Testing foqus

Here is where a list of test will go.

Start FOQUS

Open Anaconda Prompt and run the python program. This will start up FOQUS and start the select working directory dialog.

> python \ProgramData\Anaconda2\Scripts\foqus.py

Get FOQUS Source Code

> git clone https://$USERNAME:[email protected]/CCSI-Toolset/foqus
> cd foqus/examples/Optimization

Smoke test (short test to insure basic workings)

Session --> Open Session

Select foqus/examples/Optimization/BFBCostOpt.foqus

Session --> Add\Update Model to Turbine

Browse and select foqus/examples/Optimization/Model_Files/BFB_sinter_config_v6.2.json
Name this "BFB"

Remove the BFB Cost node from the FOQUS Flowsheet

Run

Detailed test

Sotorrio's Example

Installation

Go to https://github.com/CSSI-Toolset/foqus/releases
Under "Internal 2018 Feb release", click on "Source code (zip)"
Select "Save File" and click "OK"
After file is downloaded, "Extract All" in the directory "C:\Users\sotorrio1\foqus-1.0.0"
Go to Windows Terminal and type the following commands:

cd foqus-1.0.0

python setup.py install

Opening FOQUS

Type the following command in the Windows Terminal (make sure you are in the foqus-1.0.0 directory):

python foqus.py

Running FOQUS

FOQUS asks the user to provide the working directory
Once the working directory is set up, the FOQUS main window pops up
First thing I like to do is going to the "Settings" tab and making sure "Working Directory" and "PSUADE EXE" are set up correctly

Uncertainty Quantification (UQ) Module Testing

Click on the arrow right next to the "Session" tab and click "Open Session..."
FOQUS asks the user "Do you want to save your current session before loading another session?"
Since this is the first session we are loading and there is no current session, click "No" and the browsing window shows up
Go to "C:\Users\sotorrio1\foqus-1.0.0\examples\UQ" and select the file "Rosenbrock_no_vectors.foqus"
Go to "Uncertainty" tab
Click "Add New..."
In the "Add New Ensemble - Model Selection" window make sure "Use flowsheet" is selected and click "OK"
In the "Simulation Ensemble Setup" window make sure "Choose sampling scheme" is selected
In the "Distributions" tab change the "type" of variable "Rosenbrock.x2" to "Fixed" and a value equal to 5
Switch to the "Sampling Scheme" tab
Select "Latin Hypercube" as sampling scheme and 1,000 samples
Click "Generate Samples" button. Samples are generated
Go to "Preview Samples"
Select all inputs and click "Graph 1-D Scatter". Scatter plots are generated
Left click let the user zoom in on each individual scatter plot
Right click let the user zoom out back to all the plots
Close the figure and go back to "Preview Inputs" window
Select two inputs and click "Graph 2-D Scatter (2 Inputs Only)". Scatter plot is generated
Close the figure and go back to "Preview Inputs" window
Select all inputs and click "Graph Distribution". Distribution plots are generated
Left click let the user zoom in on each individual distribution plot
Right click let the user zoom out back to all the plots
Close the figure and go back to "Preview Inputs" window
Click "OK" to close the "Preview Inputs" window and go back to "Simulation Ensemble Setup" window
Click "Done" to close the "Simulation Ensemble Setup"
In the "Simulation Ensemble Table" click "Launch"
"FOQUS Run Finished" window pops up showing the amount of successful runs. Click "OK" to close it
Click "Delete Selected" to delete the ensemble from the "Simulation Ensemble Setup"
Click "Load from File..." button
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\UQ" and select the file "lptau20k_10inputs_4outputs.filtered"
In the "Simulation Ensemble Table" click "Analyze"
"Analysis of Ensemble" window opens

Wizard Mode

Make sure you are in "Wizard" mode
Click "Enable Parameter Screening"
Test every combination:

Choose an output to analyze (removalCO2, removalH2O, dPads)
Choose a parameter selection method (MARS Ranking, Sum of Trees, Delta Test, Gaussian Process)
Click "Compute input importance". A bar plot with parameter screening rankings is generated

In the Analysis section, make sure "Ensemble Data" is selected as the output to use for the analysis
Choose output variable to analyze
Select one input and click "Visualize". 1-input to 1-output scatter plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select a second input and click "Visualize". 2-input to 1-output scatter plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Uncertainty Analysis" and click "Analyze". PDF and CDF plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Sensitivity Analysis" and "First-order", and click "Analyze". First-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Second-order", and click "Analyze". Second-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Total-order", and click "Analyze". Total-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Make sure "Response Surface" is selected as the output to use for the analysis
Select "removalCO2" as the output to analyze
Select "Polynomial" and "Linear Regression" for the response surface method selection
Specify a 7.50% error envelope
Click Validate. Model Error Histogram and Actual vs. Predicted Data plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select one input and click "Visualize". 1-input to 1-output line plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select a second input and click "Visualize". 2-input to 1-output surface/contour plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select a third input and click "Visualize". 3-input to 1-output Isosurface plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Uncertainty Analysis" and click "Analyze". PDF and CDF plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Sensitivity Analysis" and "First-order", and click "Analyze". First-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Second-order", and click "Analyze". Second-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Total-order", and click "Analyze". Total-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Click "Infer..."
"Bayesian Inference of Ensemble" window opens
Select "removalCO2" and "removalH2O" as "observed" outputs
Make sure all the inputs are selected for display
Click "Infer". Input Prior and Posterior Probability plots are generated
Left click let the user zoom in on each individual distribution plot
Right click let the user zoom out back to all the plots
Close both figures and go back to "Bayesian Inference of Ensemble" window
Click "Close" to go back to "Analysis of Ensemble" window

