OpenDino - Benutzerbeiträge [de]

Screenshots

2016-09-15T21:53:02Z

Dirk: /* References */

== The Main Window ==

After starting software, the main '''GUI''' appears (Fig. 1). The figure contains an example for an optimization using 4 '''Module'''s. Each module has a name and an unique ID. The software is a strictly modular. All functionality for automated optimization and machine learning is encoded in the modules. Here:

Module 1 (<code>OptAlgCMA</code>) is an optimization algorithm. It is connected via Module 3 (<code>Protocoller</code>) with Module 2 (<code>ProblemSimple</code>). Module 2 contains the optimization problem, i.e. the function to minimize. Here, it is a simple continuous, quadratic function.
Module 3 (<code>Protocoller</code>) is a protocoling module, i.e. it stores all evaluated solutions of an optimization run. These solutions can be visualized with Module 4 (<code>DataViewer</code>).
[[Image:MainGUI.png|frame|center|Figure 1: Main GUI of OpenDino]]

== Adding, Editing, and Connecting Modules ==

Right-clicking on the dotted area or on a module opens a dialogue as shown in Fig. 2. This dialogue supports:
* adding modules (if clicked on a free space)
* removing modules or changing the options of a module when clicking on a module
* adding/removing connections between modules
* checking and running modules
In Fig. 2, the right-click is made on Module 4, e.g. for setting the options of this module.

[[Image:ModuleRightclick.png|frame|center|Figure 2: Editing Modules]]

== Running an Optimization ==

Right-clicking Module 1 and selecting the option <code>Run</code> starts an optimization. The standard output of the optimization is given in the lower part of the main window in Fig. 3.

[[Image:Optimization.png|frame|center|Figure 3: Running an Optimization]]

== Postprocessing an Optimization ==

Right-clicking Module 4 and selecting the option <code>Run</code> starts an the data viewer. Figure 4 shows the evolution of the objective function ''f'' over the number of evaluations. As the objective function is a simple quadratic function (x^2), the minimal function value is zero, which is approximated by 1e-12.

[[Image:DataViewer.png|frame|center|Figure 4: Postprocessing an Optimization with the DataViewer module. ]]

Documentation/Related

2016-09-15T21:51:57Z

Dirk: /* MATLAB Implementations */

=Links and Related Pages=

== JAVA Implementations ==

* Optimization Algorithms

* Neural Network and Data Mining
** [http://snarli.sourceforge.net/ SNARLI]
** [http://www.cs.waikato.ac.nz/~ml/weka/ WEKA]

== Multi-Language Implementations ==

* Optimization
** [https://www.lri.fr/~hansen/cmaesintro.html Evolution Strategy with Covariance Matrix Adaptation (CMA-ES)]

Documentation/Related

2016-09-15T21:51:30Z

Dirk: /* JAVA Implementations */

=Links and Related Pages=

== JAVA Implementations ==

* Optimization Algorithms

* Neural Network and Data Mining
** [http://snarli.sourceforge.net/ SNARLI]
** [http://www.cs.waikato.ac.nz/~ml/weka/ WEKA]

== MATLAB Implementations ==

* Neural Network and Data Mining
** [http://www.esat.kuleuven.ac.be/sista/lssvmlab/ Least Squares Support Vector Machines (LS-SVM)]

== Multi-Language Implementations ==

* Optimization
** [https://www.lri.fr/~hansen/cmaesintro.html Evolution Strategy with Covariance Matrix Adaptation (CMA-ES)]

Documentation/Related

2016-09-15T21:50:28Z

Dirk: /* JAVA Implementations */

=Links and Related Pages=

== JAVA Implementations ==

* Neural Network and Data Mining
** [http://snarli.sourceforge.net/ SNARLI]
** [http://www.cs.waikato.ac.nz/~ml/weka/ WEKA]
** [http://rapid-i.com/ RapidMiner]

== MATLAB Implementations ==

* Neural Network and Data Mining
** [http://www.esat.kuleuven.ac.be/sista/lssvmlab/ Least Squares Support Vector Machines (LS-SVM)]

== Multi-Language Implementations ==

* Optimization
** [https://www.lri.fr/~hansen/cmaesintro.html Evolution Strategy with Covariance Matrix Adaptation (CMA-ES)]

Tutorials

2016-09-15T21:27:46Z

Dirk: Created page with "= Using OpenDino as a Library in Other Java Code = OpenDino contains a graphical user interface for simple interaction (see e.g. the [Screenshots]). Alternatively, OpenDino can..."

= Using OpenDino as a Library in Other Java Code =

OpenDino contains a graphical user interface for simple interaction (see e.g. the [Screenshots]).

Alternatively, OpenDino can be used as a library in other Java programs. All functionality of the software can be used such as the graphical plotting environment in the Module {{{DataViewer}}}.

== Tutorials ==

The tutorials can be directly executed as they are implemented in a Java main method.

