This page details the new features that were added and bug fixes that were addressed with each minor release. To download the tools please visit the OpenMETA page.
We added a number of general improvements with this release:
- Expanded Design Variable “Range” attribute to support real-number intervals.
- Improved the documentation for the Value Aggregator (formally named Custom Processor) that makes it easy to extract values from properties and parameters in the model. Additionally we added six new “operators;” see the Value Aggregator Documentation for more details.
- Added new “CSV File” DOE Type to Parameter Study Drivers that allows for reading cases from pre-generated CSV file.
- Added a new HTML “Report” page to Test Bench and PET runs that makes it easier to navigate to logs when debugging errors in an execution.
Added three (3) additional options for launching results from the Results Browser: “Export as CSV,” “Launch in Excel,” and “Launch in JMP.” See the Results Browser Documentation for more information.
Added support for generating and populating a template schematic file in CyPhy2Schematic to allow SPICE Test Benches in a PET context to update the generated SPICE System Model according to changes in Test Bench Parameters
Added a Surrogate Modeling tab to the Visualizer that allows the generation, exploration, and training of surrogate models from results computed in a Parametric Exploration analysis.
Fixed issue that caused Autodesk EAGLE 8 to fail to launch when a Place and Route test bench was executed
PET Hierarchy was added to OpenMETA that includes the ability to nest PET Analysis Workflows that include either an Optimizer Driver or no driver at all. The concepts of Problem Inputs and Problem Outputs were added to allow for connectivity to parent PETs.
Additionally, we completed functionality of the PCC driver in a Parametric Exploration which can be used to generate a Probabilistic Certificate of Correctness (PCC). Documentation on the new PET changes can be found online and along with the installed tools.
A “Design Configuration Tree” section was added to the Visualizer that presents the Design Container structure from the OpenMETA model that allows for filtering on the choices in the configuration tree.
You can now save analysis artifacts, such as images, from individual Parametric Exploration (PET) experiments, and retrieve them later using the Visualizer.
- Results Manager and Visualizer improvements
- Usability and documentation improvements for Uncertainty Quantification dashboard
- Manage multiple dashboard "sessions" for a single set of PET result data
- Single Plot viewer now supports contour and regression overlays
- Use OpenMETA with Linux and HPC environments
- Launch editor for Python components from directly inside a PET
- Files may now be passed between PET blocks, including Test Benches
In this release the Visualizer was updated to run as a framework. We also added domain support for Simulink.
The following features were added to the Visualizer:
- Ability to configure 'user-defined' tabs by writing an R script that specifies the front-end UI and the back-end functionality.
- Ability to specify ranges exactly by double-clicking on a slider in the Filters section.
Additionally, the ability to easily configure continuous integration using an OpenMETA project was added. This will allow a specific set of test benches to be executed upon new commits being pushed to a VCS repository that is tracking an OpenMETA project.
Finally, documentation was shipped with the tools that allows for up-to-date information about all new features. It can be found using the "Meta Documentation" shortcut that is installed or navigating to:
:\Program Files (x86)\META\Documentation\HtmlDoc\index.html
This release adds physical units to the data generated from a PET within OpenMETA. These units can be assigned in the OpenMETA PET driver and appear after in parentheses after variable names in the Visualizer.
Additionally, we completed functionality of the PCC driver in a Parametric Exploration which can be used to generate a Probabalistic Certificate of Correctness (PCC).
This release improves the workflow for iterating upon a completed PET by providing a UI to adjust the configuration and a Run button to trigger the execution of a new PET from within the Visualizer without ever opening GME. The created jobs still go directly to the Results Browser as if they had been launched from within GME.
An "Uncertainty Quantification" tab was added that provides the ability to apply a normal weighting distribution (defined by mean and standard deviation) to the input variables and calculate a probability distribution function (PDF) for each of the output variables. Furthermore, constraints can be applied and a full Forward Uncertainty Quantification analysis can be executed all within the Visualizer. Finally, if multiple discrete configurations are present in the dataset, you can also rank them based on the PDFs that result from the weighting applied to each discrete configuration.
This release added the following new features to the Visualizer tool:
- A "Bayesian" tab was added that allows for Bayesian analysis techniques to be used on a certain design configuration in data.
- Combining of the "Ranking" and "Data Table" tabs for a cleaner UI.
- Numbering of enumerations in the filter section to better correlate them to the "Pairs Plot" graph panel.
This version also introduced a new workflow for running and refining PETs. When viewing data in the Visualizer, the new "Ranges" tab allows for the user to see what Design Variable ranges were used in the PET to generate the current dataset and allows the user to make adjustments to those ranges and export them back to OpenMETA a subsequent execution of the PET.
This is the initial public release of the OpenMETA tools.