Alternative optimizers

The nullspace_optimizer package provides an interface with several optimizers which are available in the nullspace_optimizer/optimizers folder.

This interface allows to run these optimizers on the Optimizable structure, which is useful to compare the results with those of the function nlspace_solve().

Method of Moving Asymptotes

The function mma_solve() solves an optimization problem using the implementation of the Method of Moving Asymptotes available on this repository.

python

from nullspace_optimizer.optimizers.MMA import mma_solve

mma_solve(problem: Optimizable, l, u, params=None, results=None)

See examples/topopt_examples/ex03_compliance_MMA.py for an example of use.

Note

The decorator bound_constraints_optimizable() should not be used when solving an optimization problem with mma_solve(). Bound constraints need to be provided with the l and u arguments.

IPOPT

The function ipopt_solve() solves an optimization problem using the optimizer IPOPT called by the Python interface cyipopt, which both need to be installed before running this function.

python

from nullspace_optimizer.optimizers.IPOPT import ipopt_solve

See examples/topopt_examples/ex04_compliance_IPOPT.py for an example of use.

Note

The decorator bound_constraints_optimizable() should not be used when solving an optimization problem with ipopt_solve(). Bound constraints need to be provided with the l and u arguments.

Optimality Criteria

The function oc_solve() solves an optimization problem using the Optimality Criteria method, based on its implementation in the topology optimization code available here. It can be used for solving optimization problems featuring a single equality constraint and no inequality constraints apart from bound constraints.

python

from nullspace_optimizer.optimizers.OC import oc_solve

oc_solve(problem: Optimizable, l, u, params=None, results=None)

An Optimality Criteria solver for volume constrained compliance minimization in topology optimization based on the script topopt_cholmod.py.

This solver applies to Optimization problems of the form

\[\begin{aligned} \min_{x\in \R^n}& \quad J(x)\\ \textrm{s.t.} & \left\{\begin{aligned} g(x)&=0,\\ l &\leqslant x \leqslant u\\ \end{aligned}\right. \end{aligned}\]

where \(x\) is a numpy array and \(g\) a single equality constraint.

See examples/topopt_examples/ex02_compliance_oc.py for an example of use.

Note

The decorator bound_constraints_optimizable() should not be used when solving an optimization problem with oc_solve(). Bound constraints need to be provided with the l and u arguments.