next up previous contents index
Next: molecular_dynamics() optimize Up: The optimizers module: optimization Previous: conjugate_gradients() optimize   Contents   Index

quasi_newton() -- optimize atoms with quasi-Newton minimization

quasi_newton(output='NO_REPORT', min_atom_shift=0.01, max_atom_shift=100.0, residue_span_range=(0, 99999), **vars)
Output:
molpdf

Requirements:
restraints

This functions in a very similar way to conjugate_gradients(), but uses a variable metric (quasi-Newton) method instead to find the minimum. The algorithm implemented in MODELLER is the BFGS or Broyden-Fletcher-Goldfarb-Shanno method [Press et al., 1992]. It takes the same keyword arguments as conjugate_gradients(), plus one additional max_atom_shift argument. This is used to limit the maximum size of an optimization move.



Ben Webb 2007-08-03