forte2.optimize.geometry_optimizer

Module Contents

class forte2.optimize.geometry_optimizer.GeometryOptimizer

Cartesian geometry optimizer using L-BFGS and method analytic gradients.

Parameters:

method_factoryCallable, optional

Callable taking a System and returning a method object that provides run(), E, and gradient(). This is retained as a valid alternative. The preferred API call is GeometryOptimizer(...)(method).run().

maxiterint, optional, default=50

Maximum L-BFGS iterations.

g_tolfloat, optional, default=1.0e-4

L-BFGS gradient convergence threshold.

max_stepfloat, optional, default=1.0

Maximum line-search step length.

project_orbitalsbool, optional, default=True

If True, project the occupied orbitals from the previous evaluated geometry into the AO basis of the next geometry and use them as the SCF initial guess.

lbfgs_kwargsdict, optional

Additional keyword arguments forwarded to LBFGS.

method_factory: Callable | None = None

maxiter: int = 50

g_tol: float = 0.0001

max_step: float = 1.0

project_orbitals: bool = True

lbfgs_kwargs: dict

executed: bool = False

converged: bool = False

iter: int = 0

E: float | None = None

gradient: numpy.ndarray | None = None

history: list[tuple] = []

coordinates: numpy.ndarray | None = None

system: forte2.system.System | None = None

method: object | None = None

parent_method: object | None = None

run(system=None)

Optimize a molecular geometry.

Parameters:

systemSystem, optional

Initial molecular system. Required only when method_factory is used directly instead of the upstream-method API.

Returns:

GeometryOptimizer

The executed optimizer object.