1 | // (C) Copyright International Business Machines Corporation and Carnegie Mellon University 2004 |
---|
2 | // All Rights Reserved. |
---|
3 | // This code is published under the Common Public License. |
---|
4 | // |
---|
5 | // Authors : |
---|
6 | // Pierre Bonami, Carnegie Mellon University, |
---|
7 | // Carl D. Laird, Carnegie Mellon University, |
---|
8 | // Andreas Waechter, International Business Machines Corporation |
---|
9 | // |
---|
10 | // Date : 12/01/2004 |
---|
11 | |
---|
12 | #ifndef __TMINLP_HPP__ |
---|
13 | #define __TMINLP_HPP__ |
---|
14 | |
---|
15 | #include "IpUtils.hpp" |
---|
16 | #include "IpReferenced.hpp" |
---|
17 | #include "IpException.hpp" |
---|
18 | #include "IpAlgTypes.hpp" |
---|
19 | #include "CoinPackedMatrix.hpp" |
---|
20 | #include "OsiCuts.hpp" |
---|
21 | #include "IpTNLP.hpp" |
---|
22 | |
---|
23 | #include "CoinHelperFunctions.hpp" |
---|
24 | namespace Ipopt |
---|
25 | { |
---|
26 | |
---|
27 | /** Base class for all MINLPs that use a standard triplet matrix form |
---|
28 | * and dense vectors. |
---|
29 | * The class TMINLP2TNLP allows the caller to produce a viable TNLP |
---|
30 | * from the MINLP (by relaxing binary and/or integers, or by |
---|
31 | * fixing them), which can then be solved by Ipopt. |
---|
32 | * |
---|
33 | * This interface presents the problem form: |
---|
34 | * \f[ |
---|
35 | * \begin{array}{rl} |
---|
36 | * &min f(x)\\ |
---|
37 | * |
---|
38 | * \mbox{s.t.}&\\ |
---|
39 | * & g^L <= g(x) <= g^U\\ |
---|
40 | * |
---|
41 | * & x^L <= x <= x^U\\ |
---|
42 | * \end{array} |
---|
43 | * \f] |
---|
44 | * Where each x_i is either a continuous, binary, or integer variable. |
---|
45 | * If x_i is binary, the bounds [xL,xU] are assumed to be [0,1]. |
---|
46 | * In order to specify an equality constraint, set gL_i = gU_i = |
---|
47 | * rhs. The value that indicates "infinity" for the bounds |
---|
48 | * (i.e. the variable or constraint has no lower bound (-infinity) |
---|
49 | * or upper bound (+infinity)) is set through the option |
---|
50 | * nlp_lower_bound_inf and nlp_upper_bound_inf. To indicate that a |
---|
51 | * variable has no upper or lower bound, set the bound to |
---|
52 | * -ipopt_inf or +ipopt_inf respectively |
---|
53 | */ |
---|
54 | class TMINLP : public ReferencedObject |
---|
55 | { |
---|
56 | public: |
---|
57 | /**@name Constructors/Destructors */ |
---|
58 | //@{ |
---|
59 | TMINLP() |
---|
60 | {} |
---|
61 | ; |
---|
62 | |
---|
63 | /** Default destructor */ |
---|
64 | virtual ~TMINLP() |
---|
65 | {} |
---|
66 | ; |
---|
67 | //@} |
---|
68 | |
---|
69 | /** Class to store sos constraints for model */ |
---|
70 | struct SosInfo |
---|
71 | { |
---|
72 | /** Number of SOS constraints.*/ |
---|
73 | int num; |
---|
74 | /** Type of sos. At present Only type '1' SOS are supported by Cbc*/ |
---|
75 | char * types; |
---|
76 | /** priorities of sos constraints.*/ |
---|
77 | int * priorities; |
---|
78 | |
---|
79 | /** \name Sparse storage of the elements of the SOS constraints.*/ |
---|
80 | /** @{ */ |
---|
81 | /** Total number of non zeroes in SOS constraints.*/ |
---|
82 | int numNz; |
---|
83 | /** For 0 <= i < nums, start[i] gives the indice of indices and weights arrays at which the description of constraints i begins..*/ |
---|
