An Efficient Lossless Compression Algorithm for Trajectories of Atom Positions and Volumetric Data.

Area of Science:

• Computational chemistry
• Data compression
• Scientific visualization

Background:

• Efficient storage and retrieval of large scientific datasets are crucial for computational chemistry and molecular dynamics simulations.
• Existing compression methods often struggle to balance compression ratios with decompression speed and random access capabilities for trajectory data.

Purpose of the Study:

• To develop and present a novel, highly efficient lossless compression algorithm tailored for atom position trajectories and volumetric scientific data.
• To introduce a robust and flexible file format (BQB) that supports fast random access to compressed trajectory frames.

Main Methods:

• A two-step compression approach utilizing polynomial extrapolation to reduce data entropy by exploiting spatial and temporal continuity.
• Application of a series of transformations including Burrows-Wheeler, move-to-front, and run-length encoding, followed by multitable canonical Huffman coding.
• Implementation in C++ and provision as free software, integrated into the TRAVIS package and available as a standalone tool or library (libbqb).

Main Results:

• Achieved compression ratios of approximately 15:1 for XYZ position trajectories and 35:1 for Gaussian Cube volumetric data.
• Demonstrated that the compression and decompression speeds are suitable for everyday use, with user-selectable data precision.
• Introduced the BQB file format, offering robustness, flexibility, efficiency, and fast random access to individual trajectory frames, outperforming other compared formats.

Conclusions:

• The developed algorithm offers a significant improvement in data compression for scientific trajectories, enabling more efficient storage and handling of large datasets.
• The BQB format provides a superior solution for storing compressed trajectory data, particularly due to its random access capabilities.
• The algorithm's availability as free software promotes its adoption and integration into various scientific workflows and projects.