Parametric System

A parametric system is the base of each algorithm. The system contains a parametric process chain. A design is send to be evaluated by a solver or process chain and then send back to the parametric system (parent system). Running a parametric system means the evaluation of the reference design.

To access the following tabs and options, double-click Parametric System on the Scenery pane.

Parameter Tab

The parameter set presents all parameters by row in a table. You can change parameter properties by double-clicking the cell and save using Enter.

When changing properties for several selected parameters use the Shift or Ctrl keys to select multiple cells.

Option	Description
Name	Displays the parameter name.
Parameter type	Defines the parameter type. Select one of the following: Optimization Stochastic Opt.+Stoch. Dependent
Reference value	Displays the reference value (for example, parsed value for text input). For dependent parameters, the reference value is the evaluated expression.
Constant	When selected, sets the parameter as a constant. If a parameter is set as constant, the reference value is used. Filtered parameters (MOP) cannot be set as constants.
Operation	Defines the expression of dependent parameters using mathematical functions. To define conditional dependent parameters, use conditional operators. The default expression is the current reference value. More complex mathematical expressions can be formulated using a Calculator component.
Value type	Defines the value type. Select one of the following: Real Integer Bool String
Deterministic Properties These properties are available only when you have selected Optimization or Opt. + Stoch. as the parameter type.
Resolution	Defines the resolution of the parameter. Select one of the following: Continuous Discrete by value Discrete values are used directly for calculations. Ordinal discrete (by index) Discrete values are translated to indices to obtain equidistant discrete states. Nominal discrete Discrete values are translated to indices but there is no order between all states. Note: The following dependencies exist for this selection. Mixed parameters can only be continuous or discrete by value. Continuous is only available for real types. Discrete by value is only available for real and integer types. Ordinal discrete (by index) and nominal discrete are available for all types.
Range	Sets either the lower and upper bounds for continuous parameters, or discrete values for discrete parameters. Bounds of continuous parameters can be defined relative to the reference value by using a %. A minus is implicitly applied to the lower bound value. For example, [10% 10%] results in: lower_bound = reference_value - reference_value * 10 % upper_bound = reference_value + reference_value * 10 % Defining both values within one cell, mixing of relative and absolute values and the multiset feature are also supported. To update the range for continuous parameters, either change the value for each cell or type both values in one cell with a space between each value. For discrete parameters that are not strings, type the discrete values with spaces in between. Discrete string parameters must be spaced using ;. If the discrete string values are put into quotation marks, the quotation marks are included part of the string. Ranges can be applied only for equidistant discrete integer or real parameter using - and optionally [], for example, [1-5] ; 1-5 ; 7,[10-12]; and negative ranges -1--5. The step size of the range can be varied with the pattern [minimum:maximum:interval], for example [8:12:2].
Range Plot	Displays an overview of the range. For discrete parameters, the displayed number of blocks corresponds to the number of discrete states. Reference values are represented as vertical gray bar.
Stochastic Properties These properties are available only when you have selected Stochastic or Opt. + Stoch. as the parameter type.
PDF	Displays the probability distribution function (PDF).
Type	Sets the distribution type. Descriptions of the types of distribution can be found in Distribution Types.
Std. Dev.	Sets the standard deviation of a distribution.
Mean	Sets the mean value of a distribution.
Distribution parameter	Appropriate distribution types are listed in Distribution Types. When defining less parameter than required, default values will be set for the missing ones.

The dependency between Parameter, Mean or Standard deviation is automatically evaluated.

Action Button

Use the multi-selection button on the bottom right-hand side to perform the following actions.

