Recent News
SNAP Version 4.0.4 Released
July 7, 2022
AptPlot 8.0.1 Updated
April 29, 2022
PyPost 4.0.2 Released
July 7, 2022

Uncertainty Plug-in Change Log

Version 1.8.1 - Released 4/29/2022:

  • The uncertainty quatification framework was updated and tested to support DAKOTA version 6.15. This version is now bundled with the plug-in for Windows installations.

Version 1.8.0 - Released 10/29/2021:

  • An issue was resolved that prevented Uncertainty reports from being written for job streams executed using containerized versions of SNAP that do not include user preferences.
  • An issue was resolved in the Uncertainty plug-in that prevented an Extract_Data application definition from being automatically created when necessary. This update also includes support for the integration of AptPlot with SNAP 4.0.0.
  • An issue was resolved that could cause the Uncertainty Quantification report process to fail when the specified Figure of Merit names were longer than three characters.
  • An issue was resolved that caused EMF files created during the Uncertainty Quantification report creation process to remain after the process completes.

Version 1.7.5 - Released 5/21/2021:

  • The Python bindings for Uncertainty Quantification were updated to support platforms (such as AWS Batch) which use remote locations for intermediate files staged between tasks.
  • The Python API documentation for Python directed Uncertainty Quantification streams was updated to include more detailed examples and descriptions. This documentation is available in:
    [SNAP_Install]/doc/python/snap/streams/uq/index.html]
  • An issue was resolved that prevented Loguniform distributions from being included in the Uncertainty Report for Python directed job streams. /li>
  • An issue was resolved that caused a failure in the uncertainty report generator when none of the jobs included in a Python directed stream completed successfully.

Version 1.7.4 - Released 12/23/2020:

  • The Uncertainty Quantification support in Python Directed streams was updated to support a specified number of "Replacement Samples" that are used to run additional tasks to replace those that fail to execute.

Version 1.7.3 - Released 11/6/2020:

  • The data extraction step was updated to support RELAP5-3D plot files and demultiplexed plot files.

Version 1.7.2 - Released 10/28/2020:

  • An issue was resolved that prevented the Uncertainty plug-in from generating a Report unless both the Title Page and Front Matter note were selected.

Version 1.7.1 - Released 10/25/2020:

  • Support for uncertainty quantification analysis using DAKOTA has been added to Python Directed streams. This feature can be accessed via the "Uncertainty Quantification" property of any Python Directed stream component.
  • An issue was resolved that could prevent atexit calls in the SNAP and PyPost python modules from being properly processed when running with Jython.

Version 1.7.0 - Released 3/4/2020:

  • Updated to work with Java 11. Methods that have been deprecated in Java 11 have been refactored to use the appropriate Java 11 equivalent.

Version 1.6.0 - Released 5/23/2019:

  • The uncertainty quatification framework was updated to support DAKOTA version 6.9. This version is now bundled with the plug-in for Windows installations.
  • The uncertainty plug-in has been updated to support defining additional variable properties when creating a new variable reference. This allows for TRACE sensitivity coefficients that include component cross references.

Version 1.5.2 - Released 10/30/2018:

  • The uncertainty quatification framework was updated to support DAKOTA version 6.8. This version is now bundled with the plug-in for Windows installations.
  • The Dakota report step has been updated to suppress an extraneous error message that was reported by Dakota when analyzing the uncertainty quantification results.
  • The legacy odt output file definition which was causing an error to be reported after executing a report job step has been removed from the uncertainty job step. This file definition was replaced by the report file definition which handles either odt or docx output files.

Version 1.5.1 - Released 6/21/2018:

  • The histogram distribution has been updated to report an error if an internal bin node has a magnitude less than 1e-15. This reflects a limitation of the DAKOTA software.

Version 1.5.0 - Released 2/28/2018:

  • The Uncertainty module's report generation has been updated to support the Microsoft Word (DOCX) file format in addition to the OpenOffice (ODT) format. The document format can be configured via the Report tab in the Uncertainty Stream's Parametric Properties editor.

Version 1.4.2 - Released 11/21/2017:

  • The DAKOTA report generator was updated to resolve an issue where the table that contains the variate data could trim the negative sign off values whose magnitude is less than 1000.0, but not essentially zero.
  • The error handling for parsing DAKOTA output files of unsupported versions now identifies why the parsing could not complete.

