Worked domain examples#

The repository’s examples/ directory contains complete, runnable scripts that apply Mergen end to end in several fields. Each is self-contained and produces a design, a quality report, and plots. They are the fastest way to see the whole workflow on a realistic problem close to your own.

Wet-lab biology#

An enzyme-assay design over pH, incubation temperature, buffer type, and substrate concentration. Buffer type is a nominal factor, so the design is scored with \(\mathrm{MaxPro_{QQ}}\); a held-out validation set is reserved for checking the fitted response on unseen conditions. See examples/11_wetlab_biology.py.

CFD engineering#

A computational-fluid-dynamics parameter study on a purely numeric, continuous space, illustrating a projection-based criterion where any subset of the inputs may drive the response. See examples/12_cfd_engineering.py.

Machine-learning hyperparameters#

A hyperparameter design over learning rate, batch size, optimiser, and weight decay. The learning rate is placed on a logarithmic ladder of explicit levels and the optimiser is a nominal factor, so a QQ-type criterion is used. See examples/13_ml_hyperparameters.py.

High-energy physics phenomenology#

A parameter scan of a beyond-the-Standard-Model two-Higgs-doublet model, the kind of high-dimensional, constrained scan for which space-filling coverage is essential. See examples/14_bsm_2hdm_phenomenology.py.

Accelerator beamline optics#

A beamline-optics tuning design, showing Mergen on an instrument-configuration problem. See examples/15_beamline_optics.py.

The full set#

Script	Focus
`01_quickstart.py`	The five-statement workflow, end to end
`02_parameter_types.py`	All five factor types in one space
`03_advanced_design.py`	Prescribed points, focus region, exclusion zone
`04_choosing_criteria.py`	`compare()` across criteria, saving the winner
`05_choosing_algorithm.py`	One criterion, three optimisers, score vs time
`06_sample_size.py`	Budget choices around the 10 x d guideline
`07_extra_sets.py`	External labelled sets and validation hold-outs
`08_resume_existing.py`	`load_design`: reusing a previous campaign
`09_sequential_workflow.py`	extend, run_order, subsample, k_fold_split
`10_outputs_and_reports.py`	Every export format and the quality report
`11_wetlab_biology.py`	Enzyme assay with a nominal buffer factor
`12_cfd_engineering.py`	Continuous CFD parameter study
`13_ml_hyperparameters.py`	Log-ladder learning rate, nominal optimiser
`14_bsm_2hdm_phenomenology.py`	Constrained high-dimensional physics scan
`15_beamline_optics.py`	Instrument-configuration tuning

Running them#

Each script runs directly:

python examples/11_wetlab_biology.py

The lower-numbered scripts (01 through 10) build up the same features tutorial by tutorial, from the quickstart to full export and reporting.