Skip to content

Recorded study report

Source SHA-256: 74836746178d941efcbd03272ae420deb805db64cf41a24612addf40703bbdd3

All observations come from verified persisted runs. This report does not rerun optimization.

Problem Algorithm Budget Repeats HV median [Q1, Q3] ↑ IGD+ median [Q1, Q3] ↓
zdt1 moead 200 10 8.976699 [8.530672, 9.343050] 1.569691 [1.113530, 1.714475]
zdt1 moead 800 10 9.893841 [9.616116, 10.558809] 0.827681 [0.417799, 0.956596]
zdt1 moead 2000 10 11.232371 [11.000206, 11.519566] 0.189720 [0.062229, 0.274339]
zdt1 nsgaii 200 10 9.768761 [9.673728, 9.915476] 1.118603 [0.896094, 1.320738]
zdt1 nsgaii 800 10 11.320392 [11.211871, 11.383002] 0.268171 [0.222940, 0.396848]
zdt1 nsgaii 2000 10 11.707686 [11.692251, 11.718976] 0.030541 [0.025193, 0.044977]
zdt2 moead 200 10 8.163663 [8.030817, 8.650843] 2.720717 [2.318674, 2.756765]
zdt2 moead 800 10 9.951560 [9.792049, 10.171434] 1.239909 [1.029809, 1.305593]
zdt2 moead 2000 10 10.988770 [10.932504, 11.087343] 0.344353 [0.215804, 0.394602]
zdt2 nsgaii 200 10 9.165179 [8.981931, 9.233088] 1.829512 [1.687968, 1.963302]
zdt2 nsgaii 800 10 10.509070 [10.486238, 10.640125] 0.710542 [0.596607, 0.780407]
zdt2 nsgaii 2000 10 11.039830 [11.008922, 11.247592] 0.298105 [0.117587, 0.326099]

What these observations show

At 2000 evaluations on zdt1, moead: median HV 11.232371, median IGD+ 0.189720 (10 repeats); nsgaii: median HV 11.707686, median IGD+ 0.030541 (10 repeats).

At 2000 evaluations on zdt2, moead: median HV 10.988770, median IGD+ 0.344353 (10 repeats); nsgaii: median HV 11.039830, median IGD+ 0.298105 (10 repeats).

Read higher HV and lower IGD+ as better under the declared references. Compare the spreads in the table and plots, then limit any preference to these problems, settings, budgets and observed repeats; it is not evidence of general superiority.

Budget comparison

Final distributions

Interpretation limits

These are per-problem descriptive observations, not a general algorithm ranking. Quartiles describe variability; they are not confidence intervals. Shared seed labels do not create paired samples. The budget comparisons use fresh runs and charge every run in full; no continuous convergence history was recorded. Only two problems are included. No p-value, significance claim, or algorithm-wide superiority is reported. Do not average raw HV across different problems. See metrics.csv for each task, selected attempt, run ID and manifest hash.

Download the observations · Task metrics · Grouped summaries

Generated from the recorded protocol in Study setup. Follow T08 to rebuild this report without rerunning optimization.