Convex Optimization. 4. Convex Optimization Problems. Prof. Ying Cui. Department of Electrical Engineering Shanghai Jiao Tong University

Size: px
Start display at page:

Download "Convex Optimization. 4. Convex Optimization Problems. Prof. Ying Cui. Department of Electrical Engineering Shanghai Jiao Tong University"

Transcription

1 Conve Optimization 4. Conve Optimization Problems Prof. Ying Cui Department of Electrical Engineering Shanghai Jiao Tong University 2017 Autumn Semester SJTU Ying Cui 1 / 58

2 Outline Optimization problems Conve optimization Linear optimization problems Quadratic optimization problems Geometric programg Semidefinite prograg Vector optimization SJTU Ying Cui 2 / 58

3 Optimization problem in standard form f 0 () s.t. f i () 0, i = 1,,m optimization variable: R n objective function: f 0 : R n R h i () = 0, i = 1,,p inequality constraint functions: f i : R n R, i = 1,,m equality constraint functions: h i : R n R, i = 1,,p domain: D = m i=0 domf i p i=0 domh i feasible point : D and satisfies all constraints feasible set or constraint set X: set of all feasible points feasible problem: problem with nonempty feasible set SJTU Ying Cui 3 / 58

4 Implicit constraints the standard form optimization problem has an implicit constraint m p D = domf i domh i i=0 i=1 eplicit constraints: f i () 0, i = 1,,m, h i () = 0,i = 1,,p a problem is unconstrained if it has no eplicit constraints (m = p = 0) eample: f 0 () = k log(b i ai T ) is an unconstrained problem with implicit constraints a T i < b i i=1 SJTU Ying Cui 4 / 58

5 Optimal and locally optimal points optimal value: p = inf{f 0 () f i () 0, i = 1,,m, h i () = 0,i = 1,,p} p = if problem is infeasible p = if problem is unbounded below (globally) optimal point : is feasible and f 0 () = p optimal set X opt : set of optimal points if Xopt is nonempty, the optimal value is achieved; otherwise not achieved (always occurs when problem unbounded below) locally optimal point : R > 0 such that is optimal for f 0 (z) z s.t. f i (z) 0, i = 1,,m h i (z) = 0, i = 1,,p z 2 R SJTU Ying Cui 5 / 58

6 Optimal and locally optimal points eamples (with n = 1, m = p = 0) f 0 () = 1/, domf 0 = R ++ : p = 0, no optimal point f 0 () = log, domf 0 = R ++ : p = f 0 () = log, domf 0 = R ++ : p = 1/e, unique optimal point = 1/e f 0 () = 3 3, p =, locally optimal point = 1 SJTU Ying Cui 6 / 58

7 Feasibility problems The feasibility problem is to detere whether the constraints are consistent, and if so, find a point that satisfies them, i.e., find s.t. f i () 0, i = 1,...,m h i () = 0, i = 1,...,p It can be considered a special case of the general problem with f 0 () = 0, i.e., 0 s.t. f i () 0, i = 1,...,m h i () = 0, i = 1,...,p p = 0 if constraints are feasible; any feasible is optimal p = if constraints are infeasible SJTU Ying Cui 7 / 58

8 Conve optimization problems in standard form s.t. f 0 () f i () 0, i = 1,...,m a T i = b i, i = 1,...,p (or A = b) objective function f 0 and inequality constraint functions f 1,...,f m are conve; equality constraints are affine problem is quasiconve if f0 is quasiconve and f 1,...,f m conve feasible set of a conve optimization problem is conve intersection of domain D = m i=0 domf i with m sub level sets { f i () 0} and p hyperplanes { a T i = b i } (all conve) SJTU Ying Cui 8 / 58

9 Abstract form conve optimization problem eample f 0 () = s.t. f 1 () = 1 /(1+2) 2 0 h 1 () = ( ) 2 = 0 imize a conve function over a conve set f0 is conve, feasible set {( 1, 2 ) 1 = 2 0} is conve not a conve optimization problem in standard form (according to our definition) f1 is not conve, h 1 is not affine equivalent (but not identical) to the conve problem s.t = 0 SJTU Ying Cui 9 / 58

10 Local and global optima any locally optimal point of a conve problem is (globally) optimal proof: Suppose is locally optimal, i.e., is feasible and f 0 () (a) = inf{f 0 (z) z feasible, z 2 R} for some R > 0. Suppose is not globally optimal, i.e., there eists a feasible y such that f 0 (y) < f 0 (). Evidently y 2 > R. Consider z = θy +(1 θ) with θ = Then, we have z 2 = R/2 < R, and by conveity of the feasible set, z is feasible. By conveity of f 0, we have which contradicts (a). f 0 (z) θf 0 (y)+(1 θ)f 0 () < f 0 () R 2 y 2. SJTU Ying Cui 10 / 58

11 Optimality criterion for differentiable f 0 is optimal iff it is feasible and f 0 () T (y ) 0 for all y X X f0() Figure 4.2 Geometric interpretation of the optimality condition (4.21). The feasible set X is shown shaded. Some level curves of f0 are shown as dashed lines. The point is optimal: f0() defines a supporting hyperplane (shown as a solid line) to X at. geometric interpretation: if f 0 () 0, f 0 () defines a supporting hyperplane to feasible set X at SJTU Ying Cui 11 / 58

12 Optimality criterion for differentiable f 0 unconstrained problem: f 0 () is optimal iff domf 0, f 0 () = 0 equality constrained problem: f 0 () s.t. A = b is optimal iff there eists a v such that domf 0, A = b, f 0 ()+Av = 0 imization over nonnegative orthant: f 0 () s.t. 0 is optimal iff domf 0, 0, f 0 () 0, ( f 0 ()) i i = 0 condition ( f0 ()) i i = 0 is called complementarity: the sparsity patterns (i.e., the set of indices corresponding to nonzero components) of the vectors and f 0 () are complementary (i.e., have empty intersection) ( { f0 () i 0 i =0 f 0 () i =0 i >0 SJTU Ying Cui 12 / 58 )

13 Equivalent conve problems two problems are (informally) equivalent if the solution of one is readily obtained from the solution of the other, and vice-versa some common transformations that preserve conveity SJTU Ying Cui 13 / 58

14 Equivalent conve problems eliating equality constraints s.t. f 0 () f i () 0, i = 1,...,m A = b is equivalent to z f 0 (Fz + 0 ) s.t. f i (Fz + 0 ) 0, i = 1,...,m where F and 0 are such that A = b = Fz + 0 for some z in principle, we can restrict our attention to conve optimization problems without equality constraints in many cases, however, it is better to retain equality constraints, for ease of analysis, or not to ruin efficiency of an algorithm that solves it SJTU Ying Cui 14 / 58

15 Equivalent conve problems introducing equality constraints is equivalent to f 0 (A 0 +b 0 ) s.t. f i (A i +b i ) 0, i = 1,...,m,y s.t. f 0 (y 0 ) f i (y i ) 0, i = 1,...,m y i = A i +b i, i = 0,...,m SJTU Ying Cui 15 / 58

16 Equivalent conve problems introducing slack variables for linear inequalities s.t. f 0 () a T i b i, i = 1,...,m is equivalent to,s s.t. f 0 () a T i +s i = b i, i = 1,...,m s i 0, i = 1,...,m s i is called the slack variable associated with a T i b i each inequality constraint is replaced with an equality constraint and a nonnegativity constraint SJTU Ying Cui 16 / 58

17 Equivalent conve problems epigraph problem form s.t. f 0 () f i () 0, i = 1,...,m A = b is equivalent to,t t s.t. f 0 () t 0 f i () 0, i = 1,...,m A = b an optimization problem in the graph space (, t): imize t over the epigraph of f 0, subject to the constraints on. a linear objective is universal for conve optimization can simplify theoretical analysis and algorithm development SJTU Ying Cui 17 / 58

18 Equivalent conve problems imizing over some variables is equivalent to f 0 ( 1, 2 ) 1, 2 s.t. f i ( 1 ) 0, i = 1,...,m 1 f i ( 2 ) 0, i = 1,...,m 2 1 f0 ( 1 ) s.t. f i ( 1 ) 0, i = 1,...,m 1 where f 0 ( 1 ) = inf{f 0 ( 1,z) f i (z) 0, i = 1,...,m 2 } SJTU Ying Cui 18 / 58

19 Quasiconve optimization p = s.t. f 0 () f i () 0, i = 1,...,m A = b where f 0 : R n R is quasiconve, and f 1,...,f m are conve locally optimal solutions and optimality conditions Quasiconve optimization problem can have locally optimal points that are not (globally) optimal (,f()) Figure 4.3 A quasiconve function f on R, with a locally optimal point that is not globally optimal. This eample shows that the simple optimality condition f () = 0, valid for conve functions, does not hold for quasiconve functions. SJTU Ying Cui 19 / 58

20 Quasiconve optimization conve representation of sublevel sets of f 0 if f 0 is quasiconve, there eists a family of functions φ t such that: φ t () is conve in for fied t t-sublevel set of f 0 is 0-sublevel set of φ t,i.e., f 0 () t φ t () 0 for each, φ t () is nondecreasing in t eample (conve over concave function) f 0 () = p() q() with p conve, q concave, and p() 0,q() > 0 on domf 0 can take φ t () = p() tq() : for t 0,φ t conve in p()/q() t if and only if φ t () 0 for each, φ t () is nondecreasing in t SJTU Ying Cui 20 / 58

21 Quasiconve optimization quasiconve optimization via conve feasibility problems conve feasibility problem in for any fied t find φ t () 0, f i () 0, i = 1,...,m A = b if feasible, t p ; if infeasible, t p Bisection method for quasiconve optimization ( log 2 ((u l)/ǫ) iterations) given l p,u p, tolerance ǫ > 0. repeat 1. t := (l +u)/2. 2. Solve the conve feasibility problem (1). 3. if feasible, u := t; else l := t. until u l ǫ SJTU Ying Cui 21 / 58

22 Linear program (LP) c T +d s.t. G h A = b where c R n, d R, G R m n, h R m, A R p n and b R p objective and constraint functions are all affine linear programs are, of course, conve optimization problems omit d, i.e., linear objective function (not affect X opt or X) feasible set is a polyhedron c P Figure 4.4 Geometric interpretation of an LP. The feasible set P, which is a polyhedron, is shaded. The objective c T is linear, so its level curves are hyperplanes orthogonal to c (shown as dashed lines). The point is optimal; it is the point in P as far as possible in the direction c. SJTU Ying Cui 22 / 58

23 Linear program (LP) Two special cases of LP: Standard form LPs c T s.t. A = b 0 only inequalities are component-wise nonnegativity constraints Inequality form LPs no equality constraints c T s.t. A b SJTU Ying Cui 23 / 58

24 Linear program (LP) converting LPs to standard form sometimes useful to transform a general LP to a standard form LP (e.g., in order to use an algorithm for standard form LPs) introduce slack variables s i,i = 1,,m for inequalities:,s c T +d s.t. G +s = h A = b s 0 epress + and, i.e., = +, +, 0: +,,s c T + c T +d s.t. G + G +s = h A + A = b + 0, 0, s 0 SJTU Ying Cui 24 / 58

25 LP-eamples diet problem: choose quantities 1,, n of n foods one unit of food j costs c j, contains amount a ij of nutrient i healthy diet requires nutrient i in quantity at least b i to find cheapest healthy diet: piecewise-linear imization: c T s.t. A b, 0 ma i=1,,m (at i +b i ) equivalent to an LP by first forg the epigraph problem and then epressing the inequality as a set of m separate inequalities,t t s.t. a T i +b i t, i = 1,,m SJTU Ying Cui 25 / 58

26 LP-eamples Chebyshev center of a polyhedron: find the center c of the largest Euclidean ball B = { c +u u 2 r} that lies in a polyhedron P = { a T i b i,i = 1,,m} cheb Figure 8.5 Chebyshev center of a polyhedron C, in the Euclidean norm. The center cheb is the deepest point inside C, in the sense that it is farthest from the eterior, or complement, of C. The center cheb is also the center of the largest Euclidean ball (shown lightly shaded) that lies inside C. a T i b i for all B iff sup{a T i ( c +u) u 2 r} = a T i c +sup{a T i u u 2 r} = a T i c +r a i 2 b i Chebyshev center of polyhedron P can be detered by solving the LP c,r r s.t. a T i c +r a i 2 b i, i = 1,,m SJTU Ying Cui 26 / 58

27 Linear-fractional program imize a ratio of affine functions over a polyhedron: f 0 () = ct +d e T +f s.t. G h A = b (domf 0 () = { e T +f > 0}) f 0 () is quasiconve (quasilinear) so linear-fractional programs are quasiconve problems and can be solved by bisection method Transforg to an LP: let y = and z = 1 e T +f e T +f y,z c T y +dz s.t. Gy hz Ay = bz e T y +fz = 1 z 0 SJTU Ying Cui 27 / 58

28 Linear-fractional program generalized linear-fractional program ci T +d i f 0 () = ma i=1,,r ei T, domf 0 () = { ei T +f i > 0,i = 1,,r} +f i f 0 () is the pointwise maimum of r quasiconve functions, and therefore quasiconve, so this problem is quasiconve and can be solved by bisection method eample Von Neumann growth problem: allocate activity to maimize growth rate of slowest growing sector ma, + ) i=1,,n + i / i s.t. + 0, B + A, + R n : activity levels of n sectors, in current, net period (A) i, (B + ) i : produced, consumed amounts of good i + i / i : growth rate of sector i SJTU Ying Cui 28 / 58

29 Quadratic program (QP) (1/2) T P +q T +r s.t. G h A = b where P S n +, q Rn, r R, G R m n and A R p n objective function is conve quadratic and constraints are affine imize a conve quadratic function over a polyhedron QPs include LPs as a special case, by taking P = 0 f0( ) P Figure 4.5 Geometric illustration of QP. The feasible set P, which is a polyhedron, SJTU is shown shaded. The contour Ying lines Cuiof the objective function, which 29 / 58 is conve quadratic, are shown as dashed curves. The point is optimal.

30 QP-eamples least-squares: unconstrained QP (A R k n ) A b 2 2 = T A T A 2b T A +b T b analytical solution = A b singular value decomposition of A: A = UΣV T pseudo-inverse of A: A = VΣ 1 U T R n k constrained least-squares: add linear constraints, e.g., upper and lower bounds on A b 2 2 s.t. l u SJTU Ying Cui 30 / 58

31 QP-eamples linear program with random cost: imize risk-sensitive cost (a linear combination of epected cost and cost variance), captureing a trade-off between small epected cost and small cost variance c T +γ T Σ = Ec T +γvar(c T ) s.t. G h, A = b c R n is random vector with mean c and covariance E(c c)(c c) T ) = Σ for given, c T R is random variable with mean c T and variance E(c T Ec T ) 2 = T Σ risk aversion parameter γ > 0: control the trade-off between epected cost and cost variance (risk) SJTU Ying Cui 31 / 58

32 Quadratically constrained quadratic program (QCQP) (1/2) T P 0 +q T 0 +r 0 s.t. (1/2) T P i +q T i +r i 0, i = 1,,m A = b where P i S n +,i = 1,,m objective and inequality constraint functions are conve quadratic if P 1,,P m S n ++, feasible region is intersection of m ellipsoids and an affine set QCQPs include LPs as a special case, by taking P i = 0,i = 1,,m SJTU Ying Cui 32 / 58

33 Second-order cone programg (SOCP) f T s.t. A i +b i 2 ci T +d i, i = 1,,m F = g where f,c i R n, A i R n i n, d i R, F R p n and g R p each inequality constraint is a second-order cone (SOC) constraint: (A i +b i,c T i +d i ) second-order cone in R n i+1 more general than QCQP (and of course LP) if ci = 0, SOCP reduces to QCQP, by squaring each inequality constraint if Ai = 0, SOCP reduces to LP SJTU Ying Cui 33 / 58

