RmultConvergence¶
-
class
mybigdft.workflows.convergences.RmultConvergence(base_job, reference, delta, n_jobs=10, precision_per_atom=0.00036749322480854655)[source]¶ Bases:
mybigdft.workflows.convergences.AbstractConvergenceBigDFT uses wavelets on a grid to represent the wavefunctions and density of the system. One key parameter of a BigDFT calculation therefore is the grid extension (noted rmult). There are actually two grids centered on the atoms of the system: a coarse grid, with a shorter extension and one fine grid, with a longer extension.
This class allows to run all the necessary calculations to determine the smallest rmult which must be used so that the energy error compared to the reference calculation (with the largest grid extension considered) lie within the required precision per atom.
One must provide a base Job instance defining the system considered and possibly non-default input parameters. Then, one must give the reference value of the input parameter for which the convergence is considered. A delta value of this input parameter is then used to initialize a given number of jobs (n_jobs) of decreasing quality. Finally, a precision per atom (in Ha) is required (default to 0.3675 mHa (= 10 meV) par atom).
- Parameters
base_job (Job) – Template for all the other jobs of this workflow.
reference – Reference input parameter, giving high quality results.
delta – Variation of of the input parameter between two runs.
n_jobs (int) – Maximal number of jobs to be run.
precision_per_atom (float) – Precision per atom for a job to be considered as converged, the reference energy being that of the job using the reference input parameter (units: Ha).