Tutorials#

This part presents complete, runnable examples. The scripts live in the examples folder of the repository (run them from that folder) and in the regression suite tests/test_basic.py; each tutorial shows the output they produce.

The basic problem from YAML

The basic two-discipline problem from a YAML file: running it, reading the console output, the result object and the output files.

The Sellar problem

The Sellar problem built with the Python API, and the effect of the inner solver (SLSQP, OMADS poll, OMADS MADS).

The Session API

Driving a run one outer iteration at a time, observers, early stopping, and solving the subproblems yourself with ask/tell.

Serial, parallel and custom execution

Gauss–Seidel and Jacobi sweeps, process pools, custom executors, and why parallel runs give the same answer whatever the execution order.

Extending DMDO

Analyses as Discipline objects, named blackboxes, custom subproblem solvers, executors and coordination policies, and plugins.

The HTTP service and remote workers

The HTTP service: sessions over HTTP, live events, and remote workers pinned to the machines that hold the analysis codes.

The post-processing dashboard

The post-processing dashboard: interactive plots of the coordination history and of the subproblem solves.

Note

The outputs shown in these tutorials were produced with DMDO 2612.0 on Windows with Python 3.12. Runs are reproducible for a given seed and platform; other platforms may differ in the last digits.