34 SOCP-eample Robust linear programg: when parameters c,a i,b i c T s.t. ai T b i, i = 1,,m can be uncertain, two common approaches to handling uncertainty (in a i, for simplicity) deteristic model: a T i b i must hold for all a i E i c T s.t. ai T b i,for all a i E i i = 1,,m stochastic model: a i is random variable, and each constraint must hold with probability η (chance constraint) c T s.t. prob(ai T b i ) η i = 1,,m SJTU Ying Cui 34 / 58

35 SOCP-eample deteristic approach via SOCP choose an ellipsoid E i with center ā i R n, semi-aes detered by singular values/vectors of P i R n n : E i = {ā i +P i u u 2 1} a T i b i for all a i E i iff sup u 2 1(ā i +P i u) T = ā T i +sup u 2 1(P i u) T = ā T i + P T i 2 < b i robust LP c T s.t. ai T b i for all a i E i, i = 1,,m is equivalent to the SOCP c T s.t. āi T + Pi T 2 b i, i = 1,,m SJTU Ying Cui 35 / 58

36 SOCP-eample stochastic approach via SOCP Assume a i R n is Gaussian with mean ā i and covariance Σ i, i.e., a i N(ā i,σ i ) ai T is Gaussian r.v. with mean āi T and variance T Σ i : ( ) prob(ai T b i āi T b i ) = Φ Σ 1/2 i 2 where Φ() = (1/ 2π) ep t2 /2 dt is CDF of N(0,1) robust LP c T s.t. prob(a T i b i ) η, i = 1,,m with η 1/2, is equivalent to the SOCP c T s.t. ā T i +Φ 1 (η) Σ 1/2 i 2 b i i = 1,,m SJTU Ying Cui 36 / 58

37 Geometric programg (GP) monomial function: f : R n R with c > 0 and a i R f() = c a 1 1 a 2 2 an n, dom f = R n ++ monomials are closed under multiplication and division posynomial function (sum of monomials): f : R n R f() = K k=1 c k a 1k 1 a 2k 2 a nk n, dom f = R n ++ posynomials are closed under addition, multiplication, and nonnegative scaling a posynomial multiplied by a monomial is a posynomial a posynomial divided by a monomial is a posynomial SJTU Ying Cui 37 / 58

38 Geometric programg (GP) geometric program (GP) f 0 () s.t. f i () 1, i = 1,,m h i () = 1, i = 1,,p with f i posynomial and h i monomial, implying domain D = R n ++ (i.e., implicit constraint 0) etensions of GP: f posinomial and h (h i ) monomial f() h() f()/h() 1 (f/h posinomial) f() a (a > 0) f()/a 1 (f/a posinomial) h 1 () = h 2 () h 1 ()/h 2 () 1 (h 1 /h 2 monomial) maimize a nonzero monomial objective function, by imizing its inverse (which is also a monomial) SJTU Ying Cui 38 / 58

39 Geometric program in conve form GPs are not (in general) but can be transformed to conve problems by changing variables to y i = log i and taking log monomial f() = c a 1 1 a 2 2 an n transforms to logf(e y 1,,e yn ) = a T y +b (b = logc) posynomial f() = K k=1 c k a 1k 1 a 2k 2 a nk n transforms to K logf(e y 1,,e yn ) = log( e at k y+b k ) (b k = logc k ) k=1 geometric program transforms to conve problem log( s.t. log( K ep(a0k T y +b 0k)) k=1 K ep(aik T y +b ik)) 0, i = 1,,m k=1 Gy +d = 0 SJTU Ying Cui 39 / 58

40 GP-eamples design of cantilever beam: N segments with unit lengths and rectangular cross-sections of width w i and height h i, and given vertical force F applied at the right end, causing the beam to deflect (downward) and inducing stress in each segment segment 4 segment 3 segment 2 segment 1 Figure 4.6 Segmented cantilever beam with 4 segments. Each segment has unit length and a rectangular profile. A vertical force F is applied at the right end of the beam. F imize total weight subject to upper & lower bounds on w i,h i upper bound & lower bounds on aspect ratios h i /w i upper bound on stress in each segment upper bound on vertical deflection at the end of the beam variables: w i,h i for i = 1,,N SJTU Ying Cui 40 / 58

41 GP-eamples objective and constraint functions total weight w 1 h 1 + +w N h N is a posynomial aspect ratio h i /w i and inverse aspect ratio w i /h i are monomials maimum stress in segment i given by 6iF/(w i hi 2) is a monomial vertical deflection y i and slope v i of central ais at the right end of segment i are defined recursively as F v i = 12(i 1/2) Ew i hi 3 +v i+1 F y i = 6(i 1/3) Ew i hi 3 +v i+1 +y i+1 for i = N,N 1,...,1 with v N+1 = y N+1 = 0 (E: Young s modulus) v i and y i can be shown to be posynomial functions of w,h by induction SJTU Ying Cui 41 / 58

42 GP-eamples formulation as a GP w,h s.t. w 1 h w N h N wmaw 1 i 1, w w 1 i 1, i = 1,...,N hmah 1 i 1, h h 1 i 1, i = 1,...,N Smaw 1 1 i h i 1, S w i h 1 i 1, i = 1,...,N 6iFσmaw 1 1 i h 2 i 1, i = 1,...,N ymay write w w i w ma and h h i h ma as w /w i 1, w i /w ma 1, h /h i 1, h i /h ma 1 write S h i /w i S ma as S w i /h i 1, h i /(w i S ma ) 1 SJTU Ying Cui 42 / 58

43 GP-eamples Minimizing spectral radius of nonnegative matri Perron-Frobenius eigenvalue: λ pf (A) suppose matri A R n n is (elementwise) positive and irreducible (i.e., matri (I +A) n 1 is elementwise positive) Perron-Frobenius theorem: A has a positive real eigenvalue λ pf (A) equal to its spectral radius, i.e., largest magnitude of its eigenvalues ma i λ i (A) λ pf (A) deteres asymptotic growth (decay) rate of A k (i.e., A k λ k pf ) as k and (1/λ pf)a) k converges as k alternative characterization: λ pf (A) = inf{λ Av λv for some v 0} Av λv can be epressed as n A() ij v j /(λv i ) 1, i = 1,...,n j=1 SJTU Ying Cui 43 / 58

44 GP-eamples imizing spectral radius of matri of posynomials: suppose entries of matri A are posynomial functions of some underlying variable R k imize λ pf (A()) possibly subject to posynomial inequalities on equivalent geometric program: λ,v, s.t. λ n A() ij v j /(λv i ) 1, i = 1,...,n j=1 SJTU Ying Cui 44 / 58

45 Generalized inequality constraints conve problem with generalized inequality constraints f 0 () s.t. f i () Ki 0, i = 1,...,m A = b f 0 : R n R conve; f i : R n R k i K i -conve w.r.t proper cone K i R k i results for ordinary conve optimization problems still hold the feasible set, any sublevel set, and the optimal set are conve any point that is locally optimal for the problem (4.48) is globally optimal. optimality condition for differentiable f0 : is optimal iff it is feasible and f 0 () T (y ) 0 for all y X SJTU Ying Cui 45 / 58

46 Conic form problems (cone programs) simplest conve optimization problems with generalized inequalities with a linear objective and one affine inequality constraint function c T s.t. F +g K 0 A = b generalize LPs (reduce to LPs when K = R m +) conic form problem in standard form c T s.t. A = b K 0 conic form problem in inequality form c T s.t. F +g K 0 SJTU Ying Cui 46 / 58

47 Semidefinite prograg (SDP) c T s.t. 1 F F n F n +G 0 A = b with G,F i S k and A R p n (note that K is S k +) inequality constraint is a linear matri inequality (LMI) G,F i are all diagonal, then the LMI is equivalent to a set of n linear inequalities, and the SDP reduces to an LP SJTU Ying Cui 47 / 58

48 Semidefinite prograg (SDP) common to refer to a problem with linear objective, linear equality and inequality constraints, and several LMI constraints, i.e., c T s.t. F (i)() = 1 F (i) F (i) n F (i) n +G (i) 0, i = 1,,K G h, A = b as an SDP, as it is readily transformed to an SDP by forg a large block diagonal LMI c T s.t. tr(g h,f (1) (),,F (K) ()) 0 A = b SJTU Ying Cui 48 / 58

