; Transformation minimize objective: Conclusion sqrt((x+2)^2+(y+1)^2)+sqrt((x+y)^2); CPLEX 12.2.0.0: at2372.nl contains a nonlinear objective. KNITRO 6.0.0: Current feasible solution estimate cannot be improved. objective 2.12251253; 30 iterations; 209 function evaluations Motivating Example SOCP Detectionand Transforma- tion Original: Erickson, Fourer (cid:113) (cid:113) Introduction minimize (x +2)2+(y +1)2+ (x +y)2 Background Detection Transformed: Transformation minimize u+v Conclusion (x +2)2+(y +1)2 ≤ u2 (x +y)2 ≤ v2 u,v ≥ 0 Motivating Example SOCP Detectionand Transforma- tion Ampl model: Erickson, Fourer var x; var y; Introduction var u >= 0; var v >= 0; Background minimize obj: u+v; Detection s.t. C1: (x+2)^2+(y+1)^2 <= u^2; Transformation s.t. C2: (x+y)^2 <= v^2; Conclusion CPLEX 12.2.0.0: QP Hessian is not positive semi-definite. KNITRO 6.0.0: Locally optimal solution. objective 2.122027399; 3161 iterations; 3276 function evaluations Motivating Example SOCP Original: Detectionand Transforma- tion (cid:113) (cid:113) Erickson, minimize (x +2)2+(y +1)2+ (x +y)2 Fourer Introduction Transformed: Background minimize u+v Detection r2+s2 ≤ u2 Transformation Conclusion t2 ≤ v2 x +2 = r y +1 = s x +y = t u,v ≥ 0 Motivating Example SOCP Detectionand Ampl model: Transforma- tion var x; var y; Erickson, var u >= 0; var v >= 0; Fourer var r; var s; var t; Introduction minimize obj: u+v; Background s.t. C1: r^2+s^2 <= u^2; Detection s.t. C2: t^2 <= v^2; Transformation s.t. C3: x+2 = r; Conclusion s.t. C4: y+1 = s; s.t. C5: x+y = t; CPLEX 12.2.0.0: primal optimal; objective 2.121320344 5 barrier iterations KNITRO 6.0.0: Locally optimal solution. objective 2.122027305; 3087 iterations; 3088 function evaluations Generally Accepted SOCP Form SOCP Detectionand Transforma- SOCP general form: tion Erickson, minimize fTx Fourer Introduction subject to (cid:107)A x +b (cid:107) ≤ cTx +d ∀i i i i i Background Detection where Transformation x ∈ Rn is the variable Conclusion f ∈ Rn Ai ∈ Rmi,n bi ∈ Rmi c ∈ Rn i d ∈ R i

Description:
Nov 8, 2010 SOCP. Detection and. Transforma- tion. Erickson,. Fourer .. 1) Determine objective or constraint type with detection rules. 2) Apply
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.