Import parameter from system
Parameters can be imported from any other system in the project. Select one or more parameters and click OK. A copy of the parameter with all its properties is added to the current parametric system. Modifications can be applied then without changing the original parameter. The following filters for existing systems are available:
- Show all systems (no filtering is applied)
- Has result designs
- Has parameters (default)
- Has criteria
- Name contains: (shows only systems, whose names contain the specified string)
Import parameter from file
Parameters can be imported from a *.csv file. After clicking this option:
1. Browse for the file containing the parameters and click Open.
2. Select one or more parameters to use from the import table.
3. Click OK
A copy of each parameter with all its properties is merged into the current parametric system.
Limitation: You will get a failure on import if the file contains single backslashes. You must ensure that the backslashes are doubled:
- Incorrect: C:\Windows\Temp
- Correct: C:\\Windows\\Temp
Export parameter
Parameters can be exported as *.csv file. Each row in the resulting file represents one parameter. The values are separated by a user-defined delimiter.
1. Select the parameters to export.
2. Click Export Parameter.
3. In the File name field, enter a file name.
4. Click Save.
5. In the Delimiters dialog, select the character that separates the values.
6. Click OK
Use design as reference
A previously calculated design can be used to modify the reference value of the parameter. The design can be selected from any system of the project with matching parameters. Parameters must be available to use this option. If there are more parameter entries in the selected design, they are ignored. The reference value is only used for the matching parameter.
Edit correlations
Correlations can be defined for non-constant parameters with stochastic properties. Select all required parameters and select Edit correlations to open the correlation matrix. Enter the correlation coefficients into the corresponding box and click OK.

Sorting Parameter

Parameters can be sorted in the table by column header in ascending or descending order. You can switch the sorting order by clicking on the column header or by using the Sort by context menu of the parameter table. Sorting by creation order resets the parameter order to its original order.

Merging

When parameters are passed to a system using the slot IParameterManager, the resulting parameter set is composed using the following rules:

The resulting parameter set is always the union of parameters passed to the slot and parameters defined in the system.

If there is an intersection (for example, common parameter names in the defined set and the imported parameters), the following rules apply for these parameters:

Parameter merging is Prefer defined: the parameter provided at the slot is ignored and the defined parameter is used.
Parameter merging is Prefer from slot: the parameter provided at the slot is used and the defined parameter is ignored.
Parameter merging is Merge from slot: the parameter provided at the slot and parameter defined in the system are merged using following rules:
- The defined parameter is used.
- Flag Const is merged by logical Or. Thus, a non-filtered parameter can be set to const additionally.
- Flag Filtered (flag Const is fixed) is used from the parameter provided at the slot. Thus, filtered parameter cannot be set to non-const.

Criteria Tab

Criteria are used by algorithms to define goals and restrictions. Criteria are formulated as a mathematical expression of the values of the input parameters and responses. Recurring expressions can be defined as free variables.

In optiSLang, parameters and responses can be dragged and dropped to the Criteria table or Create new on the corresponding icon.

Optimization Criteria

One or more objective functions are required for optimization problems. Constraints are restrictions of an optimization problem, and be defined as inequalities. A design is only feasible if all constraints are fulfilled.

Reliability Criteria

For reliability analysis, at least one limit state function has to be defined. A design is only safe if all state functions are fulfilled.

Option	Description
Name	The criteria name
Type	The type of the criterion. Possible values are Variable, Objective, Constraint or Limit state.
Expression	The expression text field supports completion of known function and variable names. More complex mathematical expressions can be easily formulated with the help of a Calculator component. To open the calculator either select Open calculator in the context menu of the Expression cell or select the expression cell and press Ctrl+Shift+o. All parameters, variables, and responses are available and are autocompleted for easier typing.
Criterion	Possible objective criteria are minimizing MIN or maximizing MAX. Maximization problems will be transformed into minimization problems internally. Possible constraint criteria are ≤ and ≥. Internally all less equal constraints are transformed to greater equal constraints. Possible limit state criteria are ≤ and ≥.
Limit	The comparison value for the limiting expression of a constraint or limit state.
Evaluated expression	The result of the comparison expression. An infeasible or unsafe reference design is highlighted red.

Action Buttons

Use the multi-selection button on the bottom right-hand side to perform the following actions.