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Tutorial
! URL
|-
| single objective evolutionary optimization with output of the best solution
| https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/tutorial/TutorialSingleObjectiveOptimization.java
|-
| multi-objective evolutionary optimization with visualization and data storage
| https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/tutorial/TutorialMultiObjectiveOptimization.java
|}

Documentation

2016-09-15T21:15:46Z

Dirk: /* Developers' Guide */

== Requirements ==

* [[Documentation/Software/Requirements | Software Requirements]]

== Users' Guide ==

* [[Installation | Installation]]
* [[Screenshots | A First Introduction to OpenDino in the Screenshots]]
* [[Documentation/Notation | Symbols and Notation]]
* [[Documentation/Modules | Module Documentation]]

== Theory ==
* [[Documentation/DoE | Introduction to Design of Experiment]]
* [[Documentation/Optimization | Introduction to Automated Optimization]]
* [[Documentation/Machine_Learning | Introduction to Machine Learning]]

== Developers' Guide ==

* [[Download | Download]]
* [[Documentation/Compiling | Developing and Compiling]]
* Tutorials
** [[Tutorials/UseAsLibrary | Using OpenDino as a Library in other Java Code]]
* Concepts behind OpenDino
** [[Documentation/Modules/TheCode | The 3 Main Parts]]
** [[Documentation/Modules/IModule | Interface <code>IModule</code> - The Obligatory Interface for Modules]]
** [[Documentation/Modules/Interfaces | Interfaces of Modules]]
** [[Documentation/Modules/ClassesModules | Abstract Modules and Classes]]

== Various ==

* [[Documentation/Related | Links and Related Software]]

Documentation/Software/Requirements

2016-09-15T21:12:55Z

Dirk:

* Compiling: JDK 1.7 or higher
* Executing: JRE 1.7 or higher

Documentation

2016-09-15T21:11:40Z

Dirk: /* Developers' Guide */

== Requirements ==

* [[Documentation/Software/Requirements | Software Requirements]]

== Users' Guide ==

* [[Installation | Installation]]
* [[Screenshots | A First Introduction to OpenDino in the Screenshots]]
* [[Documentation/Notation | Symbols and Notation]]
* [[Documentation/Modules | Module Documentation]]

== Theory ==
* [[Documentation/DoE | Introduction to Design of Experiment]]
* [[Documentation/Optimization | Introduction to Automated Optimization]]
* [[Documentation/Machine_Learning | Introduction to Machine Learning]]

== Developers' Guide ==

* [[Download | Download]]
* [[Documentation/Compiling | Developing and Compiling]]
* Concepts behind OpenDino
** [[Documentation/Modules/TheCode | The 3 Main Parts]]
** [[Documentation/Modules/IModule | Interface <code>IModule</code> - The Obligatory Interface for Modules]]
** [[Documentation/Modules/Interfaces | Interfaces of Modules]]
** [[Documentation/Modules/ClassesModules | Abstract Modules and Classes]]

== Various ==

* [[Documentation/Related | Links and Related Software]]

Main Page

2016-02-18T20:28:28Z

Dirk: /* Welcome to OpenDino */

= Welcome to '''OpenDino''' =

'''OpenDino ''' is an open source platform for automated optimization, design of experiment and learning, written in Java.

The source code as well as as an executable can be found on [http://sourceforge.net/projects/opendino/ sourceforge]. The source code is structured in 3 layers: A core providing the basic backbone, Modules containing all algorithms, and a graphical user interface (GUI). This simplifies the integration of new algorithms as they are represented by additional "Modules".

Implemented Modules

* Evolutionary optimization algorithms (CMA-ES, (1+1)-ES, Swarm Algorithms)

* Deterministic optimization algorithm (SIMPLEX)

* Learning (feed-forward Artificial Neural Net)

* Design of Experiments (full factorial design, Latin Hypercube, I-Optimal Design)

* Problems (test functions, interface to external programs (solvers), distributed and parallel execution)

* Other modules (data storage and visualization)

Source Code
* http://sourceforge.net/projects/opendino/

Contact:
* [mailto:help@opendino.org help@opendino.org]

Download

2015-11-18T22:04:26Z

Dirk:

== Download Software ==

The executable jar file is given at

https://sourceforge.net/projects/opendino/files/

== Download Source Code ==

The software is version controlled using a Subversion (SVN) repository. Source code is located at

https://sourceforge.net/p/opendino/code

The URL specifies also the download command.

In the SVN repository, a "trunk" exist and "branches" might be added. "trunk" contains the main development and should always at least compile and be executable.
"branches" are copies of the "trunk" for temporary modifications (see the [http://subversion.apache.org/ subversion homepage]) and should not be downloaded.

Downloading requires a [http://subversion.apache.org/ subversion client] or an appropriate plug-in for your IDE.

Documentation/Modules/ConstraintHandler

2015-10-26T21:00:06Z

Dirk: /* Module Description */

[[Documentation/Modules | << BACK TO MODULE OVERVIEW <<]]

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

'''0. No Constraint Handling'''

This option turns off the constraint handling. Objective values and constraints are not changed.

