solution

class Solution(problem_size: int, batch_size: int, instance_name: str, iterations: int, objective_values: Tensor, solve_time: float, pp_time: float, optimal_value: float, best_value: float, num_frac_values: int, solution_vector: list, variables: dict, evolution_file: str | None = None, device: str = 'cpu', solution_performance: dict | None = None, best_objective_value: float | None = None)

Bases: object

Define the solution class of a solve.

Parameters:

• problem_size (int) – The size of the problem solved.

• batch_size (int) – The number of times the problem instance was solved simultaneously.

• instance_name (str) – The name of the problem instance.

• objective_values (torch.Tensor) – The objective values of the solutions found by the solver.

• iterations (int) – The iteration number for this problem size.

• solve_time (float) – The effective time to solve the problem instance only once.

• pp_time (float) – Time to post-process the problem.

• optimal_value (float) – The optimal objective value for the given problem instance.

• best_value (float) – The best objective value for the given problem instance.

• num_frac_values (int) – Number of fractional values in the solutions of given problem instance.

• solution_vector (list) – Vector of solution to the problem instance found using Gurobi.

• variables (dict) –

  A dict object to store solution related variables. Some fields might be solver-dependent, such as mu, sigma and s. - problem_variables (torch.Tensor): The values of the problem

  variables found by the solver.

• evolution_file (str, optional) – The filename of the evolution file, if it exists.

• device (str, optional) – Device to use, one of: “cpu” or “cuda”. Defaults to “cpu”.

solution_performance

A dictionary contains the following fields - optimal (float): The fraction of the solutions that were within the 0.1% of optimal value. - one_percent (float): The fraction of the solutions that were within the 1% of optimal value. - two_percent (float): The fraction of the solutions that were within the 2% of optimal value. - three_percent (float): The fraction of the solutions that were within the 3% of optimal value. - four_percent (float): The fraction of the solutions that were within the 4% of optimal value. - five_percent (float): The fraction of the solutions that were within the 5% of optimal value. - ten_percent (float): The fraction of the solutions that were within the 10% of optimal value. Defaults to None.

Type:

dict

best_objective_value

The best objective value found by the solver.

Type:

float

batch_size: int

best_objective_value: float = None

best_value: float

device: str = 'cpu'

evolution_file: str = None

get_metadata_dict() → dict

Return the metadata dictonary. Excluding the tensor fields in solution by filtering out the fields where repr sets to False.

Returns:

metadata of the solution.

Return type:

dict

get_solution_stats()

A method that calculates the fraction of solutions that were optimal, within 1%, 2%, 3%, 4%, 5% and 10% of optimal value and update the solution.

instance_name: str

iterations: int

num_frac_values: int

objective_values: Tensor

optimal_value: float

pp_time: float

problem_size: int

save_tensor_to_file(tensor_name, file_dir='.', file_name=None)

Save the tensor that exists in the solution variables dictionary to a file.

Parameters:

• tensor_name (str) – The name (key) that identifies the tensor in the solution.variable dictionary.

• file_dir (str) – The directory of the file. Default to current directory.

• file_name (str) – The name of the file storing tensor. If not provided, defaults to the same name as tensor_name.

Raises:

• Exception – Failed to create the folder path.

• Exception – Tensor_name not exists in Solution.

• Exception – A tensor object cannot be obtained by tensor_name.

solution_performance: dict = None

solution_vector: list

solve_time: float

variables: dict