Expert Mode

Switch to "Expert Mode" by clicking on the "Mode" button at the top of the "Analysis of Ensemble" window
Select "removalCO2" as output under analysis
Select "MARS Ranking" as a parameter selection method and click "Compute input importance". A bar plot with parameter screening rankings is generated
Close the figure and go back to "Analysis of Ensemble" window
Under "Ensemble Data Analysis" section, select "Uncertainty Analysis" and click "Analyze". PDF and CDF plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Correlation Analysis" and click "Analyze". Pearson and Spearman correlation coefficients plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Sensitivity Analysis" and "First-order", and click "Analyze". First-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Second-order", and click "Analyze". Second-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Total-order", and click "Analyze". Total-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select one input and click "Visualize". 1-input to 1-output scatter plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select a second input and click "Visualize". 2-input to 1-output scatter plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Under "Response Surface (RS) Based Analysis" section, select "Polynomial" and "Linear Regression"
Specify an "Error Envelope" of 10.00%
Click "Validate". Model Error Histogram and Actual vs. Predicted Data plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select one input and click "Visualize". 1-input to 1-output line plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select a second input and click "Visualize". 2-input to 1-output surface/contour plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select a third input and click "Visualize". 3-input to 1-output Isosurface plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Uncertainty Analysis" and "Aleatory Only"
Click "Analyze". PDF and CDF plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Uncertainty Analysis" and "Epistemic-Aleatory". Remember to change the "Type" to "Epistemic" of at least one of the input variables
Click "Analyze". Cumulative Distribution plots are generated
Close the figure and go back to "Analysis of Ensemble" window
Select "Sensitivity Analysis" and "First-order", and click "Analyze". First-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Second-order", and click "Analyze". Second-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window Select "Sensitivity Analysis" and "Total-order", and click "Analyze". Total-order sensitivity analysis bar plot is generated
Close the figure and go back to "Analysis of Ensemble" window
Click "Infer..."
"Bayesian Inference of Ensemble" window opens
Select "removalCO2" and "removalH2O" as "observed" outputs
Make sure all the inputs are selected for display
Click "Infer". Input Prior and Posterior Probability plots are generated
Left click let the user zoom in on each individual distribution plot
Right click let the user zoom out back to all the plots
Close both figures and go back to "Bayesian Inference of Ensemble" window
Click "Close" to go back to "Analysis of Ensemble" window

Optimization Under Uncertainty (OUU) Module Testing

Go to "OUU" tab
Select "Load Model From File" and click "Browse..."
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "ouu_optdriver.in"

Example 1

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, and variable 9-12 as Z3
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Discrete Random Variables", click "Browse..."
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x3sample.smp"
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 2

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, and variable 9-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Continuous Random Variables", select "Generate new sample for Z4"
Set "Sample Scheme" to "Latin Hypercube" and set "Sample Size" to 200
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 3

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, and variable 9-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Continuous Random Variables", select "Generate new sample for Z4"
Set "Sample Scheme" to "Latin Hypercube" and set "ample Size" to 200
Under "UQ Setup" and "Continuous Random Variables", check the "Use Response Surface" box
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 4

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, variable 9 as Z3, and variable 10-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Discrete Random Variables", click "Browse..."
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x3sample4.smp"
Under "UQ Setup" and "Continuous Random Variables", select "Generate new sample for Z4"
Set "Sample Scheme" to "Latin Hypercube" and set "Sample Size" to 100
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 5

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, variable 9 as Z3, and variable 10-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Discrete Random Variables", click "Browse..."
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x3sample4.smp"
Under "UQ Setup" and "Continuous Random Variables", select "Generate new sample for Z4"
Set "Sample Scheme" to "Latin Hypercube" and set "Sample Size" to 100
Under "UQ Setup" and "Continuous Random Variables", check the "Use Response Surface" box
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 6

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, variable 9 as Z3, and variable 10-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Discrete Random Variables", click "Browse..."
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x3sample4.smp"
Under "UQ Setup" and "Continuous Random Variables", select "Load existing sample for Z4"
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x4sample4.smp"
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Example 7

Under "Variables", set variable 1-4 as Z1, variable 5-8 as Z2, and variable 9-12 as Z4
Under "Optimization Setup" select the first objective function (default) and select "use model as optimizer" as the "Inner Solver"
Under "UQ Setup" and "Continuous Random Variables", select "Load existing sample for Z4"
Go to the directory "C:\Users\sotorrio1\foqus-1.0.0\examples\OUU\test_suite" and select the file "x4sampleLarge.smp"
Under "UQ Setup" and "Continuous Random Variables", check the "Use Response Surface" box and set "Sample Size" to 100
Go to "Launch/Progress" tab and click "Run OUU"
User can see OUU Progress

Provide feedback

Saved searches

Use saved searches to filter your results more quickly