'''1. Delete Constraints'''

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

'''2. Penalty Method'''

To each objective function, a penalty (larger or equal to zero) is added. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

penalized objective i = fi + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

'''3. Stochastic Ranking'''

Limitations:
# only for single objective problems
# requires population based search methods such as CMA-ES. There are 3 steps

Rough Description of Main Steps of Stochastic Ranking ([[#1|1]])
# For each solution, the penalty is computed as the sum of all violated constraints (i.e. the constraint value is >= 0) <code> p = sumj(max(gj,0)) </code> as in the Penalty Method.
# The solutions of the population are randomly ranked.
# Always two adjacent solutions of the population are compared either based on their objective value or based on their penalty. Typically the comparison according to the penalty should have a higher probability of about 60%. The winner of the comparison obtains the better ranking position.
# The comparision is repeated several times.

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", ''IEEE Transactions on Evolutionary Computation'', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/OptAlgMOPSO

2015-10-25T19:43:10Z

Dirk: /* Source Code */

==Summary==

This optimization module is an implementation of the particle swarm optimization algorithm for single- and multi-objective optimization [[#References | (1)]], however it contains some modifications to the publication. The algorithm reflects the natural movement of flocking birds.

The algorithm is elitist: Always the best particles are kept as guides.

This algorithm is designed for continuous variables and can not handle discrete problems.
Furthermore, the algorithm is implemented for minimizing a single and multiple objective function(s).

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| stochastic - stochastic adaptation of the velocities.
|-
! Design Variables
| Written for continuous variables. No discrete or mixed variables are possible.
|-
! Objectives
| single- and multi-objective for minimization.
|-
! Constraint handling
| no
|-
! Boundary handling
| no
|-
! Initialization
| Requires at least one of the following: initial search region or bounds.
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at this module
| Module requires exactly one connection of type <code>optimization</code>.
|-
! Ending at this module
| -
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! '''Run'''
| starts the optimization.
|-
|}

===Options===
The options are currently described in the pop-up help.

==Module Description==

=== Initialization ===

The initial particles are randomly generated within the <code>initial search region</code> (if existing) or otherwise between the <code>bounds</code>.

=== Optimization ===
The algorithm contains stochastic processes and operates with a set of particles. Parallelization on the basis of the number of particles is implemented.

==Usage==

... todo

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optAlg/OptAlgMoPso.java

==References==

(1) Sanaz Mostaghim. Multi-Objective Evolutionary Algorithms. Data Structures,
Convergence, and Diversity. Paderborn, Germany, November 2004.

Documentation/Modules/ProblemTruss

2015-10-25T19:39:39Z

Dirk: /* Summary */

==Summary==

The module <code>ProblemTruss</code> is a multi-objective optimization problem. It implements a simple construction problem with the goal to optimize a truss structure. The module constains a small Finite Element Analysis code for analysing a truss structure.

[[File:ProblemTruss.png|300px]]

The truss consists of 10 rods which connect the 6 point (labelled from 0 to 5). At point 0 and 2, a force F=45'000N is applied, which deforms the truss and generates stresses in the 10 rods.

The input of this module are the individual cross-sections of the rods (i.e. 10 cross-sections).

The output of this modules are 3 values, which can be set either as objectives or constraints:

# the weight of the 10 rods of the truss
# the maximum stress in a rod
# the maximum deflection of a node of the truss.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic test function
|-
! Design Variables
| 10 continuous variables
|-
! Objectives
| yes, 3 outputs which can be set either as objectives or constraints.
|-
! Constraints
| yes, 3 outputs which can be set either as objectives or constraints.
|-
! Boundaries
| yes, can be set as an option.
|-
! Initial Search Region
| not set.
|-
! Typical X
| not set.
|-
! Initialization
| sets problem properties.
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| -
|-
! Ending at this module
| Connections of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===

The options are currently described as "pop-up help".

==Usage==
-
==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/problems/ProblemTruss.java

==References==

-

Documentation/Modules/ProblemTruss

2015-10-25T19:38:26Z

Dirk: Created page with "==Summary== The module <code>ProblemTruss</code> contains a multi-objective optimization problem. It implements a simple Finite Element Analysis code for analysing a truss struc..."

==Summary==

The module <code>ProblemTruss</code> contains a multi-objective optimization problem. It implements a simple Finite Element Analysis code for analysing a truss structure.

[[File:ProblemTruss.png|300px]]

The truss consists of 10 rods which connect the 6 point (labelled from 0 to 5). At point 0 and 2, a force F=45'000N is applied, which deforms the truss and generates stresses in the 10 rods.

The input of this module are the individual cross-sections of the rods (i.e. 10 cross-sections).

The output of this modules are 3 values, which can be set either as objectives or constraints:

# the weight of the 10 rods of the truss
# the maximum stress in a rod
# the maximum deflection of a node of the truss.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic test function
|-
! Design Variables
| 10 continuous variables
|-
! Objectives
| yes, 3 outputs which can be set either as objectives or constraints.
|-
! Constraints
| yes, 3 outputs which can be set either as objectives or constraints.
|-
! Boundaries
| yes, can be set as an option.
|-
! Initial Search Region
| not set.
|-
! Typical X
| not set.
|-
! Initialization
| sets problem properties.
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| -
|-
! Ending at this module
| Connections of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===

The options are currently described as "pop-up help".