84 | int * starts; |
---|
85 | /** indices of elements belonging to the SOS.*/ |
---|
86 | int * indices; |
---|
87 | /** weights of the elements of the SOS.*/ |
---|
88 | double * weights; |
---|
89 | /** @} */ |
---|
90 | /** default constructor. */ |
---|
91 | SosInfo():num(0), types(NULL), priorities(NULL), numNz(0), starts(NULL), |
---|
92 | indices(NULL), weights(NULL) |
---|
93 | {} |
---|
94 | /** Copy constructor.*/ |
---|
95 | SosInfo(const SosInfo & source):num(source.num), types(NULL), priorities(NULL), |
---|
96 | numNz(source.numNz), starts(NULL),indices(NULL), |
---|
97 | weights(NULL) |
---|
98 | { |
---|
99 | |
---|
100 | if(num > 0) { |
---|
101 | assert(source.types!=NULL); |
---|
102 | assert(source.priorities!=NULL); |
---|
103 | assert(source.starts!=NULL); |
---|
104 | assert(source.indices!=NULL); |
---|
105 | assert(source.weights!=NULL); |
---|
106 | types = new char[num]; |
---|
107 | priorities = new int[num]; |
---|
108 | starts = new int[num + 1]; |
---|
109 | indices = new int[numNz]; |
---|
110 | weights = new double[numNz]; |
---|
111 | for(int i = 0 ; i < num ; i++) { |
---|
112 | source.types[i] = types[i]; |
---|
113 | source.priorities[i] = priorities[i]; |
---|
114 | source.starts[i] = starts[i]; |
---|
115 | } |
---|
116 | for(int i = 0 ; i < numNz ; i++) { |
---|
117 | source.indices[i] = indices[i]; |
---|
118 | source.weights[i] = weights[i]; |
---|
119 | } |
---|
120 | } |
---|
121 | else { |
---|
122 | assert(source.types==NULL); |
---|
123 | assert(source.priorities==NULL); |
---|
124 | assert(source.starts==NULL); |
---|
125 | assert(source.indices==NULL); |
---|
126 | assert(source.weights==NULL); |
---|
127 | } |
---|
128 | |
---|
129 | } |
---|
130 | |
---|
131 | /** destructor*/ |
---|
132 | ~SosInfo() |
---|
133 | { |
---|
134 | gutsOfDestructor(); |
---|
135 | } |
---|
136 | |
---|
137 | |
---|
138 | /** Reset information */ |
---|
139 | void gutsOfDestructor() |
---|
140 | { |
---|
141 | num = 0; |
---|
142 | numNz = 0; |
---|
143 | if(types) delete [] types; |
---|
144 | types = NULL; |
---|
145 | if(starts) delete [] starts; |
---|
146 | starts = NULL; |
---|
147 | if(indices) delete [] indices; |
---|
148 | indices = NULL; |
---|
149 | if(priorities) delete [] priorities; |
---|
150 | priorities = NULL; |
---|
151 | if(weights) delete [] weights; |
---|
152 | weights = NULL; |
---|
153 | } |
---|
154 | }; |
---|
155 | |
---|
156 | /** Stores branching priorities information. */ |
---|
157 | struct BranchingInfo |
---|
158 | { |
---|
159 | /**number of variables*/ |
---|
160 | int size; |
---|
161 | /** User set priorities on variables. */ |
---|
162 | int * priorities; |
---|
163 | /** User set preferered branching direction. */ |
---|
164 | int * branchingDirections; |
---|
165 | /** User set up pseudo costs.*/ |
---|
166 | double * upPsCosts; |
---|
167 | /** User set down pseudo costs.*/ |
---|
168 | double * downPsCosts; |
---|
169 | BranchingInfo(): |
---|
170 | size(0), |
---|
171 | priorities(NULL), |
---|
172 | branchingDirections(NULL), |
---|
173 | upPsCosts(NULL), |
---|
174 | downPsCosts(NULL) |
---|
175 | {} |
---|
176 | BranchingInfo(const BranchingInfo &other) |
---|
177 | { |
---|
178 | gutsOfDestructor(); |
---|
179 | size = other.size; |
---|
180 | priorities = CoinCopyOfArray(other.priorities, size); |
