idaes-sdoe

idaes-sdoe is a Python package for design of experiments in process systems engineering.

idaes-sdoe is part of the IDAES integrated software platform, specifically the Institute for the Design of Advanced Energy Systems Process Systems Engineering Framework (IDAES PSE Framework).

Install

Recommended setup: Conda environment

conda create -n idaes-sdoe python=3.11
conda activate idaes-sdoe
pip install -e .

Direct install into an existing environment:

pip install -e .

Optional extras:

pip install -e ".[docs]"
pip install -e ".[mars]"
pip install -e ".[notebook]"
pip install -e ".[dev]"

Use docs for local documentation builds, mars if you need the method="mars" response-surface option in imputation, notebook if you want the notebook stack managed with the project, and dev for the test stack. The docs extra installs Sphinx plus the HTML theme and code-block helpers used by the site.

To use notebooks in the recommended Conda environment:

conda activate idaes-sdoe
pip install -e ".[notebook]"
jupyter notebook

To build the docs locally:

conda activate idaes-sdoe
pip install -e ".[docs]"
python -m sphinx -b html docs docs/_build/html
open docs/_build/html/index.html

On systems without open, open docs/_build/html/index.html in a browser directly.

Quick start

The example below uses a bundled candidate set from the repository.

from pathlib import Path

from idaes_sdoe import ColumnRoles, load_csv, prepare_design_setup
from idaes_sdoe.design import design_uniform_batch

candidate = load_csv(Path("examples/supporting_data/SDOE_Ex1_Candidates.csv"))

setup = prepare_design_setup(
    candidate=candidate,
    roles=ColumnRoles(inputs=["X1", "X2"]),
)

results = design_uniform_batch(
    setup=setup,
    design_sizes=[8, 9, 10],
    num_restarts=1000,
    mode="minimax",
    random_state=7,
)

design8 = results[0]
print(design8.criterion_value)
print(design8.design.head())

Usage

The package is designed for direct use from Python modules, notebooks, and interactive sessions. Typical workflow:

• load candidate data into pandas tables
• define ColumnRoles
• call prepare_design_setup()
• run a design method from idaes_sdoe.design
• inspect the returned result
• optionally apply plotting, candidate generation, imputation, or run ordering

The main public surface is split across idaes_sdoe, idaes_sdoe.design, idaes_sdoe.ordering, and idaes_sdoe.extras.

Layout

• src/idaes_sdoe/design: core design algorithms
• src/idaes_sdoe/extras: candidate-generation and imputation helpers
• src/idaes_sdoe/plotting.py: Plotly plotting helpers
• tests: standalone test suite

License

See LICENSE.md and COPYRIGHT.md.

Contributing

By contributing to this repository, you are agreeing to all the terms set out in the LICENSE.md and COPYRIGHT.md files in this directory.

Contact

For questions about idaes-sdoe, contact Xiangyu Bi at xbi@lbl.gov.