==Usage==
-
==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/problems/ProblemTruss.java

==References==

-

Datei:ProblemTruss.png

2015-10-25T19:28:21Z

Dirk:

Documentation/Modules/ProblemSimple

2015-10-25T18:54:45Z

Dirk: /* Source Code */

==Summary==

The module <code>ProblemSimple</code> contains a very simple optimization problem, which can be used to illustrate the optimization process. It contains a time delay, which allows to reduce the speed of evaluations. This can be used e.g. to show the convergence of an optimization algorithm in a presentation.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic test function.
|-
! Design Variables
| Continuous, discrete, or mixed variables.
|-
! Objectives
| single-objective for minimization.
|-
! Constraints
| none
|-
! Boundaries
| set to fixed value.
|-
! Initial Search Region
| not set.
|-
! Typical X
| not set.
|-
! Initialization
| sets problem properties.
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| -
|-
! Ending at this module
| Connections of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are the number of discrete and continuous variables and an optional time delay for per evaluation.

==Module Description==

=== Computation of the Objective ===

The single objective function ''f'' is a quadratic function. More precisely, ''f'' is computed as the square of all continuous and discrete variables.

''f''('''x''') = sumi(xi2)

This function is also referred to as the ''sphere function''.

==Usage==
-
==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/problems/ProblemSimple.java

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T18:54:05Z

Dirk:

[[Documentation/Modules | << BACK TO MODULE OVERVIEW <<]]

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

'''0. no constraint handling'''

This option turns off the constraint handling. Objective values and constraints are not changed.

'''1. delete constraints'''

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

'''2. Penalty method'''

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

'''3. Penalty method'''

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", ''IEEE Transactions on Evolutionary Computation'', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:53:19Z

Dirk: /* Summary */

[[Documentation/Modules | << UP TO MODULE OVERVIEW <<]]

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

'''0. no constraint handling'''

This option turns off the constraint handling. Objective values and constraints are not changed.

'''1. delete constraints'''

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

'''2. Penalty method'''

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

'''3. Penalty method'''

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", ''IEEE Transactions on Evolutionary Computation'', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:52:07Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

'''0. no constraint handling'''

This option turns off the constraint handling. Objective values and constraints are not changed.

'''1. delete constraints'''

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

'''2. Penalty method'''

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

'''3. Penalty method'''

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", ''IEEE Transactions on Evolutionary Computation'', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:18:36Z

Dirk: /* References */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", ''IEEE Transactions on Evolutionary Computation'', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:18:23Z

Dirk: /* Source Code */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:18:15Z

Dirk: /* Usage */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

-

==Source Code==

ToDo:Link to SVN

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:18:01Z

Dirk: /* Usage */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==

https://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optim/tools/ConstraintHandler.java

==Source Code==

ToDo:Link to SVN

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:16:39Z

Dirk: /* References */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
(1) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:16:00Z

Dirk: /* References */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
(<div id="1">1</div>) T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:15:44Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking ([[#1|1]])

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
[<div id="1">1</div>] T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:15:26Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking [[[#1|1]]]

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
[<div id="1">1</div>] T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:14:54Z

Dirk: /* References */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking [[#1|1]]

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
[<div id="1">1</div>] T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:14:35Z

Dirk: /* References */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking [[#1|1]]

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==
[1]<div id="1">1</div> T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

Documentation/Modules/ConstraintHandler

2015-10-25T18:13:17Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking [[#1|1]]

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T18:12:11Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed: sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking algorithm of the paper

T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T18:11:51Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

While bjectives are to be minimized (<code>min('''f''')</code>), constraints are to be fulfilled (<code>'''g''' =< 0</code>).

0. no constraint handling

This option turns off the constraint handling. Objective values and constraints are not changed.

1. delete constraints

Deletes all constraints. This can be used if the goal is to minimize the objective function(s) only. The objective functions are not changed.

2. Penalty method

All objective function values are summed. All violated constraints (i.e. the constraint value is >= 0) are summed and multiplied with a penalty factor <code>p</code>:

output = sumi(fi) + sumj(max(gj,0))*p

Increasing the penalty factor <code>p</code> puts more weight on the constraints compared to the objectives.

3. Penalty method

This algorithm works only for population based search methods such as CMA-ES. There are 3 steps

a. All objective function values are summed: sumi(fi)
b. All violated constraints (i.e. the constraint value is >= 0) are summed sumj(max(gj,0))
c. Ranking of the solutions according to the Stochastic Ranking algorithm of the paper

T. P. Runarsson and X. Yao, "Stochastic Ranking for Constrained Evolutionary Optimization", '''IEEE Transactions on Evolutionary Computation''', Vol. 4, No. 3, pp. 284-294, Sep. 2000.

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T17:54:19Z

Dirk: /* Module Description */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The result of an evaluation of a solution may contain objective values '''f''' and constraint values '''g''':

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T17:53:06Z

Dirk: /* Properties */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| not affected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The optimization algorithm proposes one or several new solutions '''x'''. Some values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by thress different methods:

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/ConstraintHandler

2015-10-25T17:48:40Z

Dirk: /* Summary */

==Summary==

Some optimization problems define [[Documentation/Notation#objectives | objectives]] as well as
[[Documentation/Notation#constraints | constraints]].

For example, one can set for the optimization problem [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]] the weight of the truss to be minimized, while a constraint is set on the maximum stress and displacement of the truss.

