Compression

Compression is a general term for processes that reduce the space required to store, transmit, or represent something. In information technology, it usually refers to reducing digital data size, while in physical sciences it describes compressing a material by applying pressure. Both share the idea of removing redundancy, but differ in mechanisms and loss implications.

In data compression, an encoder analyzes input data and encodes it with fewer bits than the original. Decoding reproduces a reconstruction, ideally identical in lossless compression, or approximately so in lossy compression. Lossless methods, such as Huffman coding or LZ77-based schemes, preserve every bit and are used when exact reproduction is required. Lossy methods, such as JPEG for images or MP3 for audio, discard information to achieve greater reductions, often using perceptual models to minimize visible degradation.

Compression quality is described by metrics such as compression ratio, bit rate, and, in lossy cases, distortion measures like PSNR or SSIM. The process usually includes transforming data to a compact domain, quantization (lossy) or entropy coding, and adding headers or metadata. Theoretical limits arise from information theory; Shannon’s theorem defines the minimum average length for lossless coding given the source entropy, while the rate-distortion function bounds the trade-off between bitrate and fidelity.

Physical compression refers to reducing the volume of a gas or material by applying pressure, approximated by the ideal gas law or more complex equations of state. It is central to engines, refrigeration, and gas handling, and is measured by pressure, volume, and temperature. Compression ratio describes the original versus compressed volume.