49 Semidefinite prograg (SDP) standard form SDP with C,A i S n inequality form SDP with B,A i S k tr(cx) X s.t. tr(a i X) = b i, i = 1,,p X 0 c T s.t. 1 A A n A n B A = b SJTU Ying Cui 49 / 58

50 LP and SOCP as SDP LP and equivalent SDP LP: c T s.t. A b SDP : c T s.t. diag(a b) 0 (note different interpretation of generalized inequality ) SOCP and equivalent SDP SOCP : f T s.t. A i +b i 2 ci T +d i, i = 1,...,m SDP : s.t. f T [ ] (c T i +d i )I A i +b i (A i +b i ) T ci T 0, i = 1,...,m +d i SJTU Ying Cui 50 / 58

51 SDP-eamples eigenvalue imization R n λ ma (A()) where A() = A A n A n (with given A i S k ) equivalent SDP: t R n,t R s.t. A() ti constraint follows from λ ma (A) t A ti SJTU Ying Cui 51 / 58

52 SDP-eamples matri norm imization A() 2 = (λ ma (A() T A())) 1/2 R n where A() = A A n A n (with given A i R p q ) equivalent SDP: variables R n,t R constraint follows from t R n,t R [ ] ti A() s.t. A() T 0 ti A 2 t A T A t 2 I, t 0 [ ] ti A A T 0 ti SJTU Ying Cui 52 / 58

53 Vector optimization general vector optimization problem (w.r.t.k) f 0 () s.t. f i () 0, i = 1,...,m h i () = 0, i = 1,...,p with vector objective f 0 : R n R q imized w.r.t. proper cone K R q, and constraint functions f i : R n R, h i : R n R conve vector optimization problem (w.r.t.k) f 0 () s.t. f i () 0, i = 1,...,m A = b with f 0 K-conve, f i conve and h i affine SJTU Ying Cui 53 / 58

54 Optimal and Pareto optimal points set of achievable objective values O = {f 0 () feasible} R q f 0 () is the imum value of O (f 0 () K f 0 (y), y O): is optimal and f0 () is the optimal value most vector optimization problems do not have an optimal point (value), but this does occur in some special cases f 0 () is a imal value of O (if there eists y O such that f 0 (y) K f 0 (), then f 0 (y) = f 0 ()) is Pareto optimal and f0 () is a Pareto optimal value a vector optimization problem can have many Pareto optimal points (values) O O f0( po ) f0( ) Figure 4.7 The set O of achievable values for a vector optimization with objective values in R 2, with cone K = R 2 +, is shown shaded. In this case, the point labeled f0( ) is the optimal value of the problem, and is an optimal point. The objective value f0( ) can be compared to every other achievable value f0(y), and is better than or equal to f0(y). (Here, better than or equal to means is below and to the left of.) The lightly shaded region is f0( )+K, which is the set of all z R 2 corresponding to objective values worse than (or equal to) f0( ). Figure 4.8 The set O of achievable values for a vector optimization problem with objective values in R 2, with cone K = R 2 +, is shown shaded. This problem does not have an optimal point or value, but it does have a set of Pareto optimal points, whose corresponding values are shown as the darkened curve on the lower left boundary of O. The point labeled f0( po ) is a Pareto optimal value, and po is a Pareto optimal point. The lightly shaded region is f0( po ) K, which is the set of all z R 2 corresponding to objective values better than (or equal to) f0( po ). SJTU Ying Cui 54 / 58

55 Multicriterion optimization vector optimization problem with K = R q + f 0 () = (F 1 (),...,F q ()) q different scalar objectives F i ; roughly speaking we want all F i s to be small feasible is optimal if f 0 ( ) f 0 (y) for all feasible y if there eists an optimal point, the objectives are noncompeting feasible po is Pareto optimal if that there eists feasible y such that f 0 (y) f 0 ( po ) implies f 0 ( po ) = f 0 (y) if there are multiple Pareto optimal values, there is a trade-off between the objectives SJTU Ying Cui 55 / 58

56 Multicriterion optimization-eamples regularized least-squares (w.r.t.r 2 +) ( A b 2 2, 2 2) 15 F2() = F1() = A b 2 2 Figure 4.11 Optimal trade-off curve for a regularized least-squares problem. The shaded set is the set of achievable values ( A b 2 2, 2 2). The optimal trade-off curve, shown darker, is the lower left part of the boundary. eample for A R ; heavy line is formed by Pareto optimal points SJTU Ying Cui 56 / 58

57 Multicriterion optimization-eamples risk return trade-off in portfolio optimization (w.r.t.r 2 +) ( p T, T ) s.t. 1 T = 1, 0 R n is investment portfolio; i is fraction invested in asset i p R n is vector of relative asset price changes; modeled as a random variable with mean p and covariance p T = Er is epected return; T = var r is return var. 15% mean return 10% 5% 0% 1 0% 10% 20% (4) (3) (2) allocation 0.5 (1) 0 0% 10% 20% standard deviation of return Figure 4.12 Top. Optimal risk-return trade-off curve for a simple portfolio optimization problem. The lefthand endpoint corresponds to putting all resources in the risk-free asset, and so has zero standard deviation. The righthand endpoint corresponds to putting all resources in asset 1, which has highest mean return. Bottom. Corresponding optimal allocations. SJTU Ying Cui 57 / 58

58 Scalarization general vector optimization problems: find Pareto optimal points by choosing λ K 0 and solving scalar problem λ T f 0 () s.t. f i () 0, i = 1,...,m h i () = 0, i = 1,...,p optimal point for the scalar problem is Pareto optimal for vector optimization problem conve vector optimization problems: can find (almost) all Pareto optimal points by varying λ K 0 O f0(1) λ1 f0(3) f0(2) λ2 Figure 4.9 Scalarization. The set O of achievable values for a vector optimization problem with cone K = R 2 +. Three Pareto optimal values f0(1), f0(2), f0(3) are shown. The first two values can be obtained by scalarization: f0(1) imizes λ T 1 u over all u O and f0(2) imizes λ T 2 u, where λ1, λ2 0. The value f0(3) is Pareto optimal, but cannot be found by scalarization. SJTU Ying Cui 58 / 58

Convex optimization problems. Optimization problem in standard form

Convex optimization problems. Optimization problem in standard form Convex optimization problems optimization problem in standard form convex optimization problems linear optimization quadratic optimization geometric programming quasiconvex optimization generalized inequality

More information

4. Convex optimization problems

4. Convex optimization problems Convex Optimization Boyd & Vandenberghe 4. Convex optimization problems optimization problem in standard form convex optimization problems quasiconvex optimization linear optimization quadratic optimization

More information

Lecture: Convex Optimization Problems

Lecture: Convex Optimization Problems 1/36 Lecture: Convex Optimization Problems http://bicmr.pku.edu.cn/~wenzw/opt-2015-fall.html Acknowledgement: this slides is based on Prof. Lieven Vandenberghe s lecture notes Introduction 2/36 optimization

More information

4. Convex optimization problems

4. Convex optimization problems Convex Optimization Boyd & Vandenberghe 4. Convex optimization problems optimization problem in standard form convex optimization problems quasiconvex optimization linear optimization quadratic optimization

More information

Convex Optimization and Modeling

Convex Optimization and Modeling Convex Optimization and Modeling Convex Optimization Fourth lecture, 05.05.2010 Jun.-Prof. Matthias Hein Reminder from last time Convex functions: first-order condition: f(y) f(x) + f x,y x, second-order

More information

Convex Optimization Problems. Prof. Daniel P. Palomar

Convex Optimization Problems. Prof. Daniel P. Palomar Conve Optimization Problems Prof. Daniel P. Palomar The Hong Kong University of Science and Technology (HKUST) MAFS6010R- Portfolio Optimization with R MSc in Financial Mathematics Fall 2018-19, HKUST,

More information

4. Convex optimization problems (part 1: general)

4. Convex optimization problems (part 1: general) EE/AA 578, Univ of Washington, Fall 2016 4. Convex optimization problems (part 1: general) optimization problem in standard form convex optimization problems quasiconvex optimization 4 1 Optimization problem