This module aggregates objectives and constraints into a single output.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| Design variables values outside the boundaries are corrected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The optimization algorithm proposes one or several new solutions '''x'''. Some values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by thress different methods:

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules

2015-10-25T17:46:41Z

Dirk: /* Optimization Problems */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>]]: Nelder-Mead Simplex Algorithm

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>]]: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>]]: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>]]: Elitist Evolution Strategy with Covariance Matrix Adaptation
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>]]: Particle Swarm Optimization Algorithm

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>]]: A Module Containing Different DoE plans
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>]]: Uniform Random Sampling (Monte Carlo)

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code>]]: For Optimization Algorithms without Bound Handling
* [[Documentation/Modules/ConstraintHandler | <code>ConstraintHandler </code>]]: For Optimization Algorithms without Constraint Handling

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code>]] A Simple Test Problem
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>]]: A Set of Single Objective Test Problem
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>]]: A Set of Multi-Objective Test Problem

* [[Documentation/Modules/ProblemTruss | <code>ProblemTruss </code>]]: The goal is to opzimize the thickness of 10 trusses. The weight of the truess, maximum stress, and displacement can each be set either as objective or constraint.

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>]]: Artificial Neural Network

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>]]: A Framework for Surrogate Managing

Documentation/Modules/ConstraintHandler

2015-10-25T17:42:24Z

Dirk: Created page with "==Summary== Some optimization problems require objective functions as well as constraints. require bounds on the design variables '''x..."

==Summary==

Some optimization problems require objective functions as well as [[Documentation/Notation#constraints | constraints]].

require bounds on the design variables '''xxx'''. E.g. for the diameter of any mechanical support must be above a certain minimum value.

As not all optimization algorithms include bound handling, the module <code>BoundHandler</code> was created.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| Design variables values outside the boundaries are corrected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The optimization algorithm proposes one or several new solutions '''x'''. Some values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by thress different methods:

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Notation

2015-10-25T17:39:39Z

Dirk: /* Terminology */

== Notation ==

As a math parser is currently not used, we write the math equations as formated text:
* Scalars are written as small italic letters, e.g. ''f''
* Vectors are written as small bold letters, e.g. '''x'''.
* Matrices are written in capital bold letters, e.g. '''C'''.

== Symbols ==

:{| cellpadding="5" width="95%"
|-
| ''a'', '''a'''
| additionals, i.e. additional value(s)
|-
| ''f'', '''f'''
| objective function(s)
|-
| ''g'', '''g'''
| constraint(s)
|-
| ''x'', '''x'''
| design variable(s)
|-
|}

== Terminology ==

:{| cellpadding="5" cellspacing="0" width="95%" border="1"
|-
| ''additionals'' <div id="additionals"></div>
| Additionals '''a''' are additional values in a '''solution''' that are for information only and do not influence the optimization. For example, if an optimization problem returns an objective function that is a sum of several values, these values could be added to the solution as additional values '''a'''.
|-
| ''constraints'' <div id="constraints"></div>
| Constraints '''g''' are criteria that have to be fulfilled. OpenDino defines a constraint ''g'' as
* fulfilled, if ''g'' =< 0
* violated, if ''g'' > 0.

The most simple constraint handling in optimization is to add a penalty to the objective function if the constraint is vialoted, resulting in

''f'' + max(0, ''g'').
|-
| ''design variables'' <div id="design_variables"></div>
| The vector of design variables '''x''' may consist of real numbers (continuous variables), integers (discrete variables) or both (mixed variables).
|-
| ''objectives'' <div id="objectives"></div>
| An optimization problem must have at least one objective or one constraint. OpenDino requires that the objective(s) '''f''' is/are to be minimized. A maximization of a function ''k'' can be converted into minimization by using ''f'' = -''k''.
|-
| ''solution'' <div id="solution"></div>
| The term solution is used in optimization. One solution ('''x''','''f''','''g''','''a''') consists of a vector of design variables '''x''', the evaluated objective(s) '''f'''('''x'''), constraint(s) '''g'''('''x'''), and optionally of additional values '''a'''.
|-|}

Documentation/Notation

2015-10-25T17:38:30Z

Dirk: /* Terminology */

== Notation ==

As a math parser is currently not used, we write the math equations as formated text:
* Scalars are written as small italic letters, e.g. ''f''
* Vectors are written as small bold letters, e.g. '''x'''.
* Matrices are written in capital bold letters, e.g. '''C'''.

== Symbols ==

:{| cellpadding="5" width="95%"
|-
| ''a'', '''a'''
| additionals, i.e. additional value(s)
|-
| ''f'', '''f'''
| objective function(s)
|-
| ''g'', '''g'''
| constraint(s)
|-
| ''x'', '''x'''
| design variable(s)
|-
|}

== Terminology ==

:{| cellpadding="5" cellspacing="0" width="95%" border="1"
|-
| ''additionals''
| Additionals '''a''' are additional values in a '''solution''' that are for information only and do not influence the optimization. For example, if an optimization problem returns an objective function that is a sum of several values, these values could be added to the solution as additional values '''a'''.
|-
| ''constraints''
| Constraints '''g''' are criteria that have to be fulfilled. OpenDino defines a constraint ''g'' as
* fulfilled, if ''g'' =< 0
* violated, if ''g'' > 0.

