class VMCopt_workflow

This class manages the workflow of VMC optimization.

Module usage

from turboworkflows.workflow_vmcopt import VMCopt_workflow

Constructor arguments

job parameters

argument	type	default value	description
server_machine_name	str	localhost	name of the server machine for job submission.
queue_label	str	(None)	queue label for job submission.
mpi	bool	False	use MPI parallelization.
version	str	stable	version of packages to use.
sleep_time	int	1800	interval of job status check in seconds.

vmcopt parameters

argument	type	default value	description
vmcopt_max_continuation	int	2	maximum number of continuation runs.
vmcopt_target_error_bar	float	1.0e-3	target error bar in Ha.
vmcopt_trial_optsteps	float	50	number of trial optimization steps.
vmcopt_trial_steps	float	50	number of trial VMC steps per optimization step.
vmcopt_minimum_blocks	int	3	minimum number of blocks required.
vmcopt_production_optsteps	int	2000	number of production optimization steps.
vmcopt_optwarmupsteps_ratio	int	0.8	ratio of warmup steps to discard.
vmcopt_bin_block	int	1	binning length.
vmcopt_warmupblocks	int	0	number of warmup blocks to discard.
vmcopt_optimizer	str	lr	optimizer type. options: sr for stochastic reconfiguration, lr for linear method.
vmcopt_learning_rate	float	0.35	optimization step size. default values is 0.05 for sr, 0.35 for lr.
vmcopt_regularization	float	0.001	regularization parameter
vmcopt_num_opt_param	int	0	number of optimized parameters. 0 means all the parameters are optimized.
vmcopt_onebody	bool	True	flag to optimize onebody Jastrow
vmcopt_twobody	bool	True	flag to optimize twobody Jastrow
vmcopt_det_mat	bool	False	flag to optimize matrix elements in the determinant part
vmcopt_jas_mat	bool	True	flag to optimize matrix elements in the Jastrow part
vmcopt_det_basis_exp	bool	False	flag to optimize exponents of the determinant basis sets
vmcopt_jas_basis_exp	bool	False	flag to optimize exponents of the Jastrow basis sets
vmcopt_det_basis_coeff	bool	False	flag to optimize coefficients of the determinant basis sets
vmcopt_jas_basis_coeff	bool	False	flag to optimize coefficients of the Jastrow basis sets
vmcopt_num_walkers	int	-1	number of walkers. -1 (default) = the number of MPI processes
vmcopt_twist_average	bool	False	flag for twist average
vmcopt_kpoints	list	[]	Monkhorst-Pack k-grids, [kx,ky,kz,nx,ny,nz], kx,ky,kz for grids, nx,ny,nz for shift=0, noshift=1.
vmcopt_maxtime	int	172000	maximum time (sec.)

Description

In the VMCopt_workflow class, the VMC optimization workflow is executed asynchronously (via async_launch()). The workflow runs VMC optimization in a continuation loop (up to vmcopt_max_continuation runs). For each continuation index \(i_{\mathrm{cont}}\):

test run (icont = 0)

Uses fixed trial parameters (vmcopt_trial_optsteps, vmcopt_trial_steps) to produce energy and error-bar data.
No step estimation is performed.
Results are used by the next run to estimate required VMC steps.

production run (icont ≥ 1)

Loads the previous run's results from the corresponding pkl file.
Estimates the VMC steps required to reach vmcopt_target_error_bar via the relation

\[N_{\mathrm{steps}} \propto \left( \frac{\sigma}{E_{\mathrm{target}}} \right)^2\]

where \(\sigma\) is the error bar from the previous run and \(E_{\mathrm{target}}\) is the target error bar.
Applies a minimum-step floor using vmcopt_minimum_blocks and vmcopt_warmupblocks.
Estimates total runtime from the previous run's estimated_time_for_1_generation and logs it.
Builds a VMCopt_genius instance with the estimated steps and fixed vmcopt_production_optsteps, then generates input and submits the job.

For each run, the method:

Waits for job submission and completion using asyncio.sleep().
Fetches output files (e.g. fort.10, fort.11, out_min_*, parminimized.d, forces.dat).
Stores and plots results; for \(i_{\mathrm{cont}} > 0\), computes parameter averages discarding an initial fraction (vmcopt_optwarmupsteps_ratio).
Persists state in pkl files under the pkl directory.

If all continuation pkl files already exist and vmcopt_rerun is False, the calculation is skipped. On success, the method returns (status, list of output file paths under the root directory, and an empty output-values dict).