OpenMDAO

3.43.0 verified Fri May 01 auth: no python

OpenMDAO is an open-source high-performance computing framework for multidisciplinary analysis and optimization. It enables efficient decomposition of coupled systems into components, automatic differentiation, and parallel computing. Current version 3.43.0, requires Python >=3.10. Release cadence is approximately monthly.

pip install openmdao

Common errors

error ImportError: cannot import name 'IndepVarComp' from 'openmdao.api' ↓

cause openmdao.api is deprecated; import path changed.

fix

Use 'from openmdao.core.indepvarcomp import IndepVarComp' or keep openmdao.api but this will break in future versions.

error NameError: name 'ExecComp' is not defined ↓

cause Forgot to import ExecComp; it's not a built-in.

fix

Add 'from openmdao.components.exec_comp import ExecComp' or use 'om.ExecComp' if using openmdao.api.

error ValueError: The model is already set up. Cannot change configuration after setup. ↓

cause Attempting to modify model after setup() call.

fix

Make all model changes before calling setup(). If needed, create a new Problem instance.

Warnings

breaking OpenMDAO 3.x removed support for Python 2.7 and 3.6. Upgrade code to Python >=3.10. ↓

fix Ensure Python interpreter is 3.10 or higher; update deprecated API calls.

breaking The 'openmdao.api' module is deprecated as of OpenMDAO 3.31.0. Direct imports from submodules are now preferred. ↓

fix Replace 'from openmdao.api import X' with the explicit import path (e.g., 'from openmdao.core.problem import Problem').

deprecated The 'openmdao.core.component.Component' base class has been renamed to 'ExplicitComponent' and 'ImplicitComponent'. Using the old name issues a deprecation warning and will be removed. ↓

fix Use 'from openmdao.core.explicitcomponent import ExplicitComponent' or 'from openmdao.core.implicitcomponent import ImplicitComponent' instead.

gotcha Do not modify a model's structure (add/remove subsystems) after calling setup() or run_model() unless you call final_setup() again. Doing so can cause crashes or undefined behavior. ↓

fix Either restructure before setup() or use a new Problem instance.

gotcha Solver tolerances and options must be set before calling setup() if you want them to take effect. Setting them after setup may be silently ignored. ↓

fix Set solver options on the system's solver object before prob.setup().

Install

pip install openmdao[all]

Imports

Problem
wrong
```
from openmdao.api import Problem
```
correct
```
from openmdao.core.problem import Problem
```
Old top-level import from openmdao.api still works but is deprecated; prefer explicit core import.

IndepVarComp

wrong

from openmdao.api import IndepVarComp

correct

from openmdao.core.indepvarcomp import IndepVarComp

Using openmdao.api is deprecated; use explicit import from submodule.

ExplicitComponent

wrong

from openmdao.api import ExplicitComponent

correct

from openmdao.core.explicitcomponent import ExplicitComponent

Same deprecation as above.

Quickstart

Minimal example: create a Problem, add an IndepVarComp and an ExecComp, run the model, and print output.

import openmdao.api as om

prob = om.Problem()
prob.model.add_subsystem('indep', om.IndepVarComp('x', 1.0), promotes=['x'])
prob.model.add_subsystem('comp', om.ExecComp('y=x+1'), promotes=['*'])
prob.setup()
prob.run_model()
print(prob.get_val('y'))