The most simple constraint handling in optimization is to add a penalty to the objective function if the constraint is vialoted, resulting in

''f'' + max(0, ''g'').
|-
| ''design variables''
| The vector of design variables '''x''' may consist of real numbers (continuous variables), integers (discrete variables) or both (mixed variables).
|-
| ''objectives''
| An optimization problem must have at least one objective or one constraint. OpenDino requires that the objective(s) '''f''' is/are to be minimized. A maximization of a function ''k'' can be converted into minimization by using ''f'' = -''k''.
|-
| ''solution'' <div id="solution"></div>
| The term solution is used in optimization. One solution ('''x''','''f''','''g''','''a''') consists of a vector of design variables '''x''', the evaluated objective(s) '''f'''('''x'''), constraint(s) '''g'''('''x'''), and optionally of additional values '''a'''.
|-|}

Documentation/Modules/BoundHandler

2015-10-25T11:26:46Z

Dirk: /* Properties */

==Summary==

Some optimizations require bounds on the design variables '''xxx'''. E.g. for the diameter of any mechanical support must be above a certain minimum value.

As not all optimization algorithms include bound handling, the module <code>BoundHandler</code> was created.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic (as no gradient handling is implemented)
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| Design variables values outside the boundaries are corrected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The optimization algorithm proposes one or several new solutions '''x'''. Some values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by thress different methods:

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/BoundHandler

2015-10-25T11:25:56Z

Dirk: /* Module Description */

==Summary==

Some optimizations require bounds on the design variables '''xxx'''. E.g. for the diameter of any mechanical support must be above a certain minimum value.

As not all optimization algorithms include bound handling, the module <code>BoundHandler</code> was created.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| Design variables values outside the boundaries are corrected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

The optimization algorithm proposes one or several new solutions '''x'''. Some values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by thress different methods:

0. no bound handling

This option turns off the bound handling.

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu)^2 / (x - xu)

else if x > xu, then x = xl + (xu- xl)^2 / (x - xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules/BoundHandler

2015-10-25T11:11:30Z

Dirk: /* Properties */

==Summary==

Some optimizations require bounds on the design variables '''xxx'''. E.g. for the diameter of any mechanical support must be above a certain minimum value.

As not all optimization algorithms include bound handling, the module <code>BoundHandler</code> was created.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| deterministic
|-
! Design Variables
| continuous variables, discrete or mixed variables are possible.
|-
! Objectives
| any number
|-
! Constraints
| any number
|-
! Boundaries
| Design variables values outside the boundaries are corrected
|-
! Initial Search Region
| not affected
|-
! Typical X
| not affected
|-
! Initialization
| not required
|-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at his module
| One connection of type <code>optimization</code>
|-
! Ending at this module
| One connection of type <code>optimization</code>
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! -
| -
|-
|}

===Options===
The options are currently described as "pop-up help".

==Module Description==

Any optimization algorithm proposes a new solution '''x'''. Some of the values of '''x''' might be outside the lower bounds xl and upper bounds xu, specified in the "Problem" module. These variables are corrected to values within the bounds by two different methods:

1. set to bounds

If one of the variables x is below or above the bounds, it is set to the bound value, i.e.

if x < xl, then x = xl
else if x > xu, then x = xu

2. reflect

If one of the variables x is below or above the bounds, it is reflected from the bound.
The reflection is done such that if x goes to infinity, x is equal to the lower bound and if x goes to
minus infinity, it is set to the upper bound.

if x < xl, then x = xu + (xl- xu) / (x - xu) * (xl- xu);

else if x > xu, then x = xl + (xu- xl) / (x - xl) * (xu- xl)

==Usage==
-
==Source Code==

ToDo:Link to SVN

==References==

-

Documentation/Modules

2015-10-25T11:10:43Z

Dirk:

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>]]: Nelder-Mead Simplex Algorithm

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>]]: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>]]: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>]]: Elitist Evolution Strategy with Covariance Matrix Adaptation
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>]]: Particle Swarm Optimization Algorithm

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>]]: A Module Containing Different DoE plans
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>]]: Uniform Random Sampling (Monte Carlo)

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code>]]: For Optimization Algorithms without Bound Handling
* [[Documentation/Modules/ConstraintHandler | <code>ConstraintHandler </code>]]: For Optimization Algorithms without Constraint Handling

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code>]] A Simple Test Problem
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>]]: A Set of Single Objective Test Problem
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>]]: A Set of Multi-Objective Test Problem

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>]]: Artificial Neural Network

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>]]: A Framework for Surrogate Managing

Documentation/Modules

2015-10-25T11:09:03Z

Dirk: /* Indirect, Stochastic Algorithms */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>]]: Nelder-Mead Simplex Algorithm

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>]]: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>]]: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>]]: Elitist Evolution Strategy with Covariance Matrix Adaptation
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>]]: Particle Swarm Optimization Algorithm

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>: A Module Containing Different DoE plans]]
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>: Uniform Random Sampling]]