Version 1.4.1 - Released 7/17/2017:

  • An issue has been resolved which would cause the variable table in the parametric properties to show a large negative number for the nominal value when a variable reference cannot be resolved.

Version 1.4.0 - Released 6/16/2017:

  • The uncertainty quatification framework was updated to support DAKOTA version 6.6. This version is now bundled with the plug-in for Windows installations.
  • Numerics variables may now be used to define the distribution parameters in an Uncertainty Quantification job stream.

Version 1.3.6 - Released 4/21/2016:

  • The Extract Data job step has been updated to handle extracting data from FRAPTRAN runs.

Version 1.3.5 - Released 8/12/2016:

  • The Extract Data and AptPlot batch job step have been updated to handle extracting data from FRAPCON runs.

Version 1.3.4 - Released 4/13/2016:

  • The validation logic has been updated to report more useful error messages when distribution properties have not been set properly and to report a warning when the number of samples is less than the number required for the current properties.
  • The validation logic no longer generates error messages for unused distributions.
  • The parametric properties editor has been modified to show the text "Invalid Values" when a distribution value has not been set.

Version 1.3.3 - Released 12/18/2015:

  • The DAKOTA Uncertainty Quantification job stream type has been updated to allow selecting multiple time values for acquiring the figures of merit. The selected times are defined in the Time Dependent editor of the DAKOTA properties tab. The Extract Data step will generate a vector of FOM values, one for each time specified. The script is written using the token to specify the current time. The results for performing the UQ analysis at each time step are combined into a single report containing detailed results at each time step and a summary section documenting the time dependent results.
  • Uncertainty stream job steps may now use tokens to embed numeric variable values in custom keywords.

Version 1.3.2 - Released 11/12/2015:

  • Updated to work with Dakota Version 6.2.
  • The full Dakota 6.2 distribution from Sandia is now installed with the plug-in on Windows platforms.
  • The plug-in has been updated to allow the user to select a different Dakota application.
  • The generated report has been modified to include a plot of the probability density function for each response variable defined.
  • The floating point values ouput in the final report are now formatted to improve readability.
  • References to the response variables in the final report now use the short description provided by the user where appropriate, to improve general readability.
  • The variates and FOM table in the report now display the model variable name that is modified by each variate, or the first variable, if more than one variable is modified by the variate.

Version 1.3.1 - Released 10/14/2015:

  • An issue has been resolved which would prevent building an uncertainty job stream that uses factor probability distributions in an engineering template.
  • The initial values for factor probability distributions were updated to center around a factor of 1.

Version 1.3.0 - Released 8/12/2015:

  • Several minor issues were resolved that prevented compiling with Java version 7. Java 7 or newer is now required to run SNAP.
  • The Lognormal Distribution has been modified to include a minimum value. This value is optional and defaults to 0.0. If supplied, this value must be greater than or equal to 0.0.

Version 1.2.5 - Released 7/11/2013:

  • The report configuration now includes an option to sort tabulated variate and response data by task number, the variates generated by a user-designated distribution, or the response data for a user-designated figure of merit.
  • The report configuration now includes an option to include the data extraction script used to pull the response data out of parametric tasks.
  • The report now lists the plug-in and its version for all plug-ins used in the uncertainty stream.
  • The report now lists the application name, description, and executable location for all steps upstream of the DAKOTA Uncertainty step.
  • The input options now list the model variables modified by the run and the figures of merit with their descriptions.
  • Various descriptive paragraphs were added to the report, generated specifically for the uncertainty analysis configuration and results: this include descriptions for the DAKOTA input file, the order statistics results, the order statistic result iterations, and a general DAKOTA results introduction.
  • The sections were restructured so that all uncertainty analysis configuration is defined in the first section, the tabulated variates and response data are listed in the second section, the order statistics results are listed in the third section, and the DAKOTA results are listed in the fourth section.
  • The statistics for each DAKOTA response function now include the minimum response value, maximum response value, and median response value. The DAKOTA input file is now included as a sub-section of the DAKOTA results and is no longer an independent section.

