oqp.library.oqp_neb

OQP nudged elastic band (NEB) for OpenQP.

Backend-agnostic (NumPy only) chain-of-states reaction-path optimizer.  The band
of images is optimized in Cartesian coordinates with:

* the improved tangent estimate (Henkelman & Jonsson, J. Chem. Phys. 113, 9978
  (2000));
* a parallel spring force on the segment-length difference and the perpendicular
  component of the true force;
* an optional climbing image (Henkelman, Uberuaga, Jonsson, J. Chem. Phys. 113,
  9901 (2000)) that drives the highest-energy image to the saddle; and
* the FIRE optimizer (Bitzek et al., Phys. Rev. Lett. 97, 170201 (2006)), which
  is robust for the non-conservative NEB force where a quasi-Newton RFO step is
  not appropriate.

The endpoints (reactant/product) are held fixed.  The caller supplies an
``energy_gradient(x_flat) -> (E, g_flat)`` callback (atomic units), so this
module is testable on analytic potentials without the compiled OQP core.

Classes

NEB

Module Contents

class NEB(images, k_spring=0.05, climbing=False, climb_fmax=0.05, logger=None)

images

nimage

k

climbing = False

climb_fmax

logger = None

energies

gradients

run(energy_gradient, fmax_tol=0.002, maxiter=200, dt=0.5, dt_max=1.0, maxmove=0.2, on_iteration=None, frms_tol=None)