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> For Optimization Algorithms without Bound Handling]]
* [[Documentation/Modules/ConstraintHandler | <code>BoundHandler</code> For Optimization Algorithms without Constraint Handling]]

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code> A Simple Test Problem]]
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>: A Set of Single Objective Test Problem]]
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>: A Set of Multi-Objective Test Problem]]

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>: Artificial Neural Network]]

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>: A Framework for Surrogate Managing]]

Documentation/Modules

2015-10-25T11:08:48Z

Dirk: /* Direct, Deterministic Algorithms */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>]]: Nelder-Mead Simplex Algorithm

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES]]
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES]]

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>]]: Elitist Evolution Strategy with Covariance Matrix Adaptation
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>]]: Particle Swarm Optimization Algorithm

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>: A Module Containing Different DoE plans]]
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>: Uniform Random Sampling]]

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> For Optimization Algorithms without Bound Handling]]
* [[Documentation/Modules/ConstraintHandler | <code>BoundHandler</code> For Optimization Algorithms without Constraint Handling]]

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code> A Simple Test Problem]]
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>: A Set of Single Objective Test Problem]]
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>: A Set of Multi-Objective Test Problem]]

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>: Artificial Neural Network]]

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>: A Framework for Surrogate Managing]]

Documentation/Modules

2015-10-25T11:08:17Z

Dirk: /* Single and Multi-Objective Optimization Algorithms */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>: Nelder-Mead Simplex Algorithm]]

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES]]
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES]]

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>]]: Elitist Evolution Strategy with Covariance Matrix Adaptation
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>]]: Particle Swarm Optimization Algorithm

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>: A Module Containing Different DoE plans]]
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>: Uniform Random Sampling]]

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> For Optimization Algorithms without Bound Handling]]
* [[Documentation/Modules/ConstraintHandler | <code>BoundHandler</code> For Optimization Algorithms without Constraint Handling]]

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code> A Simple Test Problem]]
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>: A Set of Single Objective Test Problem]]
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>: A Set of Multi-Objective Test Problem]]

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>: Artificial Neural Network]]

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>: A Framework for Surrogate Managing]]

Documentation/Modules

2015-10-25T11:07:30Z

Dirk: /* Indirect, Stochastic Algorithms */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>: Nelder-Mead Simplex Algorithm]]

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>: 1+1 Evolution Strategy with 1/5 Success Rule: the (1+1)-ES]]
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>: A Multi-member Evolution Strategy with Covariance Matrix Adaptation: the CMA-ES]]

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>: Elitist Evolution Strategy with Covariance Matrix Adaptation]]
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>: Particle Swarm Optimization Algorithm]]

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>: A Module Containing Different DoE plans]]
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>: Uniform Random Sampling]]

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> For Optimization Algorithms without Bound Handling]]
* [[Documentation/Modules/ConstraintHandler | <code>BoundHandler</code> For Optimization Algorithms without Constraint Handling]]

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code> A Simple Test Problem]]
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>: A Set of Single Objective Test Problem]]
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>: A Set of Multi-Objective Test Problem]]

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>: Artificial Neural Network]]

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>: A Framework for Surrogate Managing]]

Documentation/Modules

2015-10-25T11:06:17Z

Dirk: /* Single Objective Optimization Algorithms */

'''Modules''' contain all the functionality for optimizing and learning. One '''Modules''' may contain an optimization algorithm, an artificial neural network, or a problem to optimize.

Here is a list of documented modules in OpenDino. Further modules may exist, but may not yet be documented.

== Single Objective Optimization Algorithms ==

=== Indirect, Deterministic Algorithms ===
Indirect algorithms use gradient or higher order derivative information in the optimization.

Not implemented, yet.

=== Direct, Deterministic Algorithms ===
These algorithms neither use gradient information nor stochastic processes.

* [[Documentation/Modules/OptAlgSIMPLEX | <code>OptAlgSIMPLEX</code>: Nelder-Mead Simplex Algorithm]]

=== Indirect, Stochastic Algorithms ===

These algorithms do not use gradient information but require stochastic processes (i.e. random numbers) in their search.

* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgOpO|<code>OptAlgOpO</code>: 1+1 Evolution Strategy with 1/5 Success Rule]]
** [[Documentation/Modules/OptAlgCMA|<code>OptAlgCMA</code>: A Multi-member Evolution Strategy with Covariance Matrix Adaptation]]

== Single and Multi-Objective Optimization Algorithms ==

=== Indirect, Stochastic Algorithms ===
* Evolutionary Algorithms
** [[Documentation/Modules/OptAlgMoCMA | <code>OptAlgMoCMA</code>: Elitist Evolution Strategy with Covariance Matrix Adaptation]]
* Particle Methods
** [[Documentation/Modules/OptAlgMOPSO | <code>OptAlgMOPSO</code>: Particle Swarm Optimization Algorithm]]

== Design of Experiments ==
* [[Documentation/Modules/DoePlanner | <code>DoePlanner</code>: A Module Containing Different DoE plans]]
* [[Documentation/Modules/RandomSampling | <code>RandomSampling</code>: Uniform Random Sampling]]

