qlat_scripts.v1.params — Ensemble Simulation Parameters¶
Source: qlat/qlat_scripts/v1/params.py
Note: Update this document when updating the source file.
Outline¶
Overview¶
This module defines simulation parameters for each ensemble in dict_params. It extends the base parameters defined in rbc_ukqcd_params with values specific to the data production pipeline: trajectory lists, selection rates, AMA probabilities, smearing coefficients, and meson observable configurations.
Parameters are set via dict_params[ensemble][tag] (direct assignment) or set_param(ensemble, tag, value) (function call). All parameters are accessed at runtime via get_param(job_tag, tag).
Supported ensembles include: test-4nt8, test-4nt16, test-8nt16, 24D, 32D, 32Dfine, 24DH, 48I, 64I, 64I-pq, 16IH2, 32IfineH, 32IcoarseH1, 24IH1, 24IH2, 24IH3, 32IH1, 32IH2, 32IH3.
Trajectory Lists¶
Tag: traj_list
Each ensemble specifies a list of trajectory numbers available for analysis. These are Python range objects converted to lists.
Ensemble |
Trajectories |
|---|---|
|
1000, 1040, …, 1360 |
|
2300, 2299, …, 501 |
|
500, 510, …, 5990 |
|
5100, 5090, …, 1010 |
|
500, 510, …, 2990 |
Field Selection Parameters¶
field_selection_fsel_rate¶
Fraction of spatial sites selected per time slice for field selection.
Ensemble |
Rate |
|---|---|
|
1/16 |
|
1/16 |
|
1/32 |
field_selection_psel_rate¶
Fraction of total lattice volume for point selection. Used by get_n_points_psel.
field_selection_fsel_psrc_prop_norm_threshold¶
Threshold for adaptive point-source field selection in gen_data.py. Sites with propagator norm above this value are probabilistically added to the field selection.
Ensemble |
Threshold |
|---|---|
|
1e-3 |
|
1e-4 |
|
4e-5 |
|
2e-5 |
n_points_psel¶
Number of points in the point selection (alternative to field_selection_psel_rate).
Wall-Source AMA Parameters¶
prob_exact_wsrc¶
Probability that a given time slice receives an exact-accuracy wall-source inversion. For ensembles using n_exact_wsrc instead, the probability is computed as 1 - (1 - 1/T)^n.
Ensemble |
Probability |
|---|---|
|
1/8 |
|
1/48 |
|
1/64 |
|
1/32 |
Smeared Selection Parameters¶
n_per_tslice_smear¶
Number of points per time slice in the smeared point selection.
prob_acc_1_smear / prob_acc_2_smear¶
AMA probabilities for smeared-source medium and exact accuracy inversions.
Point-Source AMA Parameters¶
prob_acc_1_psrc / prob_acc_2_psrc¶
AMA probabilities for point-source medium and exact accuracy inversions.
Ensemble |
|
|
|---|---|---|
|
1/4 |
1/16 |
|
1/32 |
1/128 |
|
1/32 |
1/128 |
Random U(1) Parameters¶
n_rand_u1_fsel¶
Number of random U(1) volume sources per trajectory.
Ensemble |
Count |
|---|---|
|
4 |
|
16 |
Most others |
64 |
prob_acc_1_rand_u1 / prob_acc_2_rand_u1¶
AMA probabilities for random U(1) medium and exact accuracy inversions.
Propagator Smearing Parameters¶
prop_smear_coef¶
APE smearing coefficient for propagator source smearing.
Ensemble |
Coefficient |
|---|---|
Most |
0.9375 |
prop_smear_step¶
Number of APE smearing steps for propagator sources.
Ensemble |
Steps |
|---|---|
|
10 |
|
29 |
|
54 |
|
96 |
Gauge Smearing Parameters¶
gf_ape_smear_coef / gf_ape_smear_step¶
APE smearing parameters for gauge field smearing (used by run_gf_ape). Coefficient 0.5, 30 steps for all ensembles.
Fermion Parameters¶
fermion_params¶
Nested dictionary of fermion action parameters indexed by [inv_type][inv_acc]. The Ls (domain-wall extent) parameter can be overridden per ensemble.
Meson Observables¶
meson_tensor_tsep¶
Source-sink separation for meson tensor measurements.
Ensemble |
Separation |
|---|---|
|
1 |
|
8 |
|
12 |
|
18 |
meson_jwjj_threshold¶
Threshold for JWJJ (current-current) operator selection.
meson_tsep_list¶
List of source-sink separations for meson correlator measurements.
Examples¶
import qlat as q
q.begin_with_mpi([[1, 1, 1, 4]])
import qlat_scripts.v1 as qs
# Access parameters
job_tag = "test-4nt8"
total_site = qs.get_param(job_tag, "total_site")
print(f"Lattice size: {total_site}")
traj_list = qs.get_param(job_tag, "traj_list")
print(f"Trajectories: {traj_list[:5]}...")
fsel_rate = qs.get_param(job_tag, "field_selection_fsel_rate")
print(f"Field selection rate: {fsel_rate}")
q.end_with_mpi()