newton.ik.IKOptimizerLBFGS#

class newton.ik.IKOptimizerLBFGS(model, n_batch, objectives, *a, **kw)[source]#

Bases: object

L-BFGS optimizer for batched inverse kinematics.

The optimizer maintains a limited-memory quasi-Newton approximation and chooses step sizes with a parallel strong-Wolfe line search. It supports the same Jacobian backends as IKOptimizerLM.

Parameters:

  • model (Model) – Shared articulation model.

  • n_batch (int) – Number of evaluation rows solved in parallel. This is typically n_problems * n_seeds after any sampling expansion.

  • objectives (Sequence[IKObjective]) – Ordered IK objectives applied to every batch row.

  • jacobian_mode – Jacobian backend to use.

  • history_len – Number of (s, y) correction pairs retained in the L-BFGS history.

  • h0_scale – Scalar used for the initial inverse-Hessian approximation.

  • line_search_alphas – Candidate step sizes tested in parallel during the line search.

  • wolfe_c1 – Armijo sufficient-decrease constant.

  • wolfe_c2 – Strong-Wolfe curvature constant.

  • problem_idx – Optional mapping from batch rows to base problem indices for per-problem objective data.

__init__(model, n_batch, objectives, jacobian_mode=IKJacobianType.AUTODIFF, history_len=10, h0_scale=1.0, line_search_alphas=None, wolfe_c1=1e-4, wolfe_c2=0.9, *, problem_idx=None)#
compute_costs(joint_q)#

Evaluate squared residual costs for a batch of joint configurations.

Parameters:

joint_q (array(ndim=2, dtype=float32)) – Joint coordinates to evaluate, shape [n_batch, joint_coord_count].

Returns:

Costs for each batch row, shape [n_batch].

Return type:

array(ndim=1, dtype=float32)

reset()#

Clear L-BFGS history and cached line-search state.

step(joint_q_in, joint_q_out, iterations=50)#

Run several L-BFGS iterations on a batch of joint configurations.

Parameters:

  • joint_q_in (array(ndim=2, dtype=float32)) – Input joint coordinates, shape [n_batch, joint_coord_count].

  • joint_q_out (array(ndim=2, dtype=float32)) – Output buffer for the optimized coordinates, shape [n_batch, joint_coord_count]. It may alias joint_q_in for in-place updates.

  • iterations (int) – Number of L-BFGS iterations to execute.