Version 1.2.5 - Released 6/11/2013:

  • Updated to work with SNAP 2.2.4 Job Stream updates.

Version 1.2.4 - Released 3/29/2013:

  • The job stream submission processes were updated to support the new Job Stream naming constraints.

Version 1.2.3 - Released 10/31/12:

  • Uncertainty reports now include the SNAP version, model name, and MED file path (if the model has been saved) of the submitted uncertainty stream. These details are listed at the top of the table in the first section, Uncertainty Quantification Input Options.
  • An option is now provided to determine whether results matrix iterations are included in generated reports. As before, the iterations will only appear if the Order is enabled, regardless of whether the option is enabled.
  • Resolved an issue where task indexes in the results matrix iteration tables were off by one. Additionally, the text describing the structure of the results matrix iteration tables has been clarified.

Version 1.2.2 - Released 9/5/12:

  • An additional performance enhancement has been implemented for the Wilks sample count calculation. A look-up table of common probability, confidence, and FOM limit samples is included in the plug-in: these values are instantly used for the samples calculation whenever the uncertainty stream configuration matches a table entry. Additionally, whenever a value not found in the table is computed, it will be cached in the model for fast access in later calculations.
  • The Parametric Properties editor for uncertainty analysis streams will now display a warning when the user presses OK and the number of samples does not match the number required to cover the given probability and confidence levels. If the samples count is left as is and the warning is ignored, the stream can still be submitted, but the probability and confidence levels will not be written in generated reports.
  • The DAKOTA Uncertainty configuration editor now supports Undo/Redo for changes made since the window was opened. Accepting the modifications will still result in a single undoable edit in the Model Editor.
  • Figure of Merit descriptions are now broken down into a short description and a long description. The short description allows a line of text no longer than 256 characters. The long description, like the original, may be either a model note reference or a block of text. When opening older models, any explicit text descriptions shorter than 256 characters will be moved into the short description; otherwise they are used as the long description.
  • The DAKOTA Report configuration now allows plotting both figures of merit and probability distributions. At least one FOM or distribution must be selected as the dependent data, while the independent data can use either the iteration index (the default) or another FOM or distribution (FOMs can be mapped to distributions and vice versa).
  • Numerous improvements have been made to the report generated for DAKOTA Uncertainty tasks:
    • Response data is now included in the report generated in DAKOTA Uncertainty tasks. The response data is included in the variate data table if the total number of variate and response columns is five or less, otherwise it is written in distinct tables. Similarly, when the Order is set to a value greater than zero, the response data trimmed out of the results to create the results matrix will be listed after the responss data, with one table for each figure of merit. Including response data in the report is optionally enabled from the Report tab of the DAKOTA Uncertainty stream configuration.
    • The CDF graphs in the generated report, which currently draw Y axis lines for 0% and 100%, now include lines for 5%, 50%, and 95%. In addition, the X and Y axes are both labeled. - The DAKOTA-generated values for each response function have been removed from the first section and are now listed in a separate section, after the variate and response tables.
    • The generated report no longer includes the 95% confidence range values for the mean and standard deviation.
    • The response function short description is listed everywhere the response is mentioned, including response table data, results matrix iterations, etc.
    • The logic that splits tables with more than five columns into multiple tables has been improved so that subsequent tables will not have only one or two columns (6 columns are split into two tables of 3, 7 into one with 4 and one with 3, etc.).
  • When the model variable reference in a variable-to-distribution mapping can no longer be resolved (the representative component was deleted, the property is no longer enabled, etc.), stream errors now include a short description of the missing variable.
  • Uncertainty now provides a model modification context that plug-ins can use in their export logic to describe the changes made to the input. For example, the next update to the TRACE plug-in displays this data at the top of the input.
  • Resolved an issue where the generated DAKOTA input did not work with DAKOTA 5.2. The analysis_driver line defined a '<not used>' stub that DAKOTA 5.2 could not locate as a valid command, which stopped execution. This has been resolved by changing the analysis_driver argument to an empty string.
  • Resolved an issue where the user could specify a number of samples that was insufficient to correctly generate a results matrix in an uncertainty analysis stream using the order statistic method. The number of samples must now be at least one plus the number of FOM limits times the order or an error will be displayed.
  • The Uncertainty plug-in was updated to use the newly renamed CAFEAN core "Optional" editors rather than the legacy "Namelist" editors.

