novaphy.ik.IKOptimizerLM

Levenberg-Marquardt IK optimizer with adaptive trust-region damping.

IKOptimizerLM(
    model_or_bundle,
    n_batch,
    objectives,
    *,
    jacobian_mode=IKJacobianType.ANALYTIC,
    lambda_initial=0.1,
    lambda_factor=2.0,
    lambda_min=1e-5,
    lambda_max=1e10,
    rho_min=1e-3,
    problem_idx=None,
)

step(joint_q_in, joint_q_out, iterations=10, step_size=1.0) runs a fixed iteration count. lambda_values exposes the terminal damping for each row; costs stores the squared residual norm. reset() restores damping and invalidates the FK caches.

Most applications select this implementation through IKSolver(optimizer=IKOptimizer.LM).

See also

• IKOptimizerLBFGS
• IKSolver
• novaphy.ik