Coverage for /builds/kinetik161/ase/ase/optimize/precon/__init__.py: 100.00%

1""" 

2This module contains tools for preconditioned geometry optimisation. 

3 

4Code maintained by James Kermode <james.kermode@gmail.com> 

5Parts written by John Woolley, Letif Mones and Christoph Ortner. 

6 

7The preconditioned LBFGS optimizer implemented here is described in 

8the following publication: 

9 

10 D. Packwood, J. R. Kermode, L. Mones, N. Bernstein, J. Woolley, 

11 N. Gould, C. Ortner, and G. Csanyi, A universal preconditioner for 

12 simulating condensed phase materials, J. Chem. Phys. 144, 164109 (2016). 

13 DOI: https://doi.org/10.1063/1.4947024 

14 

15A preconditioned version of FIRE is also included, this is less well tested. 

16 

17Optional dependencies 

18--------------------- 

19 

20 - scipy, `pip install scipy` for efficient sparse linear algebra, 

21 important for large systems (>1000 atoms). 

22 - PyAMG, `pip install pyamg`, for iterative adaptive multi grid 

23 inversion of the preconditioner, again important for large systems. 

24""" 

25 

26from ase.optimize.ode import ODE12r 

27from ase.optimize.precon.fire import PreconFIRE 

28from ase.optimize.precon.lbfgs import PreconLBFGS 

29from ase.optimize.precon.precon import (C1, FF, Exp, Exp_FF, Pfrommer, Precon, 

30 PreconImages, SplineFit, make_precon) 

31 

32 

33class PreconODE12r(ODE12r): 

34 """ 

35 Subclass of ase.optimize.ode.ODE12r with 'Exp' preconditioning on by default 

36 """ 

37 

38 def __init__(self, *args, **kwargs): 

39 if 'precon' not in kwargs: 

40 kwargs['precon'] = 'Exp' 

41 ODE12r.__init__(self, *args, **kwargs) 

42 

43 

44__all__ = ['make_precon', 'PreconImages', 'SplineFit', 

45 'Precon', 'Exp', 'C1', 'Pfrommer', 'FF', 'Exp_FF', 

46 'PreconLBFGS', 'PreconFIRE', 'PreconODE12r']