Changes between Version 9 and Version 10 of Documentation/Reference/VehicleRoutingProblem

Timestamp:

07/01/14 09:34:10 (10 years ago)

Author:

abeham

Comment:

--

Documentation/Reference/VehicleRoutingProblem

@@ -2 +2 @@
 The vehicle routing problem (VRP) is a class of problems that frequently occurs in the field of transportation logistics. The original formulation of the problem
 has been defined in the late 1950ies and consists of a fleet of vehicles serving a set of customers with a certain demand from a single depot. The implementation in HeuristicLab covers the capacitated problem formulation with time windows (CVRPTW). Additionally, in version 3.4 pickup and delivery formulations and multiple depots are also supported. Find more information about the supported variants in a blog article: http://dev.heuristiclab.com/trac/hl/core/blog/svonolfe/vrp_34.
+
+There are several popular variants of the VRP like
+ - Capacitated VRP
+ - VRP with Time Windows 
+ - VRP with Pickup and Delivery
+ - Distance constrained VRP
+ - Multiple Depot VRP
+ - VRP with Time Dependent Travel Time
+ - Multi-Trip VRP
+ - ... and much more
 
 == Benchmark instances == 
@@ -36 +46 @@
 }}}
 
+= Implementing new VRP Variants in HeuristicLab =
+
+For an overview see our blog post on the [/blog/category/VehicleRoutingProblem new vehicle routing implementation] in HeuristicLab.
+
+Because of the variety of the VRP variants, the HeuristicLab VRP plugin is designed to be extensible and flexible. There are multiple ways of implementing a new VRP variant by extending different components of the VRP-Plugin.
+
+The following figure provides an overview about the most important components and how they interact:
+
+[[Image(1_StaticVrpOverview_Easy.png)]]
+
+== Different ways of extending the VRP ==
+
+Depending on the nature of the VRP variant there are multiple ways of implementing the variant. Because many popular VRP variants are already implemented, extending the right base-variant can save a lot of implementation effort.
+
+Three basic ways of implementing a new VRP variant are shown in the following tutorials:
+
+ - [[Documentation/Howto/Implement a New VRP Evaluator|Howto: Implement a new VRP Evaluator]]
+  * does not require additional model data
+  * does not need to optimize additional decision variables
+  * interprets existing data differently
+ - [[Documentation/Howto/Implement a New VRP Problem Instance|Howto: Implement a new VRP ProblemInstance]]
+  * require additional model data
+  * does not need to optimize additional decision variables
+ - [[Documentation/Howto/Implement a New VRP Encoding|Howto: Implement a new VRP Encoding]]
+  * need to optimize additional decision variables
+
+Depending on the VRP variant you often need multiple extension points. For example, if you implement a new {{{ProblemInstance}}} you probably need to implement a new {{{Evaluator}}} to interpret the additional data.
+
+= miscellaneous =
+
+The following figure contains a more detailed overview on how derived components integrate in the VRP design and how the components are grouped in different packages:
+
+[[Image(2_StaticVrpOverview_Full.png, 100%)]]