---|
181 | branchingDirections = CoinCopyOfArray(other.branchingDirections, size); |
---|
182 | upPsCosts = CoinCopyOfArray(other.upPsCosts, size); |
---|
183 | downPsCosts = CoinCopyOfArray(other.downPsCosts, size); |
---|
184 | } |
---|
185 | void gutsOfDestructor() |
---|
186 | { |
---|
187 | if (priorities != NULL) delete [] priorities; |
---|
188 | priorities = NULL; |
---|
189 | if (branchingDirections != NULL) delete [] branchingDirections; |
---|
190 | branchingDirections = NULL; |
---|
191 | if (upPsCosts != NULL) delete [] upPsCosts; |
---|
192 | upPsCosts = NULL; |
---|
193 | if (downPsCosts != NULL) delete [] downPsCosts; |
---|
194 | downPsCosts = NULL; |
---|
195 | } |
---|
196 | }; |
---|
197 | /** Type of the variables.*/ |
---|
198 | enum VariableType |
---|
199 | { |
---|
200 | CONTINUOUS, |
---|
201 | BINARY, |
---|
202 | INTEGER |
---|
203 | }; |
---|
204 | |
---|
205 | /** Type of the constraints*/ |
---|
206 | enum ConstraintType |
---|
207 | { |
---|
208 | LINEAR/** Constraint contains only linear terms.*/, |
---|
209 | NON_LINEAR/**Constraint contains some non-linear terms.*/ |
---|
210 | }; |
---|
211 | |
---|
212 | /**@name methods to gather information about the MINLP */ |
---|
213 | //@{ |
---|
214 | /** overload this method to return the number of variables |
---|
215 | * and constraints, and the number of non-zeros in the jacobian and |
---|
216 | * the hessian. */ |
---|
217 | virtual bool get_nlp_info(Index& n, Index& m, Index& nnz_jac_g, |
---|
218 | Index& nnz_h_lag, TNLP::IndexStyleEnum& index_style)=0; |
---|
219 | |
---|
220 | /** overload this method to set the variable type. The var_types |
---|
221 | * array will be allocated with length n. */ |
---|
222 | virtual bool get_var_types(Index n, VariableType* var_types)=0; |
---|
223 | |
---|
224 | /** overload this method to set the constraint types (linear or not)*/ |
---|
225 | virtual bool get_constraints_types(Index m, ConstraintType* const_types)=0; |
---|
226 | |
---|
227 | /** overload this method to return the information about the bound |
---|
228 | * on the variables and constraints. The value that indicates |
---|
229 | * that a bound does not exist is specified in the parameters |
---|
230 | * nlp_lower_bound_inf and nlp_upper_bound_inf. By default, |
---|
231 | * nlp_lower_bound_inf is -1e19 and nlp_upper_bound_inf is |
---|
232 | * 1e19. |
---|
233 | * An exception will be thrown if x_l and x_u are not 0,1 for binary variables |
---|
234 | */ |
---|
235 | virtual bool get_bounds_info(Index n, Number* x_l, Number* x_u, |
---|
236 | Index m, Number* g_l, Number* g_u)=0; |
---|
237 | |
---|
238 | /** overload this method to return the starting point. The bools |
---|
239 | * init_x and init_lambda are both inputs and outputs. As inputs, |
---|
240 | * they indicate whether or not the algorithm wants you to |
---|
241 | * initialize x and lambda respectively. If, for some reason, the |
---|
242 | * algorithm wants you to initialize these and you cannot, set |
---|
243 | * the respective bool to false. |
---|
244 | */ |
---|
245 | virtual bool get_starting_point(Index n, bool init_x, Number* x, |
---|
246 | Index m, bool init_lambda, |
---|
247 | Number* lambda)=0; |
---|
248 | |
---|