Import criteria from system
Import previously defined criteria from other systems. The following filters for existing systems are available:
- Show all systems (no filtering is applied)
- Has result designs
- Has parameters (default)
- Has criteria
- Name contains: (shows only systems, whose names contain the specified string)
Import criteria from file
Import criteria from an optiSLang problem definition file (*.pro) or comma-separated values file (*.csv). After clicking this option:
1. Browse for the file containing the criteria and click Open.
2. Select one or more criteria to use from the import table.
3. Click OK
Export criteria
Criteria can be exported as *.csv file. Each row in the resulting text file is representing one criterion. The values are separated by a user-defined delimiter.
1. Select the parameters to export.
2. Click Export criteria.
3. In the File name field, enter a file name.
4. Click Save.
5. In the Delimiters dialog, select the character that separates the values.
6. Click OK

Other Tab

This tab contains working directory and design directory settings.

Option	Description
Working directory
Use a manually selected working directory	When selected, allows you to select a different existing working directory than the default working directory (for example, *opd/Sensitivity).
Remove before calculating	When set to Always, ensures that data is kept consistent after modifying the system. Failed designs may not be recognized if the design directory from previous runs was kept. The Remove working directory dialog is displayed just before the run.
Remove after completion	Select if and when a design directory is removed after the design calculations have finished. If succeeded: Removes the complete directory for All or filtered file types for Any file not matching user-defined filter. If failed: Removes the complete directory for All or filtered file types for Any file not matching user-defined filter. If empty: Removes the directory if no files are contained. The user-defined filter are rules to be applied when purging. As long as no rules are defined, no files are removed. Otherwise, any file is removed, that doesn't match a filter rule. The default filter rule contains the extensions .omdb, .bin and *.txt.
Design directory format
Use a manually selected format:	When selected, design directories of the current parametric system are created with the defined name format. The format string should contain a placeholder for integer numbers representing the design number, for example, `Design%04d` creating directory Design0001 for the first design and `Design%02d` creating directory Design01. Integers are represented by %d, %i, %x, or %o. In parametric and algorithm systems, these directories are created. Reevaluation systems search for design directories of this format to read designs. If this option is not selected, the default format defined in the configuration file is used.
Postprocessing
Write optiSLang post-processor database file	Select how often the post-processor database file is written. Optimization and Reliability algorithms allow post-processor file writing for each iteration or at the end, Sampling algorithms allow post-processor file writing for each design or at the end.
Write classic database file (*.bin) additionally	When selected, an optiSLang version 3 type database file is written additionally to the omdb file.
Show as soon as available	When selected, show postprocessing during the run as soon as it is available or after a certain number of designs.
Show on termination	When selected, the postprocessing opens after the algorithm terminates. An additional Postprocessing node is not required.
Register files	When selected, defines file filter rules. Any file matching the defined filters is registered. Images: Defines a filter for common image formats such as .bmp, .gif, .ico, .jpeg, .png and so on. Ansys Viewer files: Defines a filter for Ansys Viewer .avz files. VCollab CAX/wCAX files: Defines a filter for the visualization file formats of the VCollab software. Custom*: Opens a dialog, allowing you to create custom filter settings.
Evaluation
Solve duplicated designs	When selected, identical designs are calculated again. When cleared, the results are fetched from the database.
Solve violated input constraints	When selected, constraints that depend only on parameters are checked before starting calculations. When cleared, designs that violate input constraints are not evaluated and the designs are automatically set to failed.
Solve start designs again	You can define designs using the input slot IStartDesigns or in the start designs tab. If cleared, all designs that must be evaluated are checked. If a calculated design for the requested design point already exists in start designs, a copy of the results replaces the real evaluation of the design.
Solve reference design	If selected, the reference values are calculated at the end and added to result designs. For algorithm systems, this design does not appear in the design table of Postprocessing. For reliability or robustness systems, this design is also excluded from the output slot vector of registered responses.
Solve nominal design	This option is only available for robustness and reliability systems. If selected, the nominal design is calculated at the end and added to result designs. This design does not appear in the design table of Postprocessing or in the output slot vector of registered responses.
Preserve start design IDs	This option is only available for sensitivity and robustness systems and enforces that the designs preserve the IDs of the original system.
Use manual seed	If selected, algorithms containing stochastic components will use a manually defined seed value instead of a randomly generated one.
Seed value	If the Use manual seed option is selected, enter the seed value to be used here.
Algorithm messages
Show algorithm messages in project log	Algorithm specific messages will appear in the Message log window. When opening the project again old log messages will not be loaded, but the messages are saved in the Protocol.txt embedded in the project.
Write algorithm messages to log file	Only algorithm specific messages are written into an extra protocol file. If the text file is defined as relative path the message log file is created in the current working directory. If the text file is defined as absolute path the message log file is created in the selected directory. If the file already exists all messages will be appended.
Predefined Settings: Some algorithms (for example, Particle Swarm Optimization and Evolutionary Algorithm) include predefined settings. Select the required setting from the drop-down list.