Version 1.2.1 - Released 6/18/12:

  • The Uncertainty plug-in now has a user's manual available from the Model Editor Help Contents. The Parametric Properties editor for DAKOTA Uncertainty type streams now includes a Help button that opens the manual.
  • When mapping a model variable to a distribution, an option is now available for selecting an existing distribution.
  • Model variables may now exclude certain distribution rules. When mapping a model variable that excludes rules to a distribution, the options for creating new distributions with the excluded rules are not shown. Similarly, existing distributions whose rules are excluded will not be shown in the Select and Copy drop-down lists.

Version 1.2.0 - Released 4/30/12:

  • The Variables tab in the Uncertainty Configuration window has been updated to support a new type of model data reference representing a range of values in one or more components. This can be used to reference a column of values in a table that will be modified by the Uncertainty plug-in. When such a value is selected, the Nominal Value column displays a description of the selected values.

Version 1.1.2 - Released 9/14/2011:

  • An error was corrected in the calculation of the FOM tolerance interval/limits results. The problem with the algorithm was that the limiting FOMs were not being saved as results after each sorting step, and before the rows containing them were eliminated. Instead, all the FOMs were taken from the limiting values in the remaining rows after the last sorting step. The limiting FOM values are now saved after each sorting step, and before their containing rows are eliminated.
  • The Wilks sample count computation has been further optimized to support a significantly greater range of inputs.
  • An issue was resolved with the AptPlot Extract step that caused the step to fail when connected to step outputs.

Version 1.1.1 - Released 8/16/2011:

  • The DAKOTA inputs generated by the Uncertainty Analysis plug-in now omit distributions that are not explicitly mapped to one or more model variables.
  • The results matrix in the report ODF document is now arranged with one response per row instead of per column. In addition, the label is now more prominent, using the level 2 heading style.
  • The Extract Data step now supports referencing files from their existing location without copying the plot file to the step folder. This functionality is only available for Extract Data steps created with this or later versions.

Version 1.1.0 - Released 7/11/2011:

  • The Uncertainty plug-in now supports Engineering Template models. When editing Uncertainty stream properties in an Engineering Template with multiple reference
    models, new probability distributions and variables allow selecting the target model.
  • Several new probability distributions have been added. They are: Loguniform, Exponential, Gamma, Gumbel, Frechet and Weibull.
  • The Uncertainty plug-in now supports editing model variables other than user-defined numerics. This functionality is supported on a per-plug-in basis.
  • The user interface for editing Uncertainty stream properties has been overhauled to support the order statistic approach. The order, probability level, and confidence level may all be specified in the General tab. In addition, response functions (now labeled as "figures of merit") no longer need to explicitly define the response levels. Instead, an upper and lower limit flag may be specified on each response. These new attributes are used to compute a results matrix, which appears in the generated ODF report for DAKOTA stream steps.
  • The response levels for a figure of merit are now computed automatically from the extracted results passed into a DAKOTA stream step.
  • The samples computation has been significantly optimized. Computations that previously took approximately 30 seconds now complete in a few milliseconds. Other sample computations that were prohibitively expensive now complete in a few seconds. The calculation has been updated to support computing the samples based on the specified Order, probability level and confidence level.
  • In support of the order statistic method, the Latin Hypercube sampling type is now disabled unless the Order value is set to 0.
  • An issue was resolved that would display DAKOTA-generated variants with the wrong titles in both the report and the parametric tasks dialog.

Version 1.0.4 - Released 6/10/2011:

  • Added support for Histogram probability distributions.

Version 1.0.3 - Released 5/13/2011:

  • Resolved an issue where very small values could not be used as probability distribution parameters.
  • Resolved an issue where entering values into probability distribution parameters when the model was set to British units caused unexpected values to appear.

Version 1.0.2 - Released 3/23/2011:

  • Exporting or submitting a Job Stream using the DAKOTA Uncertainty stream type now updates the progress bar for each set of model files exported.
  • Resolved an issue where updating the parametric properties of an Uncertainty stream did not update the parametric tasks of downstream steps.