Multiplates

This repository contains materials to reproduce the evaluation section of the submitted paper "Minimising Source-Plate Swaps for Robotised Compound Dispensing in Microplates"

The materials include:

1. Script to generate synthetic datasets

2. Two synthetic datasets, multiplate_dzn_files.zip and multiplate_dzn_files_2_only.zip:

   1. multiplate_dzn_files.zip is not used for the article. The number of connected source plates to a given compound combination is unrestricted
   2. multiplate_dzn_files_2_only.zip is used for the evaluation in the article. The number of connected source plates to a given compound combination is limited by a maximum of 2

3. Scripts and configuration files to run computational experiments

4. Dataframes, logs and a Python Notebook with aggregated results

Data generation

The generated dataset is contained in files multiplate_dzn_files.zip and multiplates_models_evaluation.csv (Dataset 1), multiplate_dzn_files_2_only.zip and multiplates_models_evaluation_2_only.csv (Dataset 2). Uncompress them and move to the next step of the evaluation.

If you wish to generate datasets anew then:

1. Open a bash shell and position yourself in the directory containing the file generate_multiplate_model.pl

2. Run SICStus Prolog from the terminal (version 4.6.0 or a more recent version) and execute the following commands:

   1. | ?- [generate_multiplate_model], generate_all_dzns(1), halt. to generate Dataset 1
   2. | ?- [generate_multiplate_model], generate_all_dzns(2), halt. to generate Dataset 2

Each dataset generation takes a few minutes.

Create the folder multiplate_dzn_files and/or multiplate_dzn_files_2_only if not created prior.

Executing computational experiments

To execute the computational experiments first install and configure Python (version 3.7.0 or later), MiniZinc (version 2.8.0 or later) and Gurobi (version 11.0.3 or later). Make sure that the versions of installed solvers, Chuffed, CP-SAT and Gurobi, are matched the versions written in the configuration files chuffed_config.mpc, cpsat_config.mpc and gurobi_config.mpc respectively.

Then:

1. Open a bash shell and position yourself in the directory containing files run_evaluation.py, run_evaluation_cfg1.py, run_evaluation_cfg2.py, run_evaluation_cfg3.py, run_evaluation_cfg1_2combo.py, run_evaluation_cfg2_2combo.py, run_evaluation_cfg3_2combo.py
2. Modify the file paths in run_evaluation.py, if needed
3. Execute commands, sequentially or in parallel:
   1. time python3 run_evaluation_cfg1.py to test Dataset 1 on Chuffed (1 thread) (NOT REQUIRED FOR THE ARTICLE)
   2. time python3 run_evaluation_cfg2.py to test Dataset 1 on CP-SAT (4 threads) (NOT REQUIRED FOR THE ARTICLE)
   3. time python3 run_evaluation_cfg3.py to test Dataset 1 on Gurobi (4 threads) (NOT REQUIRED FOR THE ARTICLE)
   4. time python3 run_evaluation_cfg1_2combo.py to test Dataset 2 on Chuffed (1 thread)
   5. time python3 run_evaluation_cfg2_2combo.py to test Dataset 2 on CP-SAT (4 threads)
   6. time python3 run_evaluation_cfg3_2combo.py to test Dataset 2 on Gurobi (4 threads)

Aggregating the results

Execution of computational experiments will result in the creation of files multiplates_models_evaluation_complete_chuffed_config.csv, multiplates_models_evaluation_complete_cpsat_config.csv, multiplates_models_evaluation_complete_gurobi_config.csv, multiplates_models_evaluation_two_only_chuffed_config.csv, multiplates_models_evaluation_two_only_cpsat_config.csv, multiplates_models_evaluation_two_only_gurobi_config.csv.

Open the Jupyter Notebook evaluate_multiplates.ipynb in, either, Anaconda, Google Collab or equivalents. Rerun or modify the code if needed.

Citation

The following manuscript can be used to cite this project:

R. Gindullin, M. A. Francisco Rodríguez, B. Seashore-Ludlow, and O. Spjuth Minimising Source-Plate Swaps for Robotised Compound Dispensing in Microplates In: G. Tack (editor), CP-AI-OR 2025. Lecture Notes in Computer Science, volume 15762, pages 256-273. Springer, 2025. DOI: 10.1007/978-3-031-95973-8_16

@InProceedings{PlateSwaps2025,
	author = {Gindullin, Ramiz and Francisco Rodr\'iguez, Mar\'ia Andre\'ina and Seashore-Ludlow, Brinton and Spjuth, Ola},
	title = {Minimising Source-Plate Swaps for Robotised Compound Dispensing in Microplates},
	year = {2025},
	doi = {10.1007/978-3-031-95973-8_16},
	URL = {https://doi.org/10.1007/978-3-031-95973-8_16},
	editor = {Tack, Guido},
	booktitle = {Integration of Constraint Programming, Artificial Intelligence, and Operations Research},
	publisher = {Springer Nature Switzerland},
	pages = {256--273},
	isbn = {978-3-031-95973-8}
}