Empire Technologies — Synthetic Compression Test Inputs (raw_bias_1gb and random_100mb)

Two synthetic data files from the Empire Technologies standardized benchmark test set, used as fixed reference inputs for compression testing. They are published separately because, as synthetic inputs, they have no third-party canonical source. The full test set includes additional public-corpus files (Silesia, enwik, etc.) which already have their own canonical hosts; the two files here are the synthetic subset that requires permanent Zenodo anchoring to be externally verifiable. These two files serve as bounds-testing anchors at opposite ends of the entropy spectrum: raw_bias_1gb.bin (1,073,741,824 bytes) — a bias-dominant synthetic input representing highly structured, low-entropy data. Used to demonstrate preprocessor effectiveness on structured streams. In Empire's pipeline this file compresses at approximately 471.52x ratio with byte-identical roundtrip. random_100mb.bin (104,857,600 bytes) — a pseudo-random byte stream representing the maximum-entropy boundary condition. Used to prove graceful pipeline behavior when no compression is possible. In Empire's pipeline the compressed output is approximately 0.26% larger than the input, with byte-identical roundtrip preserved. Publishing these files as a permanent dataset with canonical SHA256 hashes allows independent researchers to: 1. Download the exact byte sequences Empire uses in its published benchmarks 2. Verify claims from Empire's Provenance system (compression ratios, roundtrip fidelity) 3. Cross-check against other compressors' behavior on the same inputs Canonical SHA256 anchors (full 64-character): raw_bias_1gb.bin: 5c7c67ddc1caf760208f9ee7026e2056a734d9462cac431179e000e605e980a2 random_100mb.bin: dee92f51b751c54061fad47d553c0ad3f6fd2f059ee129f1902a88a74834de70 Files are deterministic and fixed. No updates planned; this is a stable reference release. Related work:"Compression's Least Discussed Limit: Architectural Choice" — compression architecture thesis (open decoder, proprietary encoder pattern, structural bounds); directly relevant to this dataset's role in the benchmark system.Available in the Empire Zenodo preprint bundle: https://doi.org/10.5281/zenodo.18238100