More information

Lecture: Examples of LP, SOCP and SDP

Lecture: Examples of LP, SOCP and SDP 1/34 Lecture: Examples of LP, SOCP and SDP Zaiwen Wen Beijing International Center For Mathematical Research Peking University http://bicmr.pku.edu.cn/~wenzw/bigdata2018.html wenzw@pku.edu.cn Acknowledgement:

More information

CSCI : Optimization and Control of Networks. Review on Convex Optimization

CSCI : Optimization and Control of Networks. Review on Convex Optimization CSCI7000-016: Optimization and Control of Networks Review on Convex Optimization 1 Convex set S R n is convex if x,y S, λ,µ 0, λ+µ = 1 λx+µy S geometrically: x,y S line segment through x,y S examples (one

More information

Lecture 10. ( x domf. Remark 1 The domain of the conjugate function is given by

Lecture 10. ( x domf. Remark 1 The domain of the conjugate function is given by 10-1 Multi-User Information Theory Jan 17, 2012 Lecture 10 Lecturer:Dr. Haim Permuter Scribe: Wasim Huleihel I. CONJUGATE FUNCTION In the previous lectures, we discussed about conve set, conve functions

More information

A Brief Review on Convex Optimization

A Brief Review on Convex Optimization A Brief Review on Convex Optimization 1 Convex set S R n is convex if x,y S, λ,µ 0, λ+µ = 1 λx+µy S geometrically: x,y S line segment through x,y S examples (one convex, two nonconvex sets): A Brief Review

More information

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics 1. Introduction Convex Optimization Boyd & Vandenberghe mathematical optimization least-squares and linear programming convex optimization example course goals and topics nonlinear optimization brief history

More information

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics 1. Introduction Convex Optimization Boyd & Vandenberghe mathematical optimization least-squares and linear programming convex optimization example course goals and topics nonlinear optimization brief history

More information

Lecture 7: Weak Duality

Lecture 7: Weak Duality EE 227A: Conve Optimization and Applications February 7, 2012 Lecture 7: Weak Duality Lecturer: Laurent El Ghaoui 7.1 Lagrange Dual problem 7.1.1 Primal problem In this section, we consider a possibly

More information

Convexity II: Optimization Basics

Convexity II: Optimization Basics Conveity II: Optimization Basics Lecturer: Ryan Tibshirani Conve Optimization 10-725/36-725 See supplements for reviews of basic multivariate calculus basic linear algebra Last time: conve sets and functions

More information

Geometric problems. Chapter Projection on a set. The distance of a point x 0 R n to a closed set C R n, in the norm, is defined as

Geometric problems. Chapter Projection on a set. The distance of a point x 0 R n to a closed set C R n, in the norm, is defined as Chapter 8 Geometric problems 8.1 Projection on a set The distance of a point x 0 R n to a closed set C R n, in the norm, is defined as dist(x 0,C) = inf{ x 0 x x C}. The infimum here is always achieved.

More information

Lecture 4: September 12

Lecture 4: September 12 10-725/36-725: Conve Optimization Fall 2016 Lecture 4: September 12 Lecturer: Ryan Tibshirani Scribes: Jay Hennig, Yifeng Tao, Sriram Vasudevan Note: LaTeX template courtesy of UC Berkeley EECS dept. Disclaimer:

More information

Convex Optimization and l 1 -minimization

Convex Optimization and l 1 -minimization Convex Optimization and l 1 -minimization Sangwoon Yun Computational Sciences Korea Institute for Advanced Study December 11, 2009 2009 NIMS Thematic Winter School Outline I. Convex Optimization II. l

More information

12. Interior-point methods

12. Interior-point methods 12. Interior-point methods Convex Optimization Boyd & Vandenberghe inequality constrained minimization logarithmic barrier function and central path barrier method feasibility and phase I methods complexity

More information

CS675: Convex and Combinatorial Optimization Fall 2016 Convex Optimization Problems. Instructor: Shaddin Dughmi

CS675: Convex and Combinatorial Optimization Fall 2016 Convex Optimization Problems. Instructor: Shaddin Dughmi CS675: Convex and Combinatorial Optimization Fall 2016 Convex Optimization Problems Instructor: Shaddin Dughmi Outline 1 Convex Optimization Basics 2 Common Classes 3 Interlude: Positive Semi-Definite

More information

Tutorial on Convex Optimization for Engineers Part II

Tutorial on Convex Optimization for Engineers Part II Tutorial on Convex Optimization for Engineers Part II M.Sc. Jens Steinwandt Communications Research Laboratory Ilmenau University of Technology PO Box 100565 D-98684 Ilmenau, Germany jens.steinwandt@tu-ilmenau.de

More information

ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications

ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications Professor M. Chiang Electrical Engineering Department, Princeton University March

More information

Operations Research Letters

Operations Research Letters Operations Research Letters 37 (2009) 1 6 Contents lists available at ScienceDirect Operations Research Letters journal homepage: www.elsevier.com/locate/orl Duality in robust optimization: Primal worst

More information

Conve functions 2{2 Conve functions f : R n! R is conve if dom f is conve and y 2 dom f 2 [0 1] + f ( + (1 ; )y) f () + (1 ; )f (y) (1) f is concave i

Conve functions 2{2 Conve functions f : R n! R is conve if dom f is conve and y 2 dom f 2 [0 1] + f ( + (1 ; )y) f () + (1 ; )f (y) (1) f is concave i ELEC-E5420 Conve functions 2{1 Lecture 2 Conve functions conve functions, epigraph simple eamples, elementary properties more eamples, more properties Jensen's inequality quasiconve, quasiconcave functions

More information

Convex Optimization. (EE227A: UC Berkeley) Lecture 6. Suvrit Sra. (Conic optimization) 07 Feb, 2013

Convex Optimization. (EE227A: UC Berkeley) Lecture 6. Suvrit Sra. (Conic optimization) 07 Feb, 2013 Convex Optimization (EE227A: UC Berkeley) Lecture 6 (Conic optimization) 07 Feb, 2013 Suvrit Sra Organizational Info Quiz coming up on 19th Feb. Project teams by 19th Feb Good if you can mix your research

More information

8. Geometric problems

8. Geometric problems 8. Geometric problems Convex Optimization Boyd & Vandenberghe extremal volume ellipsoids centering classification placement and facility location 8 Minimum volume ellipsoid around a set Löwner-John ellipsoid

More information

Convex Optimization M2

Convex Optimization M2 Convex Optimization M2 Lecture 3 A. d Aspremont. Convex Optimization M2. 1/49 Duality A. d Aspremont. Convex Optimization M2. 2/49 DMs DM par email: dm.daspremont@gmail.com A. d Aspremont. Convex Optimization

More information

Optimisation convexe: performance, complexité et applications.

Optimisation convexe: performance, complexité et applications. Optimisation convexe: performance, complexité et applications. Introduction, convexité, dualité. A. d Aspremont. M2 OJME: Optimisation convexe. 1/128 Today Convex optimization: introduction Course organization

More information

Exercises. Exercises. Basic terminology and optimality conditions. 4.2 Consider the optimization problem

Exercises. Exercises. Basic terminology and optimality conditions. 4.2 Consider the optimization problem Exercises Basic terminology and optimality conditions 4.1 Consider the optimization problem f 0(x 1, x 2) 2x 1 + x 2 1 x 1 + 3x 2 1 x 1 0, x 2 0. Make a sketch of the feasible set. For each of the following

More information

Midterm 1 Solutions. 1. (2 points) Show that the Frobenius norm of a matrix A depends only on its singular values. Precisely, show that

Midterm 1 Solutions. 1. (2 points) Show that the Frobenius norm of a matrix A depends only on its singular values. Precisely, show that EE127A L. El Ghaoui YOUR NAME HERE: SOLUTIONS YOUR SID HERE: 42 3/27/9 Midterm 1 Solutions The eam is open notes, but access to the Internet is not allowed. The maimum grade is 2. When asked to prove something,

More information

LECTURE 7. Least Squares and Variants. Optimization Models EE 127 / EE 227AT. Outline. Least Squares. Notes. Notes. Notes. Notes.

