Read DataA.csv and return an :class:`Instance`. The CSV must contain at least the columns documented in the README. Pay-rate columns (`DutyCo

Get the provider name from a model identifier.

Extracts the base model name from a model identifier for file paths.

Convert separate date+time strings into a naive datetime object.

Calculate solve@i metrics for the first i iterations.

Compute the total cost of a crew-pairing solution. Parameters ---------- input_file : str Path to **DataA.csv** (instance file).

Find all iteration directories and their sample subdirectories.

Builds two maps for 0-indexed domain IDs: 1. id_to_genotype: 0-idx ID -> frozenset({allele1, allele2}) 2. genotype_to_id: frozenset({alle

Determines the number of unique alleles (n_alleles). It considers the maximum allele value explicitly observed (e.g., if allele '4' is seen,

Map resource type to internal representation

Parse the JSON file and build the graph. Returns: A tuple of: - Dictionary mapping node IDs to Node objects - Dictionary

Parse the schedule file. Returns: Dictionary mapping node IDs to their scheduled start times

Perform topological sorting using DFS and set ASAP/ALAP times

Pre-compute helper maps for fast constraint checks.

Calculate As Late As Possible (ALAP) times

Calculate As Soon As Possible (ASAP) times

Raise RuntimeError if `abc` binary is not on PATH.

Raise RuntimeError if `abc` binary is not on PATH.

Check if the candidate BLIF file satisfies the LUT size constraint. Args: candidate: Path to the candidate BLIF file max

Helper function for DFS traversal

Formats the problem information section for the prompt.

Get the actual model name to use with the API.

Returns a default prompt template if prompt.md is not found.

Gets an example program template for the problem.

Run ABC `ps` and return the LUT count (nodes).

Run ABC `ps` and return the LUT count (nodes).

Loads the prompt template from prompt.md.

Parse the reasoning_effort from the model name if specified.

Read the solver output file. Each non-blank, non-comment line represents **one pairing** and contains the space-separated list of leg tokens

Read schedule file → list of pairings (each is list of leg tokens).

Run `cec` in ABC and return the CompletedProcess.

Run `cec` in ABC and return the CompletedProcess.

Saves API information to a JSON file.

Build candidate pairings more efficiently with iterative approach and pruning. Uses a breadth-first approach with aggressive pruning to avoid

Builds the graph G for the min-cut problem.

Calculate the cost for a given model and token usage.

Calculates pairwise distances between residues.

Calculate geometric mean, coverage, and QCI for an iteration's results using baseline values.

Calculates the percentage agreement between two sequences.

Calculates SASA using Bio.PDB.SASA.ShrakeRupley. Args: structure (Bio.PDB.Structure): The parsed PDB structure object. re

Classify error message into operator scheduling specific categories.

Return (input_csv, output_txt) based on repo layout.

Cost calculation function: calculates the cost. Args: input_file: Path to the input file solution_file: Path to the outp

Cost calculation function: calculates the cost. Suppose the input has been verified by the verifier, which means the input to this function i

Cost calculation function: calculates the solution cost. Args: input_file: Path to the input file containing the instance solutio

Evaluate the solution for a given input file and output file. Args: input_file (str): Path to the input file. output_fil

Sum the 'cost' field of every selected node that actually exists.

Run heurigym evaluator and return the total cost.

Cost calculation function: calculates the cost. Suppose the input has been verified by the verifier, which means the input to this function i

Runs the Python program and returns success status and output.

int index for elcass id and enode id

Extract error information from .cost files in the iteration directory.

Find all potential datasets under the given base directory and group them by base name. Args: base_dir (str): Base directory to sear

Find all .result files in the given directory.

Find all run.py files in iteration directories and their sample subdirectories.

Formats the problem description into a prompt for the LLM.

Create a Graph instance from a JSON file. Args: json_file: Path to the JSON file constrained_latency: Optiona

Sigmoidal function g(d_ij) from the paper.

Generate a bar chart for latency comparison.

Generate ILP for resource-constrained problems

Generate a formatted summary table comparing all models' performance.