Result Designs Tab

All designs awaiting calculation or that have been calculated appear in the result designs table. You can see design information in this tab before opening the postprocessing to visualize results.

Displayed float values are formatted using options provided in the Table/Tree Views section of the optiSLang configuration settings.

Table Column	Description
Id	Displays the design number in the corresponding algorithm. This column is read-only.
Feasible	Highlights the designs that fulfil all constraints (green). When you clear the Solve violated input constraints on the Other tab, violated designs are marked false before the calculation starts. This column is read-only.
Duplicates	Displays the design ID of a duplicated design. If the current design is identical to a previous design, the ID of the previous design is shown in this cell. The equality check is done before calculating designs. If the Solve duplicated designs check box is cleared on the Other tab, only the first design is calculated and the results are copied to all duplicated designs. This column is read-only.
Status	Displays the designs state (Succeeded, Failed, Pending, Needs Reevaluation). The state can be changed to calculate a specific design again.
Parameter	Displays read-only values for each parameter. The number of columns in the design table differs depending on the number of parameters.
Responses	Displays read-only values for each response. The number of columns in the design table differs depending on the number of responses.
Criteria	Displays read-only values for each criteria. The number of columns in the design table differs depending on number of criteria.

You can configure the table view (show/hide columns) using the context menu of the table header.

You can select the following options below the design table.

Selection mode: Set how you select information in the design table.
- Designs (select a row)
- Columns
- Individual Cells
Instant visualization: Displays visual information of the selected values. A graphical and a textual representation is available by switching the tab of the visualization widget. Multiple values can be visualized simultaneously.
Use as start design(s): Selected designs are copied to the Start designs tab.
Export designs: Selected designs are exported as a comma-separated value file (*.csv) or JSON file (*.json).
- *.csv file
  When creating a *.csv file, enter an output file name and select the delimiters (comma, semicolon, tab, or spaces).
- *.json file
  When creating a *.json file, enter an output file name. This option allows you to export complete design attributes and is useful for interchanging designs between optiSLang instances.

Additional Options

To access the options shown in the following table, in any tab, click Show additional options.

Option	Description
Limit maximum in parallel	Controls the resource usage of nodes in the system. When the check box is cleared (default), a value is chosen to ensure the best possible utilization of the child nodes. When the check box is selected, set the value manually to specify how many designs are sent to child nodes, limiting the maximum degree of parallelism for all children. Ansys recommends keeping the check box clear.
Auto-save behavior	Select one of the following options: No auto-save Actor execution finished Every n-th finished design (then select the number of designs from the text field) Iteration finished (for optimization, reliability) The project, including the database, is auto-saved (depending on defined interval) after calculating this node/system (either when the calculation succeeds or fails). By default, all parametric and algorithm systems have Every nth finished design 1 design(s) selected, all other nodes have No auto-save selected.

Option

Description

Limit maximum in parallel

Controls the resource usage of nodes in the system.

When the check box is cleared (default), a value is chosen to ensure the best possible utilization of the child nodes. When the check box is selected, set the value manually to specify how many designs are sent to child nodes, limiting the maximum degree of parallelism for all children. Ansys recommends keeping the check box clear.

Auto-save behavior

Select one of the following options:

No auto-save
Actor execution finished
Every n-th finished design (then select the number of designs from the text field)
Iteration finished (for optimization, reliability)

The project, including the database, is auto-saved (depending on defined interval) after calculating this node/system (either when the calculation succeeds or fails).

By default, all parametric and algorithm systems have Every nth finished design 1 design(s) selected, all other nodes have No auto-save selected.