PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 12, 20260 citationsOpen Access

360-DQHM: A Phase-Topology-Based Memory Architecture Achieving O(1) Search Complexity at Billion-Entry Scale

View Full Paper
SMShigeyuki Matsunaga

Key Points

  • The aim is to develop a memory architecture that enables constant-time retrieval for a large number of entries.
  • Developed a 360-dimensional complex-valued hypervector space for storing data.
  • Encoded key-value pairs as phase-shifted superpositions within a single vector.
  • Evaluated performance on an AMD Ryzen 7 5700X CPU without GPU across varying dataset sizes.
  • Achieved retrieval accuracy of 100% across seven orders of magnitude (10^3 to 10^9).
  • Maintained O(1) retrieval time at 27-28 µs irrespective of dataset size.
  • Reduced memory usage to only 2,880 bytes for up to 10^9 entries.

Abstract

We present 360-DQHM (360-Dimensional Dynamic Quanta Hierarchical Memory), a holographic associative memory architecture based on a 360-dimensional complex-valued hypervector space. By encoding key-value pairs as phase-shifted superpositions within a single C³60 vector, the system stores up to 10⁹ entries in only 2, 880 bytes of memory while maintaining constant-time O (1) retrieval at 27-28 µs regardless of dataset size. On an AMD Ryzen 7 5700X CPU (no GPU), empirical evaluation across seven orders of magnitude (10³ to 10⁹) demonstrates 100% retrieval accuracy under the proposed Phase-Resonance Similarity metric.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Shigeyuki Matsunaga (2026) studied this question.

synapsesocial.com/papers/69b2588496eeacc4fcec83a5https://doi.org/10.5281/zenodo.18922676
Ask AI
Helpful
Bookmark
Share
View Full Paper