{
  "abstract": "Mamba-2 compresses all context into a fixed-size recurrent state that dominates decoding memory and bandwidth. Existing compressors quantize state values or prune state coordinates, scoring each channel independently in the native basis. We examine trained Mamba-2 states and find that energy is almost rank-one, but the output does not read the energetic directions: energy concentrates in a bias-like direction accumulated by the recurrence, while the content the output reads lives in a low-dimensional subspace that is not aligned with the state's coordinates. Even output-aware scores are computed coordinate by coordinate, so they miss the directions that determine the input--output map. Compression should therefore target the write--read subspace rather than individual coordinates. We propose \\method{}, a post-training transform that instantiates balanced truncation with full controllability and observability Gramians and applies the rotation exactly via Mamba-2's scalar-identity transition. Because the useful rank varies widely across layers, we further introduce HARP, which allocates ranks per layer from the Hankel spectrum and protects retrieval-sensitive layers, improving both perplexity and retrieval over uniform budgets. Calibration takes a single forward pass, an order of magnitude faster than GHOST's released protocol, with lower peak memory. With HARP, \\method{} halves the state dimension at an average WikiText-2 perplexity increase of $0.3$ points ($1.8\\%$) across 130M--2.7B, versus $1.6$ ($9.3\\%$) for GHOST, and the gap widens at 25\\% retention.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.36617",
  "author_total": 2,
  "authors": [
    {
      "name": "Shang Tingyu"
    },
    {
      "name": "Cao Wenfei"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/0t8z6y04c1can.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.36617",
  "fulltext_url": null,
  "id": "0t8z6y04c1can",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.36617",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.AP"
    ],
    [
      "arxiv",
      "stat.ME"
    ],
    [
      "openalex",
      "1707|Advanced Neural Network Applications"
    ]
  ],
  "openalex": "W7224123059",
  "pdf_url": "https://arxiv.org/pdf/2610.36617",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-27",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/0t8z6y04c1can.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:35:59Z",
      "native_id": "2610.36617",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.36617"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.AP",
      "subfield": "statistics"
    }
  ],
  "title": "Balanced State Subspaces for Mamba-2 State Compression",
  "title_text": "Balanced State Subspaces for Mamba-2 State Compression",
  "type": "preprint",
  "updated_at": "2026-10-30T12:35:59Z",
  "url": "https://research.draconitic.com/paper/0t8z6y04c1can",
  "venue": "arXiv",
  "versions": []
}