== Optimization in General ==
* [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> For Optimization Algorithms without Bound Handling]]
* [[Documentation/Modules/ConstraintHandler | <code>BoundHandler</code> For Optimization Algorithms without Constraint Handling]]

== Optimization Problems ==
* [[Documentation/Modules/ProblemSimple | <code>ProblemSimple</code> A Simple Test Problem]]
* [[Documentation/Modules/ContinuousTestProblems | <code>ContinuousTestProblems </code>: A Set of Single Objective Test Problem]]
* [[Documentation/Modules/ContinuousMOTestProblems | <code>ContinuousMOTestProblems </code>: A Set of Multi-Objective Test Problem]]

== Machine Learning ==
* [[Documentation/Modules/NeuralNetwork | <code>NeuralNetwork</code>: Artificial Neural Network]]

== Miscellaneous Modules ==
* [[Documentation/Modules/SurrogateManager | <code>SurrogateManager</code>: A Framework for Surrogate Managing]]

Documentation/Modules/OptAlgCMA

2015-10-25T11:00:06Z

Dirk: /* Source Code */

==Summary==

This optimization module is an implementation of the Evolution Strategy with Covariance Matrix Adaptation (CMA-ES). It uses the source code of Nikolaus Hansen ( https://www.lri.fr/~hansen/cmaes_inmatlab.html#java ).

Among the tree evolutionary operators (Recombination, Mutation, Selection), the mutation is considered the most important. Mutation is performed by sampling a Covariance Matrix. The parameters of the matrix (variances and correlation coefficients) are adapted by tracking the path of successful mutations.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| stochastic - generates new solutions by sampling a probability function, however deterministic adaptation of the covariance matrix.
|-
! Design Variables
| Written for continuous variables. No discrete or mixed variables are possible.
|-
! Objectives
| single-objective for minimization.
|-
! Constraint handling
| no, use e.g. the Module [[Documentation/Modules/ConstraintHandler | <code>ConstraintHandler</code>.

|-
! Boundary handling
| yes, the module [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> is integrated in this module.
|-
! Initialization
| Requires at least one of the following: initial solution, initial search region, or bounds. |-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at this module
| Module requires exactly one connection of type <code>optimization</code>.
|-
! Ending at this module
| -
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! '''Run'''
| starts the optimization.
|-
|}

===Options===
The options are currently only described in the references (see below).

==Module Description==

=== Initialization ===

The algorithm distinguishes several cases, depending on the problem properties (<code>initial solution, initial search region, bounds</code>).

=== Optimization ===
The algorithm contains stochastic processes and operates with a population. Parallelization on the basis of the population size is implemented.

==Usage==

... todo

==Source Code==

http://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/fr/inria/optimization/cmaes/

http://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optAlg/OptAlgCMA.java

==References==

Detailed information is given at http://www.lri.fr/~hansen/ .

Documentation/Modules/OptAlgCMA

2015-10-25T10:59:54Z

Dirk: /* Source Code */

==Summary==

This optimization module is an implementation of the Evolution Strategy with Covariance Matrix Adaptation (CMA-ES). It uses the source code of Nikolaus Hansen ( https://www.lri.fr/~hansen/cmaes_inmatlab.html#java ).

Among the tree evolutionary operators (Recombination, Mutation, Selection), the mutation is considered the most important. Mutation is performed by sampling a Covariance Matrix. The parameters of the matrix (variances and correlation coefficients) are adapted by tracking the path of successful mutations.

==Properties==

===General===

{| class="wikitable" style="text-align:left" border="2" cellpadding="5" cellspacing="5"
|-
! Algorithm
| stochastic - generates new solutions by sampling a probability function, however deterministic adaptation of the covariance matrix.
|-
! Design Variables
| Written for continuous variables. No discrete or mixed variables are possible.
|-
! Objectives
| single-objective for minimization.
|-
! Constraint handling
| no, use e.g. the Module [[Documentation/Modules/ConstraintHandler | <code>ConstraintHandler</code>.

|-
! Boundary handling
| yes, the module [[Documentation/Modules/BoundHandler | <code>BoundHandler</code> is integrated in this module.
|-
! Initialization
| Requires at least one of the following: initial solution, initial search region, or bounds. |-
|}

===Connections===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Starting at this module
| Module requires exactly one connection of type <code>optimization</code>.
|-
! Ending at this module
| -
|-
|}

===Actions===
{| class="wikitable" style="text-align:left" border="1" cellpadding="5" cellspacing="0"
|-
! Name !! Description
|-
! '''Run'''
| starts the optimization.
|-
|}

===Options===
The options are currently only described in the references (see below).

==Module Description==

=== Initialization ===

The algorithm distinguishes several cases, depending on the problem properties (<code>initial solution, initial search region, bounds</code>).

=== Optimization ===
The algorithm contains stochastic processes and operates with a population. Parallelization on the basis of the population size is implemented.

==Usage==

... todo

==Source Code==

http://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/fr/inria/optimization/cmaes/
http://sourceforge.net/p/opendino/code/HEAD/tree/trunk/src/org/opendino/modules/optAlg/OptAlgCMA.java

==References==

Detailed information is given at http://www.lri.fr/~hansen/ .