This database contains the results from various solvers applied to Kings graph with 50 rows and 50 columns and unit cells of size 1 (50x50x1) and 2 (50x50x2) unit cells. The purpose of this structure is to allow efficient access to results based on the size and structure of instances.
Each result file includes the instance ID, the corresponding lowest energy (solution quality), and the state (spin configuration) representing the solution. The spin configurations are represented with the following values:
1: Spin up
-1: Spin down
0: Missing spin
------------------------------------------------------------------------
System Sizes
------------------------------------------------------------------------
Size description:
50x50x1 - Kings graph with 50 rows, 50 columns, and a unit cell with one spin.
50x50x2 - Kings graph with 50 rows, 50 columns, and a unit cell with two spins.
All instances have biases set to zero and couplings generated uniformly at random from range [-1, 1]
A detailed description of instances and unit cells can be found in the publication this dataset was named after.
------------------------------------------------------------------------
Directory Structure
------------------------------------------------------------------------
Database:
├── 50x50x1
│ ├── instances
│ ├── CPLEX.csv
│ ├── SBM.csv
│ └── TN.csv
├── 50x50x2
│ ├── instances
│ ├── CPLEX.csv
│ ├── SBM.csv
│ └── TN.csv
-----------------------------------------------------------------------
File Contents
------------------------------------------------------------------------
Each CSV file (CPLEX.csv, SBM.csv, TN.csv) contains the results from the corresponding solver.
TN.csv: The best energy results from the tensor network based solver without dimensional reduction of cluster degrees of freedom
SBM.csv: The best energy results from the Simulated Bifurcation Machine (SBM)
CPLEX.csv: The best energy results from IBMs CPLEX optimization software
----------------------------------------------------------------------------------------------------
The CSV files have the following format:
Instance, Energy, State
Where:
Instance: The identifier for each problem instance, numbered in ascending order (e.g., 001, 002, 003).
Energy: The lowest energy obtained by the solver for the corresponding instance.
State: A string of comma-separated spin values (1, -1, 0), representing the spin configuration for that instance's solution. The length of the state vector corresponds to the number of spins used by the solver. Missing spins (where a particular qubit wasn't used) are represented by 0.
Example .csv file:
Instance,Energy,State
001,-469.0,"1,1,1,-1,1,0,1,-1,1,1,1,1,1,-1,1"
002,-492.0,"1,1,-1,1,1,0,-1,1,1,-1,1,1,1,1,1"
---------------------------------------------------------------------------------------
Instances Folder:
Each folder contains a subfolder named instances, which holds the .txt files defining the problem instances for each solver run.
(2024)