The Canonical Data Atlas
Every claim HolonomiX makes is grounded in a systematic evidence framework. The Atlas classifies 475 computational workloads across four axes and assigns an empirical verdict: does QTT compression govern this workload, or not?
41 predictive rules. 3 formal theorems. 64 of 64 taxonomy cells covered. No workload class left uncharacterized.
Four verdicts. No ambiguity.
Every workload in the Atlas receives a verdict based on empirical profiling. The verdict tells you whether QTT compression governs the workload, under what conditions, and with what confidence.
Governable
QTT rank stays bounded. Compression at or above 100x. Use it.
Compressible
Compresses with caveats or specific configuration. Production-viable with tuning.
Conditional
Governor saturates at some scales. Conditional on rank budget and workload size.
Weak
QTT does not provide useful compression. The technology is not the right tool for this workload.
Effectiveness is a joint function of four axes
QTT compression does not have a single answer for any workload. Performance depends on the intersection of data structure, task type, tensorization ordering, and algorithm choice. The Atlas maps this entire space.
Data structure
smooth, separable, discontinuous, chaotic, stochastic
Task type
represent, solve, evolve, apply, query, optimize, control, compose
Tensorization ordering
lexicographic, bit-reversed, Hilbert, interleaved
Algorithm choice
forward Euler, backward Euler, RK4, and more
The evidence is open
The Canonical Data Atlas is the empirical foundation behind every HolonomiX product. Request access to explore the full dataset, predictive rules, and formal proofs.