Desktop GUI

D²MTOLab ships an experimental desktop GUI built with DearPyGui. It exposes the full experiment workflow — choose problems and algorithms, configure parameters, run, and view analysis tables and figures — without writing any code.

The GUI lives in the ui/ directory of the repository. It is not part of the PyPI package, so use a source checkout to run it.

Installation

The GUI needs the core package plus a few UI-only dependencies:

# From a source checkout of the repository
git clone https://github.com/JiangtaoShen/DDMTOLab.git
cd DDMTOLab
pip install -e .

# UI dependencies (DearPyGui, Pillow, pandas, openpyxl)
pip install -r ui/requirements.txt

# Launch
python ui/main.py

Note

The GUI requires a graphical display. On a headless machine (e.g. a remote server) run experiments through the Python API or BatchExperiment instead.

Layout

The window opens on two tabs — Test Mode and Batch Experiment — that share the same three-column layout:

Problem Selection (left) — pick a category, suite, and problem, then set problem parameters (D, M, K, Kp) where the suite allows them.
Algorithm Selection (middle) — pick an algorithm category and click algorithms to add them. Each selected algorithm becomes a collapsible card where you can rename it, edit its parameters, and reorder or remove it.
Toolbar & Results (right) — run controls and analysis settings on top, a scrollable results panel (tables and figures) below.

The algorithm category follows the problem category automatically, and only algorithms compatible with the chosen problem are offered. Incompatible combinations (for example, a single-objective algorithm on a multiobjective problem, or an equal-dimension algorithm on unequal-dimension tasks) are reported before the run starts rather than failing midway.

Test Mode

Test Mode runs each selected algorithm once on a single problem and shows the per-run analysis: a results table, convergence curves, non-dominated solution plots (for multiobjective problems), and a runtime comparison.

Workflow

Select a problem category, suite, and problem on the left.
Adjust problem parameters if needed.
Select one or more algorithms in the middle column and edit their parameters.
Choose analysis options (metric, log scale, figure format, …).
Click Run.

Toolbar

Button	Action
Run	Run the selected algorithms sequentially and generate analysis.
Stop	Interrupt the current run.
Clean Data	Clear the working `Data`/`Results` folders (optionally backing up first).
Load Data	Re-analyze a previously saved data folder without re-running.
Animation	Generate a convergence animation (GIF/MP4) from the last run’s data.

Batch Experiment

Batch Experiment runs multiple algorithms on multiple problems for a configurable number of independent runs, using parallel worker processes, then produces statistical tables (mean/median with standard deviation and rank-sum significance markers) and aggregated plots.

Workflow

Select a problem category and suite, then click problems to add them.
Select algorithms and edit their parameters.
Set Runs (independent repetitions) and Workers (parallel processes).
Configure analysis settings (metric, table format, statistic type, …).
Click Run.

Save / Load Config

The Save Config and Load Config buttons persist the full experiment setup — selected problems and algorithms, their parameters, run settings, and analysis options — to a YAML file. The saved file is compatible with from_config(), so a GUI-designed experiment can also be launched from a script.

Results

Both modes render results on a light panel: statistical tables (best values highlighted), convergence plots, non-dominated fronts, and runtime bars. Use Open Results Folder to open the output directory, or right-click a figure to copy it or reveal it in the file browser. All outputs are also written to the tests/Data and tests/Results folders next to the ui/ package.