LECTURE 7. Least Squares and Variants. Optimization Models EE 127 / EE 227AT. Outline. Least Squares. Notes. Notes. Notes. Notes. Optimization Models EE 127 / EE 227AT Laurent El Ghaoui EECS department UC Berkeley Spring 2015 Sp 15 1 / 23 LECTURE 7 Least Squares and Variants If others would but reflect on mathematical truths as deeply

More information

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 17

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 17 EE/ACM 150 - Applications of Convex Optimization in Signal Processing and Communications Lecture 17 Andre Tkacenko Signal Processing Research Group Jet Propulsion Laboratory May 29, 2012 Andre Tkacenko

More information

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics 1. Introduction mathematical optimization least-squares and linear programming convex optimization example course goals and topics nonlinear optimization brief history of convex optimization 1 1 Mathematical

More information

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics 1. Introduction Convex Optimization Boyd & Vandenberghe mathematical optimization least-squares and linear programming convex optimization example course goals and topics nonlinear optimization brief history

More information

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics

1. Introduction. mathematical optimization. least-squares and linear programming. convex optimization. example. course goals and topics 1. Introduction Convex Optimization Boyd & Vandenberghe mathematical optimization least-squares and linear programming convex optimization example course goals and topics nonlinear optimization brief history

More information

Introduction to Alternating Direction Method of Multipliers

Introduction to Alternating Direction Method of Multipliers Introduction to Alternating Direction Method of Multipliers Yale Chang Machine Learning Group Meeting September 29, 2016 Yale Chang (Machine Learning Group Meeting) Introduction to Alternating Direction

More information

12. Interior-point methods

12. Interior-point methods 12. Interior-point methods Convex Optimization Boyd & Vandenberghe inequality constrained minimization logarithmic barrier function and central path barrier method feasibility and phase I methods complexity

More information

Semidefinite and Second Order Cone Programming Seminar Fall 2012 Project: Robust Optimization and its Application of Robust Portfolio Optimization

Semidefinite and Second Order Cone Programming Seminar Fall 2012 Project: Robust Optimization and its Application of Robust Portfolio Optimization Semidefinite and Second Order Cone Programming Seminar Fall 2012 Project: Robust Optimization and its Application of Robust Portfolio Optimization Instructor: Farid Alizadeh Author: Ai Kagawa 12/12/2012

More information

Sparse Optimization Lecture: Basic Sparse Optimization Models

Sparse Optimization Lecture: Basic Sparse Optimization Models Sparse Optimization Lecture: Basic Sparse Optimization Models Instructor: Wotao Yin July 2013 online discussions on piazza.com Those who complete this lecture will know basic l 1, l 2,1, and nuclear-norm

More information

Lecture 4: January 26

Lecture 4: January 26 10-725/36-725: Conve Optimization Spring 2015 Lecturer: Javier Pena Lecture 4: January 26 Scribes: Vipul Singh, Shinjini Kundu, Chia-Yin Tsai Note: LaTeX template courtesy of UC Berkeley EECS dept. Disclaimer:

More information

8. Geometric problems

8. Geometric problems 8. Geometric problems Convex Optimization Boyd & Vandenberghe extremal volume ellipsoids centering classification placement and facility location 8 1 Minimum volume ellipsoid around a set Löwner-John ellipsoid

More information

I.3. LMI DUALITY. Didier HENRION EECI Graduate School on Control Supélec - Spring 2010

I.3. LMI DUALITY. Didier HENRION EECI Graduate School on Control Supélec - Spring 2010 I.3. LMI DUALITY Didier HENRION henrion@laas.fr EECI Graduate School on Control Supélec - Spring 2010 Primal and dual For primal problem p = inf x g 0 (x) s.t. g i (x) 0 define Lagrangian L(x, z) = g 0

More information

Robust-to-Dynamics Linear Programming

Robust-to-Dynamics Linear Programming Robust-to-Dynamics Linear Programg Amir Ali Ahmad and Oktay Günlük Abstract We consider a class of robust optimization problems that we call robust-to-dynamics optimization (RDO) The input to an RDO problem

More information

Semidefinite Programming Basics and Applications

Semidefinite Programming Basics and Applications Semidefinite Programming Basics and Applications Ray Pörn, principal lecturer Åbo Akademi University Novia University of Applied Sciences Content What is semidefinite programming (SDP)? How to represent

More information

LMI MODELLING 4. CONVEX LMI MODELLING. Didier HENRION. LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ. Universidad de Valladolid, SP March 2009

LMI MODELLING 4. CONVEX LMI MODELLING. Didier HENRION. LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ. Universidad de Valladolid, SP March 2009 LMI MODELLING 4. CONVEX LMI MODELLING Didier HENRION LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ Universidad de Valladolid, SP March 2009 Minors A minor of a matrix F is the determinant of a submatrix

More information

Chapter 2: Linear Programming Basics. (Bertsimas & Tsitsiklis, Chapter 1)

Chapter 2: Linear Programming Basics. (Bertsimas & Tsitsiklis, Chapter 1) Chapter 2: Linear Programming Basics (Bertsimas & Tsitsiklis, Chapter 1) 33 Example of a Linear Program Remarks. minimize 2x 1 x 2 + 4x 3 subject to x 1 + x 2 + x 4 2 3x 2 x 3 = 5 x 3 + x 4 3 x 1 0 x 3

More information

9. Geometric problems

9. Geometric problems 9. Geometric problems EE/AA 578, Univ of Washington, Fall 2016 projection on a set extremal volume ellipsoids centering classification 9 1 Projection on convex set projection of point x on set C defined

More information

EE5138R: Problem Set 5 Assigned: 16/02/15 Due: 06/03/15

EE5138R: Problem Set 5 Assigned: 16/02/15 Due: 06/03/15 EE538R: Problem Set 5 Assigned: 6/0/5 Due: 06/03/5. BV Problem 4. The feasible set is the convex hull of (0, ), (0, ), (/5, /5), (, 0) and (, 0). (a) x = (/5, /5) (b) Unbounded below (c) X opt = {(0, x

More information

Applications of Robust Optimization in Signal Processing: Beamforming and Power Control Fall 2012

Applications of Robust Optimization in Signal Processing: Beamforming and Power Control Fall 2012 Applications of Robust Optimization in Signal Processing: Beamforg and Power Control Fall 2012 Instructor: Farid Alizadeh Scribe: Shunqiao Sun 12/09/2012 1 Overview In this presentation, we study the applications

More information

Convex Optimization in Communications and Signal Processing

Convex Optimization in Communications and Signal Processing Convex Optimization in Communications and Signal Processing Prof. Dr.-Ing. Wolfgang Gerstacker 1 University of Erlangen-Nürnberg Institute for Digital Communications National Technical University of Ukraine,

More information

1 Robust optimization

1 Robust optimization ORF 523 Lecture 16 Princeton University Instructor: A.A. Ahmadi Scribe: G. Hall Any typos should be emailed to a a a@princeton.edu. In this lecture, we give a brief introduction to robust optimization

More information

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 18

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 18 EE/ACM 150 - Applications of Convex Optimization in Signal Processing and Communications Lecture 18 Andre Tkacenko Signal Processing Research Group Jet Propulsion Laboratory May 31, 2012 Andre Tkacenko

More information

Chapter 2: Preliminaries and elements of convex analysis

Chapter 2: Preliminaries and elements of convex analysis Chapter 2: Preliminaries and elements of convex analysis Edoardo Amaldi DEIB Politecnico di Milano edoardo.amaldi@polimi.it Website: http://home.deib.polimi.it/amaldi/opt-14-15.shtml Academic year 2014-15

More information

Duality Uses and Correspondences. Ryan Tibshirani Convex Optimization

Duality Uses and Correspondences. Ryan Tibshirani Convex Optimization Duality Uses and Correspondences Ryan Tibshirani Conve Optimization 10-725 Recall that for the problem Last time: KKT conditions subject to f() h i () 0, i = 1,... m l j () = 0, j = 1,... r the KKT conditions

More information

Canonical Problem Forms. Ryan Tibshirani Convex Optimization

Canonical Problem Forms. Ryan Tibshirani Convex Optimization Canonical Problem Forms Ryan Tibshirani Convex Optimization 10-725 Last time: optimization basics Optimization terology (e.g., criterion, constraints, feasible points, solutions) Properties and first-order