249 | /** overload this method to return the value of the objective function */ |
---|
250 | virtual bool eval_f(Index n, const Number* x, bool new_x, |
---|
251 | Number& obj_value)=0; |
---|
252 | |
---|
253 | /** overload this method to return the vector of the gradient of |
---|
254 | * the objective w.r.t. x */ |
---|
255 | virtual bool eval_grad_f(Index n, const Number* x, bool new_x, |
---|
256 | Number* grad_f)=0; |
---|
257 | |
---|
258 | /** overload this method to return the vector of constraint values */ |
---|
259 | virtual bool eval_g(Index n, const Number* x, bool new_x, |
---|
260 | Index m, Number* g)=0; |
---|
261 | |
---|
262 | /** overload this method to return the jacobian of the |
---|
263 | * constraints. The vectors iRow and jCol only need to be set |
---|
264 | * once. The first call is used to set the structure only (iRow |
---|
265 | * and jCol will be non-NULL, and values will be NULL) For |
---|
266 | * subsequent calls, iRow and jCol will be NULL. */ |
---|
267 | virtual bool eval_jac_g(Index n, const Number* x, bool new_x, |
---|
268 | Index m, Index nele_jac, Index* iRow, |
---|
269 | Index *jCol, Number* values)=0; |
---|
270 | |
---|
271 | /** overload this method to return the hessian of the |
---|
272 | * lagrangian. The vectors iRow and jCol only need to be set once |
---|
273 | * (during the first call). The first call is used to set the |
---|
274 | * structure only (iRow and jCol will be non-NULL, and values |
---|
275 | * will be NULL) For subsequent calls, iRow and jCol will be |
---|
276 | * NULL. This matrix is symmetric - specify the lower diagonal |
---|
277 | * only */ |
---|
278 | virtual bool eval_h(Index n, const Number* x, bool new_x, |
---|
279 | Number obj_factor, Index m, const Number* lambda, |
---|
280 | bool new_lambda, Index nele_hess, |
---|
281 | Index* iRow, Index* jCol, Number* values)=0; |
---|
282 | //@} |
---|
283 | |
---|
284 | /** @name Solution Methods */ |
---|
285 | //@{ |
---|
286 | /** This method is called when the algorithm is complete so the TNLP can store/write the solution */ |
---|
287 | virtual void finalize_solution(SolverReturn status, |
---|
288 | Index n, const Number* x, const Number* z_L, const Number* z_U, |
---|
289 | Index m, const Number* g, const Number* lambda, |
---|
290 | Number obj_value)=0; |
---|
291 | //@} |
---|
292 | |
---|
293 | virtual const BranchingInfo * branchingInfo() const = 0; |
---|
294 | |
---|
295 | virtual const SosInfo * sosConstraints() const = 0; |
---|
296 | private: |
---|
297 | /**@name Default Compiler Generated Methods |
---|
298 | * (Hidden to avoid implicit creation/calling). |
---|
299 | * These methods are not implemented and |
---|
300 | * we do not want the compiler to implement |
---|
301 | * them for us, so we declare them private |
---|
302 | * and do not define them. This ensures that |
---|
303 | * they will not be implicitly created/called. */ |
---|
304 | //@{ |
---|
305 | /** Default Constructor */ |
---|
306 | //TMINLP(); |
---|
307 | |
---|
308 | /** Copy Constructor */ |
---|
309 | TMINLP(const TMINLP&); |
---|
310 | |
---|
311 | /** Overloaded Equals Operator */ |
---|
312 | void operator=(const TMINLP&); |
---|
313 | //@} |
---|
314 | |
---|
315 | |
---|
316 | }; |
---|
317 | |
---|
318 | } // namespace Ipopt |
---|
319 | |
---|
320 | #endif |
---|