Deborah.Miriam.TOMLConfigMiriam

struct BootstrapConfig

Defines bootstrap resampling parameters (block bootstrap).

Fields

• ranseed::Int : Random seed for reproducibility.
• N_bs::Int : Number of bootstrap replicates to generate (N_bs > 0).
• blk_size::Int : Block length (blk_size $\ge 1$). Use 1 for i.i.d. bootstrap.
• method::String : Block-bootstrap scheme to use. Accepted values:
  • "nonoverlapping" : Nonoverlapping Block Bootstrap (NBB). Partition the series into disjoint blocks of length blk_size, then resample those blocks with replacement to reconstruct a series of approximately the original length (last block may be truncated).
  • "moving" : Moving Block Bootstrap (MBB). Candidate blocks are all contiguous length-blk_size windows; resample these with replacement.
  • "circular" : Circular Block Bootstrap (CBB). Like MBB, but windows wrap around the end of the series (circular indexing).

Notes

• Only the three literal strings above are recognized; other values should raise an error.
• Resampled series length should match the original; if it overshoots, truncate the final block.
• Choose blk_size based on dependence strength (larger for stronger autocorrelation).

Example (TOML)

[bootstrap]
ranseed  = 850528
N_bs     = 1000
blk_size = 500
method   = "moving"  # one of: "nonoverlapping", "moving", "circular"

struct FullConfigMiriam

Top-level configuration struct combining all TOML sections.

• data: See TraceDataConfig
• input_meta: See InputMetaConfig
• solver: See SolverConfig
• jk: See JackknifeConfig
• bs: See BootstrapConfig
• traj: See TrajectoryConfig
• abbrev: See StringTranscoder.AbbreviationConfig

struct InputMetaConfig

Configuration for the [input_meta] section.

Defines the lattice geometry and physics parameters for the ensemble setup.

• ns, nt: Lattice size ($V = N_S^3 \times N_T$)
• nf: Number of quark flavors $N_{\text{f}}$
• beta: Gauge coupling $\beta$
• csw: Clover coefficient $c_{\text{SW}}$
• kappa_list: List of $\kappa$ values used in the simulation

struct JackknifeConfig

Configuration for the [jackknife] section.

• bin_size: Number of configurations per jackknife bin

struct SolverConfig

Configuration for the [solver] section.

Parameters that control the trace estimation solver behavior.

• maxiter: Maximum number of iterations allowed
• eps: Convergence tolerance

struct TraceDataConfig

Configuration for the [data] section.

Specifies the ensemble trace file layout, trace data inputs, and modeling targets.

• location: Base directory path for data storage
• multi_ensemble: Identifier for the current ensemble group
• ensembles: List of ensemble names
• analysis_header: Base name for generated output
• TrM1_X, TrM2_X, ...: feature components for each trace
• TrM1_Y, TrM2_Y, ...: Target trace (scalar) for each data
• TrM1_model, ...: Learning model tag for each trace
• LBP: Labeled set percentage (as integer, e.g., 5 means $5\%$ of data used for labeled set among the total data set)
• TRP: Training set percentage (e.g., 95 means $95\%$ of data used for training set among the labeled set)
• dump_original: Whether to save original prediction data
• use_abbreviation: If true, uses abbreviation codes for filenames

struct TrajectoryConfig

Configuration for the [trajectory] section.

• nkappaT: Number of $\kappa$ values scanned in reweighting

parse_full_config_Miriam(
    toml_path::String, 
    jobid::Union{Nothing, String}=nothing
) -> FullConfigMiriam

Parses the full configuration dictionary into a FullConfigMiriam object.

This function is typically called after loading a .toml file using TOML.parsefile, and it constructs strongly-typed configuration sections used throughout the simulation pipeline.

Arguments

• toml_path::String: Path to the TOML configuration file.
• jobid::Union{Nothing, String}: Optional job ID for contextual logging.

Returns

• FullConfigMiriam: Struct with all configuration sections, used as the central config object.

Notes

• If a required field is missing or has the wrong type, the function will throw a KeyError or MethodError.
• Assumes all values in the TOML file are well-formed. No defaulting or validation is performed.

Example

cfg = parse_full_config_Miriam("config.toml")

println_benji(cfg.input_meta.beta)
println_benji(cfg.solver.maxiter)