More information

EE364 Review Session 4

EE364 Review Session 4 EE364 Review Session 4 EE364 Review Outline: convex optimization examples solving quasiconvex problems by bisection exercise 4.47 1 Convex optimization problems we have seen linear programming (LP) quadratic

More information

Nonlinear Optimization: The art of modeling

Nonlinear Optimization: The art of modeling Nonlinear Optimization: The art of modeling INSEAD, Spring 2006 Jean-Philippe Vert Ecole des Mines de Paris Jean-Philippe.Vert@mines.org Nonlinear optimization c 2003-2006 Jean-Philippe Vert, (Jean-Philippe.Vert@mines.org)

More information

COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS. Didier HENRION henrion

COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS. Didier HENRION   henrion COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS Didier HENRION www.laas.fr/ henrion October 2006 Geometry of LMI sets Given symmetric matrices F i we want to characterize the shape in R n of the LMI set F

More information

Convex Optimization & Lagrange Duality

Convex Optimization & Lagrange Duality Convex Optimization & Lagrange Duality Chee Wei Tan CS 8292 : Advanced Topics in Convex Optimization and its Applications Fall 2010 Outline Convex optimization Optimality condition Lagrange duality KKT

More information

Lecture 4: Linear and quadratic problems

Lecture 4: Linear and quadratic problems Lecture 4: Linear and quadratic problems linear programming examples and applications linear fractional programming quadratic optimization problems (quadratically constrained) quadratic programming second-order

More information

Preliminaries Overview OPF and Extensions. Convex Optimization. Lecture 8 - Applications in Smart Grids. Instructor: Yuanzhang Xiao

Preliminaries Overview OPF and Extensions. Convex Optimization. Lecture 8 - Applications in Smart Grids. Instructor: Yuanzhang Xiao Convex Optimization Lecture 8 - Applications in Smart Grids Instructor: Yuanzhang Xiao University of Hawaii at Manoa Fall 2017 1 / 32 Today s Lecture 1 Generalized Inequalities and Semidefinite Programming

More information

Lecture: Duality.

Lecture: Duality. Lecture: Duality http://bicmr.pku.edu.cn/~wenzw/opt-2016-fall.html Acknowledgement: this slides is based on Prof. Lieven Vandenberghe s lecture notes Introduction 2/35 Lagrange dual problem weak and strong

More information

EE Applications of Convex Optimization in Signal Processing and Communications Dr. Andre Tkacenko, JPL Third Term

EE Applications of Convex Optimization in Signal Processing and Communications Dr. Andre Tkacenko, JPL Third Term EE 150 - Applications of Convex Optimization in Signal Processing and Communications Dr. Andre Tkacenko JPL Third Term 2011-2012 Due on Thursday May 3 in class. Homework Set #4 1. (10 points) (Adapted

More information

A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution:

A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution: 1-5: Least-squares I A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution: f (x) =(Ax b) T (Ax b) =x T A T Ax 2b T Ax + b T b f (x) = 2A T Ax 2A T b = 0 Chih-Jen Lin (National

More information

minimize x x2 2 x 1x 2 x 1 subject to x 1 +2x 2 u 1 x 1 4x 2 u 2, 5x 1 +76x 2 1,

minimize x x2 2 x 1x 2 x 1 subject to x 1 +2x 2 u 1 x 1 4x 2 u 2, 5x 1 +76x 2 1, 4 Duality 4.1 Numerical perturbation analysis example. Consider the quadratic program with variables x 1, x 2, and parameters u 1, u 2. minimize x 2 1 +2x2 2 x 1x 2 x 1 subject to x 1 +2x 2 u 1 x 1 4x

More information

MOSEK Modeling Cookbook Release 3.0. MOSEK ApS

MOSEK Modeling Cookbook Release 3.0. MOSEK ApS MOSEK Modeling Cookbook Release 3.0 MOSEK ApS 13 August 2018 Contents 1 Preface 1 2 Linear optimization 3 2.1 Introduction................................... 3 2.2 Linear modeling.................................

More information

A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution:

A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution: 1-5: Least-squares I A : k n. Usually k > n otherwise easily the minimum is zero. Analytical solution: f (x) =(Ax b) T (Ax b) =x T A T Ax 2b T Ax + b T b f (x) = 2A T Ax 2A T b = 0 Chih-Jen Lin (National

More information

3. Convex functions. basic properties and examples. operations that preserve convexity. the conjugate function. quasiconvex functions

3. Convex functions. basic properties and examples. operations that preserve convexity. the conjugate function. quasiconvex functions 3. Convex functions Convex Optimization Boyd & Vandenberghe basic properties and examples operations that preserve convexity the conjugate function quasiconvex functions log-concave and log-convex functions

More information

Agenda. 1 Cone programming. 2 Convex cones. 3 Generalized inequalities. 4 Linear programming (LP) 5 Second-order cone programming (SOCP)

Agenda. 1 Cone programming. 2 Convex cones. 3 Generalized inequalities. 4 Linear programming (LP) 5 Second-order cone programming (SOCP) Agenda 1 Cone programming 2 Convex cones 3 Generalized inequalities 4 Linear programming (LP) 5 Second-order cone programming (SOCP) 6 Semidefinite programming (SDP) 7 Examples Optimization problem in

More information

5. Duality. Lagrangian

5. Duality. Lagrangian 5. Duality Convex Optimization Boyd & Vandenberghe Lagrange dual problem weak and strong duality geometric interpretation optimality conditions perturbation and sensitivity analysis examples generalized

More information

9. Interpretations, Lifting, SOS and Moments

9. Interpretations, Lifting, SOS and Moments 9-1 Interpretations, Lifting, SOS and Moments P. Parrilo and S. Lall, CDC 2003 2003.12.07.04 9. Interpretations, Lifting, SOS and Moments Polynomial nonnegativity Sum of squares (SOS) decomposition Eample

More information

Subgradients. subgradients. strong and weak subgradient calculus. optimality conditions via subgradients. directional derivatives

Subgradients. subgradients. strong and weak subgradient calculus. optimality conditions via subgradients. directional derivatives Subgradients subgradients strong and weak subgradient calculus optimality conditions via subgradients directional derivatives Prof. S. Boyd, EE364b, Stanford University Basic inequality recall basic inequality

More information

HW1 solutions. 1. α Ef(x) β, where Ef(x) is the expected value of f(x), i.e., Ef(x) = n. i=1 p if(a i ). (The function f : R R is given.

HW1 solutions. 1. α Ef(x) β, where Ef(x) is the expected value of f(x), i.e., Ef(x) = n. i=1 p if(a i ). (The function f : R R is given. HW1 solutions Exercise 1 (Some sets of probability distributions.) Let x be a real-valued random variable with Prob(x = a i ) = p i, i = 1,..., n, where a 1 < a 2 < < a n. Of course p R n lies in the standard

More information

UNCONSTRAINED OPTIMIZATION PAUL SCHRIMPF OCTOBER 24, 2013

UNCONSTRAINED OPTIMIZATION PAUL SCHRIMPF OCTOBER 24, 2013 PAUL SCHRIMPF OCTOBER 24, 213 UNIVERSITY OF BRITISH COLUMBIA ECONOMICS 26 Today s lecture is about unconstrained optimization. If you re following along in the syllabus, you ll notice that we ve skipped

More information

Convex Optimization in Classification Problems

Convex Optimization in Classification Problems New Trends in Optimization and Computational Algorithms December 9 13, 2001 Convex Optimization in Classification Problems Laurent El Ghaoui Department of EECS, UC Berkeley elghaoui@eecs.berkeley.edu 1

More information

3. Convex functions. basic properties and examples. operations that preserve convexity. the conjugate function. quasiconvex functions

3. Convex functions. basic properties and examples. operations that preserve convexity. the conjugate function. quasiconvex functions 3. Convex functions Convex Optimization Boyd & Vandenberghe basic properties and examples operations that preserve convexity the conjugate function quasiconvex functions log-concave and log-convex functions

