Version 6.0.0

June 2012

New Examples

Evolver 6.0 includes a completely revised set of example files, designed and written by leading MBA professor and author Dr. Chris Albright of Indiana University.  Each example includes plain-language descriptions, and may be edited to suit your situation.

New Video Tutorials

New video tutorials have also been added by Dr. Albright.  A new, interactive Quick Start tutorial shows new users how to build and understand a simple optimization model in less than 30 minutes.  In addition, other video tutorial resources have been developed to help experienced users get the most out of their software.

OptQuest Optimization Engine Added

The OptQuest optimization engine has been added to Evolver, providing a powerful alternative to the existing Genetic Algorithm available in previous versions.  Evolver is able to examine your model and automatically choose which of the engines is most appropriate to optimize it.

Linear Programming Added

If a problem is linear, Evolver will detect this and will use linear programming methods to quickly find a solution.  This makes the optimization very fast, and also ensures that the best solution found is the best possible solution.

"Discrete" Adjustable Cells

In previous versions of Evolver adjustable cells could be defined as taking "Integer" or "Any" values; now there is also an option to define adjustable cell values as "Discrete", and to specify the "Step".  For example, suppose we want to only consider multiples of 10 between 100 and 1000; we can define a Discrete variable with these values as the minimum and the maximum, and 10 as the Step size.  The use of "discrete" adjustable cells reduces dramatically the number of possible solutions compared to adjustable cells for which "Any" values are allowed; this will often result in faster optimizations.  "Discrete" variables are supported with OptQuest, but as of version 6.0.0 they are not supported with the Genetic Algorithm.  With the Genetic Algorithm variables defined as discrete will be treated as non-discrete.

New License Manager

A new License Manager has been created to help you view your license information, activate software you have purchased, help you move licenses from one machine to another, and other related tasks.

Other Changes of Note

Handling of Linear Constraints

If a constraint is linear, Evolver will detect this.  As long as the OptQuest engine is used, generated solutions will almost always meet the constraints.  This makes optimizations faster, since time is not spent recalculating the workbook for solutions that do not meet linear constraints.

Handling of Non-Linear Constraints

With the OptQuest engine non-linear constraints are also handled efficiently, including situations in which the original values of adjustable cells are invalid (do not meet the specified constraints).  The Genetic Algorithm generally requires the original cell values to meet the constraints.  The "Constraint Solver" tool was added in version 5: it finds a valid solution, if the original solution is invalid, providing a starting point for an optimization.  OptQuest optimizations do not require the use of the Constraint Solver.  If the original solution is invalid, OptQuest may start the optimization by generating a sequence of invalid solutions.  However, during this stage it collects information about how much each solution falls short of meeting the constraints, with the objective of finding valid solutions.

Support for Excel-Defined Names

If a range of cells is named using Excel's interface, Evolver dialogs and reports will show this name instead of the range address (this applies to workbook-level names, not to worksheet-level names).

Process of Defining Constraints Simplified

With the "Simple" constraint "Entry Style", constraint limits no longer need to be fixed values.  In version 5, when defining a constraint with Simple Entry Style, constraint limits had to be fixed values.  For instance, if A1:C1 were adjustable cells, one could not specify A1>B1>C1 using the Simple Entry Style.  The Entry Style had to be changed to "Formula" before the constraint could be entered (and it had to be entered as two separate formulas).  This limitation no longer exists.

Numeric Information about Constraints Provided in the Log

In version 5 the log specified only whether a constraint was met on each trial; in version 6, additional numeric information is provided.  For instance, if we have a constraint saying B3>1000, the log provides the value of cell B3 on each trial.  Note it is not always possible to report the result of the evaluation of a constraint as single number.  For example, consider a constraint saying A1<B1, where both cells are adjusted during optimization; in such cases the program still reports only whether the constraint was met.

Reporting Complex Constraints as Individual Constraints

A constraint can be specified in terms of ranges of cells; for example, we may have a constraint saying that A1:A3 < B1:B3.  In version 5 a constraint like that would be reported in one column in the log.  In version 6 it is reported in 3 separate columns (A1<B1, A2<B2, A3<B3), providing more detailed information as to which parts of the original constraint are met, and which ones are not.

Single Set of Genetic Algorithm Parameters

This change only affects users with models created with versions of software released before version 6, and only if multiple groups of adjustable cells were defined in the model.  In version 5 it was possible to specify a different mutation rate, crossover rate and genetic operators for each group of adjustable cells.  In version 6 there is one mutation rate, crossover rate and one selection of genetic operators for all adjustable cells.  Models created with version 5 or earlier versions are automatically converted to the new format by version 6.  For example, if different mutation rates are specified for different groups of adjustable cells, after opening with version 6 the same mutation rate will be used for all adjustable cells (and it will be the mutation rate specified for the first group of adjustable cells).  If the workbook is saved and subsequently used with version 5 of the software (or older), then the mutation rate will be set to the same value for all the groups of adjustable cells.

Precision Parameter Added to Constraint Settings Dialog

The definition of a hard constraint now includes "Precision".  This refers to violations of constraints that are so small that Evolver will disregard them, and will treat a solution as valid despite these small violations.  This small imprecision in the handling of constraints relates to the fact that computers can only handle mathematical operations with finite precision.  In vast majority of optimizations this setting can be left as "Automatic."  With the automatic option, the optimization summary report includes the specific precision value that was selected automatically.

Licensing Changes

Ability to Select From Multiple Software Licenses

In some cases, you may have more than one license for a Palisade product.  For example, if you are both employed by a company that gave you a professional license, but also have a student license from a university you attend, there is more than one possible license that could be used.  In version 5.x you were given no choice in this matter; the software would choose one of the licenses for you and proceed to run.  In version 6.x, the Activate License Dialog now allows you to look at all your existing licenses, and choose which one to use.

Improved Support for SSD Installations

New licensing capabilities ease installation and licensing when SSD drives are used.