documents/intro_paper/eclipse.tex

6 % in compliance with the License.  You may obtain a copy of the License
70 Thus the platform supports experimentation with different hybrid
121 may vary with the time of day, and anyway the poor salesmen need  some
139 with different hybrid algorithms until an appropriate one is found
191 concerned with {\em constructive} search, and the second with {\em
198 In this case the new choice is typically compared with the previous
202 necessarily consistent with the constraints.  The idea of changing the
232 As an example consider the map colouring program, with four countries
332 We briefly compare \ECL as a modelling language with formal
416 and inequalities, together with integer constraints.
502 Sadly the attempt to tackle combinatorial problems with mainstream
573 {\em task} with certain attributes is useful.
589 [eclipse 3]:   T=task with [id:a,duration:10].
593 [eclipse 4]:  T1=task with [id:a3,start:S3,end:(time with hour:H3)],
594               T2=task with [id:a4,start:S3,end:(time with hour:H4)],
614 Query 3 shows how the user can equate a variable with a structured
669 An explicit finite domain is associated with  each decision variable
753 associates a range (between 0 and 99) with each of the variables.
838 change up to 59 pence (replacing 99 with 59 in the above program).
890 and \verb0X #= Y0 and replace \verb0=0 in the above model with
912 The first is a library for handling symbolic finite domains, with
940 The domains associated with the countries were {\em red, green} and
942 These were declared as finite domains, with the usual syntax:
1016 associates a symbolic finite domain with the variable $X$.
1027 domain associated with a variable.
1064 often best to make the choice with the fewest alternatives first.
1120 *       Found a solution with cost 6
1151 This query is the same as query 3, with an extra constraint imposed
1182 propagation than would be possible with the disequalities.
1186 However if two variables $X$ and $Y$ have the same domain, with $M>1$
1197 variable with value different from $Val$, but the resulting number of
1203 The {\em fd} library supports such a facility with the constraint
1214 extended with an extra {\em 0/1} variable in the same way.
1225 By contrast with the finite domain library, the {\em range} library
1228 The library enables the programmer to associate a range with one or
1268 the ranges associated with the variables as far as possible based on
1343 {\em ria} requires its constraints to be written with a specific
1631 The notion of solver handles encourages experiments with
1675 and inequalities, together with integer constraints.
1681 Indeed experience with practical applications suggests that the
1687 variables increases very fast with their number.\footnote{Using the
1689 case, exponentially with the number of constraints and variables.}
1746 variable are consistent with the constraint.
1788 These behaviours, together with other facilities of the {\em propia}
1822 A rule with a guard is not executed until its guard is entailed, until
1923 handling rule programs which are distributed with \ECL.
1968 Each suspension is associated with a set of variables, and as soon as
1982 ($ X \rightarrow inst$).  When woken the goal will be scheduled with a
2010 domain associated with $X$ is tightened ($X \rightarrow min$).
2012 There are other events which can wake suspended goals associated with
2022 variable, or those associated with a given event on a given variable.
2144 because search algorithms with a breadth-first component quickly grow
2206 control with \ECL, it allows the \ECL programmer to define new search
2237 {\bf Search with the {\em fdplex} Library}
2281 *       Found a solution with cost -11
2307 Now {\em deleteff} selects the variable with the smallest
2314 optimum with new suggested values for the other variables.
2316 This time the variable with the smallest domain is $W$, and this is
2330 The branch and bound search yields better solutions with longer and
2383 incomplete search, constructive search with good heuristics can
2390 The problem may be changed because the user is unsatisfied with the
2415 detecting a variable whose tentative value is in conflict with a
2430 of other tentative values with which it conflicts according to the
2457 We illustrate this repair algorithm (with an example from the IC-Parc
2478 Initially only the constraint $Y \verb0#=0 3$ is inconsistent with the
2508 be instantiated to any value that is consistent with the variables
2529 For problems with constraints, changing the value of a variable will
2536 The program has been compared with a pure simulated annealing approach
2537 which simply associates a cost with violated constraints and otherwise
2543 together with the {\em eplex} linear constraint library.
2556 repair-based search in combination with the different solvers of \ECL.
2623 Based on growing experience with hybrid algorithms, we have been able
2639 Ongoing applications, with partners such as
2647 that are already encapsulated as libraries, but from the ease with
2686 \newblock A new approach to integrating mixed integer programming with