M. Minoux. Mathematical Programming. Theory and Algorithms. Wiley, 1986.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....updates [7] or those of Huang [19] or those used in the ABS projection algorithms [1] As proved in [16, 17] although Broyden s methods converge in a finite number of iterations for linear systems, the updates do not always converge to A 1 . For reviews of update methods, see [23] and [22]; on preconditioning techniques, see [8, 13, 14] We shall now explore another possibility. Let us consider the two following sequences of variable preconditioners C n 1 = C n U n D n (7) C # n 1 = U # n C # n D n (8) where (D n ) is an arbitrary sequence of matrices, U n = I AD n and U # n ....

M. Minoux, Mathematical Programming, Theory and Algorithms, Wiley, Chichester, 1986.

....a sequence of matrices (C n ) converging to J . Quasi Newton methods for the solution of systems of nonlinear equations were introduced by Broyden [13] There exists an extensive literature on the subject and it is not our purpose to review it here and we refer the interested reader to [56, 55, 32, 45], for example. The idea behind such procedures is to compute recursively a sequence of matrices H n approximating J or a sequence of matrices C n approximating J . For example, H n 1 can be computed from H n by a rank one modification such as in Broyden s method (y n Gamma H n s n )s = H n ....

....C n (s n Gamma C n y n )y (y n ; y n ) All these acceleration procedures and the corresponding iterative methods remain to be studied both from the theoretical and the practical point of view. For more details, the interested reader is referred to books on unconstrained optimization such as [32, 55] or books on nonlinear equations such as [45, 56, 57, 58] 5 Fixed points and differential equations Usually, various notions from the theory of dynamical systems are used in the analysis of numerical methods for solving initial value problems over long time intervals [58, 64] In this section, ....

M. Minoux, Mathematical Programming. Theory and Algorithms, Wiley, Chichester, 1986.

....updates [7] or those of Huang [19] or those used in the ABS projection algorithms [1] As proved in [16, 17] although Broyden s methods converge in a finite number of iterations for linear systems, the updates do not always converge to A . For reviews of update methods, see [23] and [22]; on preconditioning techniques, see [8, 13, 14] We shall now explore another possibility. Let us consider the two following sequences of variable preconditioners C n 1 = C n U n D n (7) and C n C n D n (8) where (D n ) is an arbitrary sequence of matrices, U n = I Gamma AD n and ....

M. Minoux, Mathematical Programming, Theory and Algorithms, Wiley, Chichester, 1986.

....in this thesis. This will include a discussion of dynamic programming in Section 2.1 and a brief presentation, in Section 2.2, of the Lagrangian relaxation technique. Section 2.3 describes a method for non smooth optimization. 2. 1 Dynamic Programming Dynamic programming, see for example [Min86] or [Win94] can be applied to many constrained optimization problems with a certain property of being decomposable. The term dynamic programming is derived from the fact that 3 4 CHAPTER 2. MATHEMATICAL PROGRAMMING the method was rst applied to the optimization of dynamic systems, that is ....

....i.e. f 0 (u) 0. After reaching the desired state in the last stage it is easy to back trace the path and hence nd the optimal state at every stage. 2. 2 Lagrangian Relaxation Lagrangian relaxation is a decomposition technique used to solve large scale optimization problems, see for example [Min86] or the classical paper by Marshall Fisher [Fis85] The method applies preferably to mathematical programs where the objective function and at least some constraints are additive separable, i.e. when (2.1) is on the form min x n n 0 P i=1 f i (x i ) s:t: n n 0 P i=1 g i;j (x i ) 0 j ....

[Article contains additional citation context not shown here]

Michel Minoux. Mathematical Programming, Theory and Algorithms. John Wiley and Sons Ltd., 1986.

....mapping problem formulation is a nonlinear integer programming NP problem with Boolean variables. 5.3 Mapping algorithms Several algorithms for mapping problem solution are presented in this section below. 5.3. 1 An exact searching algorithm An exact algorithm is based on a branch bound method [Min86] and takes into account the peculiarities of the mapping problem by building the search tree, choosing the bounding function to be assigned with each leaf of this tree and searching for the optimal vertex on each step that corresponds to the component that has to be mapped. Actually, the exact ....

....the faster and less accurate results we ll get) For example, when the search tree grows in breadth rapidly thereby slowing the algorithm down, it may be justified to set the larger eps in order to make the tree grow in depth . Obviously, if eps = then so called depth first search method [Min86] which chooses a vertex of maximum depth among those vertices not yet branched. If there is more than one, then one could choose that which corresponds to the lowest bounding value. This method aims at exhibiting a (good) solution of the problem as soon as possible. Then value F o of the got ....

[Article contains additional citation context not shown here]

M.Minoux. Mathematical Programming. Theory and Algorithms. John Wiley and Sons, 1986

....Using BFGS Optimization procedure. In this method we chose the same error function as in the BPTT algorithm, equation (15) and we also use backpropagation of the error function through the different blocks to evaluate its gradient. 3 SYSTEM IDENTIFICATION AND NEURAL NETWORKS 13 The BFGS method [17], developed independently in the 70s by Broyden, Fletcher, Goldfarb and Shanno, estimates the Hessian of the error function using the formula: H k 1 = H k 1 fl T k H k fl k ffi T k fl k ffi k ffi T k ffi T k fl k Gamma ffi k fl T k H k H k fl k ffi T k ffi T k fl k ....

Michel Minoux. Mathematical Programming. Theory and Algorithms. John Wiley & Sons Ltd., 1986.

....case with an Armijo type rule for the ff k s. It is rather straightforward to prove boundedness of the sequence and optimality of the accumulation points assuming boundedness of the level sets of f and Lipschitz continuity of its gradient, and this result can be found in several text books (e.g. [15] and [18] Without such assumptions, i.e. assuming just convexity and continuous differentiability of f , together with existence of solutions, convergence of the whole sequence to one solution has been established in [4] 6] and [13] for finite dimensional spaces, and in [21] for Hilbert ....

Minoux, M. Mathematical Programming, Theory and Algorithms. John Wiley, New York (1986).

....constraints (14) and (15) The objective function is the total production cost, i.e. the sum of all units start up costs and production costs. Notice that no start up cost is associated with the Heat water storage. 3 Solution Strategy The solution procedure is based on Lagrangian relaxation, see [Min86]. Relaxing all unit coupling constraints, the short term production planning problem (16) decomposes into Optimal Scheduling of Cogeneration Plants 7 q i;D t f i;D t r i;res HWC HWS DUM HEP CHP ELH Figure 2: The unit configuration. one separate problem for each single unit. This methodology ....

Michel Minoux. Mathematical Programming, Theory and Algorithms. John Wiley and Sons Ltd., 1986.

No context found.

M. Minoux. Mathematical Programming. Theory and Algorithms. Wiley, 1986.

No context found.

M. Minoux. Mathematical Programming. Theory and Algorithms. Wiley, 1986.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC