Note
Go to the end to download the full example code.
Outputs and reports in one place#
Produce every plot type, every export and the quality report from a single finished design.
One finished design usually has to be communicated to several audiences at once: slides want images, a lab notebook wants a readable table, a paper wants LaTeX, and a downstream script wants raw data. This example takes a single design and produces the full range of Mergen’s plots and export formats, so each audience gets the artefact that suits it. It is the one example that deliberately generates everything; every other example stays minimal.
Parameters#
factor_a, factor_b, factor_c (0.0-1.0, continuous, 15-level grid, rounded to 3 decimals): three generic normalised inputs, enough dimensions for an interesting pairplot and correlation view. The moderate grid lets the optimiser reach a high-quality design at its full default effort while still running quickly.
What to look at#
The saved plots (one PNG per type): pairplot for coverage, 1d for per-factor spread, 2d for a single pair, distances for the pairwise profile, correlation for pairwise independence, and quality for the metric percentiles. Any of these can go straight onto a slide.
quality_report()(printed): the numeric quality summary.The six export files, each for a different consumer: PNG for slides, Markdown and HTML for a lab notebook, LaTeX for a paper, CSV and JSON for downstream code.
Mergen features used#
criteria='phi_p': a maximin criterion chosen here because it gives strong, balanced percentiles across all six quality metrics, so the quality plot reads well as a showcase.result.plot('all', save=True): render every plot type at once.Every export format: to_csv, to_json, to_markdown, to_latex, to_html, to_excel.
result.quality_report()for the printed numeric summary.
Estimated runtime: a few seconds to a minute.
# sphinx_gallery_thumbnail_number = 5
from mergen import ParameterSpace, Sampler
# 1. Define a three-factor space and build one design.
space = ParameterSpace({
'factor_a': ('continuous', 0.0, 1.0, {'resolution': 15, 'round': 3}),
'factor_b': ('continuous', 0.0, 1.0, {'resolution': 15, 'round': 3}),
'factor_c': ('continuous', 0.0, 1.0, {'resolution': 15, 'round': 3}),
})
sampler = Sampler(space)
sampler.set_design(n_samples=30)
result = sampler.run(criteria='phi_p')
# 2. Printed numeric summary.
result.quality_report()
# 3. Every plot type, saved as PNG (slides).
result.plot('all', save=True)
# 4. Every export format, each for a different audience.
result.to_csv('design.csv') # downstream code
result.to_json('design.json') # downstream code
result.to_markdown('design.md') # lab notebook
result.to_latex('design.tex') # paper
result.to_html('design.html') # lab notebook / web
result.to_excel('design.xlsx') # spreadsheet users
════════════════════════════════════════════════════════════
MERGEN — Space-filling Design
════════════════════════════════════════════════════════════
Parameters : 3
Candidates : 3,375
n_samples : 30 (prescribed_in=0, focus_in=0, optimised_slots=30)
Total design : 30
Validation : 6
Criterion : phi_p
Algorithm(s) : sa
────────────────────────────────────────────────────────────
[MERGEN] Optimising (criterion=phi_p, algorithm=sa)...
[SA] Restart 1/5
[SA] Tuning temperature...
[SA] Start log(score)=2.362 T=2.0938e-02 iters=3000 swappable=30 hybrid=0.50
iter 500/3000 T=4.511e-03 best log(score)=1.451
iter 1000/3000 T=9.718e-04 best log(score)=1.335
iter 1500/3000 T=2.094e-04 best log(score)=1.285
iter 2000/3000 T=4.511e-05 best log(score)=1.277
iter 2500/3000 T=9.718e-06 best log(score)=1.260
iter 3000/3000 T=2.094e-06 best log(score)=1.252
[SA] Done log(score)=1.252 (accepted=207/3000, rate=6.9%)
[SA] Restart 1: new best log(score)=1.252
[SA] Restart 2/5
[SA] Tuning temperature...
[SA] Start log(score)=1.914 T=4.8265e+00 iters=3000 swappable=30 hybrid=0.50
iter 500/3000 T=1.040e+00 best log(score)=1.851
iter 1000/3000 T=2.240e-01 best log(score)=1.851
iter 1500/3000 T=4.827e-02 best log(score)=1.824
iter 2000/3000 T=1.040e-02 best log(score)=1.570
iter 2500/3000 T=2.240e-03 best log(score)=1.385
iter 3000/3000 T=4.827e-04 best log(score)=1.365
[SA] Done log(score)=1.365 (accepted=1584/3000, rate=52.8%)
[SA] Restart 3/5
[SA] Tuning temperature...
[SA] Start log(score)=1.835 T=2.1039e+00 iters=3000 swappable=30 hybrid=0.50
iter 500/3000 T=4.533e-01 best log(score)=1.583
iter 1000/3000 T=9.765e-02 best log(score)=1.583
iter 1500/3000 T=2.104e-02 best log(score)=1.583
iter 2000/3000 T=4.533e-03 best log(score)=1.391
iter 2500/3000 T=9.765e-04 best log(score)=1.287
iter 3000/3000 T=2.104e-04 best log(score)=1.244
[SA] Done log(score)=1.244 (accepted=1385/3000, rate=46.2%)
[SA] Restart 3: new best log(score)=1.244
[SA] Restart 4/5
[SA] Tuning temperature...
[SA] Start log(score)=2.285 T=1.6745e-01 iters=3000 swappable=30 hybrid=0.50
iter 500/3000 T=3.608e-02 best log(score)=1.759
iter 1000/3000 T=7.772e-03 best log(score)=1.494
iter 1500/3000 T=1.675e-03 best log(score)=1.373
iter 2000/3000 T=3.608e-04 best log(score)=1.281
iter 2500/3000 T=7.772e-05 best log(score)=1.268
iter 3000/3000 T=1.675e-05 best log(score)=1.256
[SA] Done log(score)=1.256 (accepted=617/3000, rate=20.6%)
[SA] Restart 5/5
[SA] Tuning temperature...
[SA] Start log(score)=2.292 T=3.4002e-01 iters=3000 swappable=30 hybrid=0.50
iter 500/3000 T=7.326e-02 best log(score)=1.799
iter 1000/3000 T=1.578e-02 best log(score)=1.601
iter 1500/3000 T=3.400e-03 best log(score)=1.380
iter 2000/3000 T=7.326e-04 best log(score)=1.296
iter 2500/3000 T=1.578e-04 best log(score)=1.289
iter 3000/3000 T=3.400e-05 best log(score)=1.264
[SA] Done log(score)=1.264 (accepted=775/3000, rate=25.8%)
[MERGEN] sa done -- score=3.471 (elapsed 13.2s)
────────────────────────────────────────────────────────────
MERGEN — Final Design
────────────────────────────────────────────────────────────
Prescribed (in) : 0
Prescribed (out) : 0
Focus (in) : 0
Focus (out) : 0
Optimised : 30
Total design : 30
Validation : 6
════════════════════════════════════════════════════════════
[METRICS] Computing MC baseline (300 designs)...
[METRICS] MC baseline complete (300 designs).
════════════════════════════════════════════════════════════════════════
MERGEN Design Metrics (n=30, d=3)
════════════════════════════════════════════════════════════════════════
Metric Value Baseline Better when Rank
────────────────────────────────────────────────────────────────────────
Min distance 0.3275 0.0720 higher 100th pct *
Minimax distance 0.3768 0.5394 lower 100th pct
Max |correlation| 0.0467 0.2321 lower 99th pct
2D projection CD2 0.1326 0.1187 lower 29th pct
CV distances 0.3324 0.3711 lower 100th pct *
Mean distance 0.8458 0.7094 higher 100th pct *
────────────────────────────────────────────────────────────────────────
Criterion scores
────────────────────────────────────────────────────────────────────────
PHI_P 3.4710 13.8943 100th pct lower
────────────────────────────────────────────────────────────────────────
* = primarily optimised by 'phi_p'
For other priorities, see: mergen.criteria.list_criteria()
────────────────────────────────────────────────────────────────────────
Baseline: 300 MC designs from feasible space | Rank = percentile among baseline designs
════════════════════════════════════════════════════════════════════════
Saved: outputs/pairplot_12.png
Saved: outputs/1d_1.png
Saved: outputs/1d_2.png
Saved: outputs/1d_3.png
Saved: outputs/2d_1.png
Saved: outputs/2d_2.png
Saved: outputs/2d_3.png
Saved: outputs/distances_2.png
Saved: outputs/correlation_1.png
Saved: outputs/quality_2.png
Saved: outputs/design.csv (36 rows)
Saved: outputs/design.json (36 rows)
Saved: outputs/design.md
Saved: outputs/design.tex
Saved: outputs/design.html
Saved: outputs/design.xlsx (36 rows)
Total running time of the script: (0 minutes 21.410 seconds)