More information

Extreme Abridgment of Boyd and Vandenberghe s Convex Optimization

Extreme Abridgment of Boyd and Vandenberghe s Convex Optimization Extreme Abridgment of Boyd and Vandenberghe s Convex Optimization Compiled by David Rosenberg Abstract Boyd and Vandenberghe s Convex Optimization book is very well-written and a pleasure to read. The

More information

Convex Optimization. 9. Unconstrained minimization. Prof. Ying Cui. Department of Electrical Engineering Shanghai Jiao Tong University

Convex Optimization. 9. Unconstrained minimization. Prof. Ying Cui. Department of Electrical Engineering Shanghai Jiao Tong University Convex Optimization 9. Unconstrained minimization Prof. Ying Cui Department of Electrical Engineering Shanghai Jiao Tong University 2017 Autumn Semester SJTU Ying Cui 1 / 40 Outline Unconstrained minimization

More information

A Tutorial on Convex Optimization

A Tutorial on Convex Optimization A Tutorial on Conve Optimization Haitham Hindi Palo Alto Research Center (PARC), Palo Alto, California email: hhindi@parc.com Abstract In recent years, conve optimization has become a computational tool

More information

Economics 205 Exercises

Economics 205 Exercises Economics 05 Eercises Prof. Watson, Fall 006 (Includes eaminations through Fall 003) Part 1: Basic Analysis 1. Using ε and δ, write in formal terms the meaning of lim a f() = c, where f : R R.. Write the

More information

Lecture 1: January 12

Lecture 1: January 12 10-725/36-725: Convex Optimization Fall 2015 Lecturer: Ryan Tibshirani Lecture 1: January 12 Scribes: Seo-Jin Bang, Prabhat KC, Josue Orellana 1.1 Review We begin by going through some examples and key

More information

Lecture 14: Optimality Conditions for Conic Problems

Lecture 14: Optimality Conditions for Conic Problems EE 227A: Conve Optimization and Applications March 6, 2012 Lecture 14: Optimality Conditions for Conic Problems Lecturer: Laurent El Ghaoui Reading assignment: 5.5 of BV. 14.1 Optimality for Conic Problems

More information

Shiqian Ma, MAT-258A: Numerical Optimization 1. Chapter 4. Subgradient

Shiqian Ma, MAT-258A: Numerical Optimization 1. Chapter 4. Subgradient Shiqian Ma, MAT-258A: Numerical Optimization 1 Chapter 4 Subgradient Shiqian Ma, MAT-258A: Numerical Optimization 2 4.1. Subgradients definition subgradient calculus duality and optimality conditions Shiqian

More information

CHAPTER 2: CONVEX SETS AND CONCAVE FUNCTIONS. W. Erwin Diewert January 31, 2008.

CHAPTER 2: CONVEX SETS AND CONCAVE FUNCTIONS. W. Erwin Diewert January 31, 2008. 1 ECONOMICS 594: LECTURE NOTES CHAPTER 2: CONVEX SETS AND CONCAVE FUNCTIONS W. Erwin Diewert January 31, 2008. 1. Introduction Many economic problems have the following structure: (i) a linear function

More information

Convex Optimization. Newton s method. ENSAE: Optimisation 1/44

Convex Optimization. Newton s method. ENSAE: Optimisation 1/44 Convex Optimization Newton s method ENSAE: Optimisation 1/44 Unconstrained minimization minimize f(x) f convex, twice continuously differentiable (hence dom f open) we assume optimal value p = inf x f(x)

More information

LMI Methods in Optimal and Robust Control

LMI Methods in Optimal and Robust Control LMI Methods in Optimal and Robust Control Matthew M. Peet Arizona State University Lecture 02: Optimization (Convex and Otherwise) What is Optimization? An Optimization Problem has 3 parts. x F f(x) :

More information

Convex Optimization for Signal Processing and Communications: From Fundamentals to Applications

Convex Optimization for Signal Processing and Communications: From Fundamentals to Applications Convex Optimization for Signal Processing and Communications: From Fundamentals to Applications Chong-Yung Chi Institute of Communications Engineering & Department of Electrical Engineering National Tsing

More information

CSCI 1951-G Optimization Methods in Finance Part 10: Conic Optimization

CSCI 1951-G Optimization Methods in Finance Part 10: Conic Optimization CSCI 1951-G Optimization Methods in Finance Part 10: Conic Optimization April 6, 2018 1 / 34 This material is covered in the textbook, Chapters 9 and 10. Some of the materials are taken from it. Some of

More information

Lecture Note 5: Semidefinite Programming for Stability Analysis

Lecture Note 5: Semidefinite Programming for Stability Analysis ECE7850: Hybrid Systems:Theory and Applications Lecture Note 5: Semidefinite Programming for Stability Analysis Wei Zhang Assistant Professor Department of Electrical and Computer Engineering Ohio State

More information

Linear and non-linear programming

Linear and non-linear programming Linear and non-linear programming Benjamin Recht March 11, 2005 The Gameplan Constrained Optimization Convexity Duality Applications/Taxonomy 1 Constrained Optimization minimize f(x) subject to g j (x)

More information

Introduction and Math Preliminaries

Introduction and Math Preliminaries Introduction and Math Preliminaries Yinyu Ye Department of Management Science and Engineering Stanford University Stanford, CA 94305, U.S.A. http://www.stanford.edu/ yyye Appendices A, B, and C, Chapter

More information

Optimization and Optimal Control in Banach Spaces

Optimization and Optimal Control in Banach Spaces Optimization and Optimal Control in Banach Spaces Bernhard Schmitzer October 19, 2017 1 Convex non-smooth optimization with proximal operators Remark 1.1 (Motivation). Convex optimization: easier to solve,

More information

Convex Functions. Daniel P. Palomar. Hong Kong University of Science and Technology (HKUST)

Convex Functions. Daniel P. Palomar. Hong Kong University of Science and Technology (HKUST) Convex Functions Daniel P. Palomar Hong Kong University of Science and Technology (HKUST) ELEC5470 - Convex Optimization Fall 2017-18, HKUST, Hong Kong Outline of Lecture Definition convex function Examples

More information

Statistical Geometry Processing Winter Semester 2011/2012

Statistical Geometry Processing Winter Semester 2011/2012 Statistical Geometry Processing Winter Semester 2011/2012 Linear Algebra, Function Spaces & Inverse Problems Vector and Function Spaces 3 Vectors vectors are arrows in space classically: 2 or 3 dim. Euclidian

More information

Constrained Optimization and Lagrangian Duality

Constrained Optimization and Lagrangian Duality CIS 520: Machine Learning Oct 02, 2017 Constrained Optimization and Lagrangian Duality Lecturer: Shivani Agarwal Disclaimer: These notes are designed to be a supplement to the lecture. They may or may

More information

Convex Optimization Boyd & Vandenberghe. 5. Duality

Convex Optimization Boyd & Vandenberghe. 5. Duality 5. Duality Convex Optimization Boyd & Vandenberghe Lagrange dual problem weak and strong duality geometric interpretation optimality conditions perturbation and sensitivity analysis examples generalized

More information

Lecture Semidefinite Programming and Graph Partitioning

Lecture Semidefinite Programming and Graph Partitioning Approximation Algorithms and Hardness of Approximation April 16, 013 Lecture 14 Lecturer: Alantha Newman Scribes: Marwa El Halabi 1 Semidefinite Programming and Graph Partitioning In previous lectures,

More information

Likelihood Bounds for Constrained Estimation with Uncertainty

Likelihood Bounds for Constrained Estimation with Uncertainty Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference 5 Seville, Spain, December -5, 5 WeC4. Likelihood Bounds for Constrained Estimation with Uncertainty

More information

Approximation Algorithms

Approximation Algorithms Approximation Algorithms Chapter 26 Semidefinite Programming Zacharias Pitouras 1 Introduction LP place a good lower bound on OPT for NP-hard problems Are there other ways of doing this? Vector programs

More information

Lecture: Introduction to LP, SDP and SOCP

Lecture: Introduction to LP, SDP and SOCP Lecture: Introduction to LP, SDP and SOCP Zaiwen Wen Beijing International Center For Mathematical Research Peking University http://bicmr.pku.edu.cn/~wenzw/bigdata2015.html wenzw@pku.edu.cn Acknowledgement:

More information