pith. sign in

arxiv: 2606.24531 · v1 · pith:JAHFOA7Bnew · submitted 2026-06-23 · 💻 cs.DC

Unified Position-Invariant Random Access Through Two Compression Layers via Absolute-Offset Coordinates: A Bit-Perfect Device-Resident Proof

Yakiv Shavidze This is my paper

Pith reviewed 2026-06-25 22:47 UTC · model grok-4.3

classification 💻 cs.DC
keywords random accesscompression layersabsolute offsetsmatch layerentropy layerbit-perfect decodingparallel parsersdevice-resident verification
0
0 comments X

The pith

Absolute-offset coordinates let any block decode bit-perfectly through match and entropy layers using one coordinate.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that resolving all back-references to absolute positions at encode time in the match layer, while applying entropy coding independently per block, removes the need for context from other blocks. This makes it possible to seek to and decode an arbitrary block through both layers with a single coordinate, producing an exact match to the original data. The claim is supported by a three-phase verification process that handles edge cases such as empty buffers. Practical results include a 0.334 ms seek time for a 16 KB block and the ability to run up to 25,344 independent parsers on one H100. The work is framed as a round-trip correctness proof for two layers rather than a complete storage format.

Core claim

ACEAPEX's absolute-offset design provides exactly this: because the match layer resolves every back-reference to an absolute position at encode time, and the entropy layer is applied per block, an arbitrary block can be decoded through both layers using one coordinate, bit-perfect, in isolation. We prove this with a three-phase verification that closes the empty-buffer trap. The seek of one 16KB block through ANS-entropy and match completes in 0.334ms. We verify the full entropy+match pipeline end-to-end on four data profiles and characterize the hardware ceiling the format reaches: the absolute-offset structure unrolls to as many as 25,344 independent parsers on one H100, which sequential L

What carries the argument

Absolute-offset coordinates that resolve every match-layer back-reference to a fixed position at encode time, combined with per-block entropy coding.

If this is right

  • Any single block can be decoded through the full match-plus-entropy pipeline without loading neighboring blocks.
  • The approach supports up to 25,344 independent parsers on one H100 GPU.
  • Seek time for a 16 KB block is 0.334 ms through both layers.
  • The result holds for two compression layers, with three-layer generalization stated as a hypothesis.
  • The verification is a round-trip correctness proof, not a claim about disk-archive formats or throughput.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The independence of blocks could reduce memory traffic in large-scale parallel decompression systems.
  • Similar absolute-position techniques might apply to other multi-layer codecs to enable isolated access.
  • Hardware designs could allocate separate parser units per block to reach the reported parallelism ceiling.
  • Testing the three-layer hypothesis would require extending the three-phase verification to an additional compression stage.

Load-bearing premise

The match layer resolves every back-reference to an absolute position at encode time with no residual dependencies between blocks that would require additional context for isolated decoding.

What would settle it

A block whose decoding produces a bit mismatch or requires state from another block despite using only the supplied absolute coordinate.

read the original abstract

Random access into compressed data is normally confined to a single layer. Entropy-layer methods (Recoil) seek within rANS by storing intermediate decoder states; dictionary/match-layer methods seek within LZ-style references. We are not aware of a format that supports a single position-invariant seek through both an entropy layer and a match layer addressed by one coordinate. We show that ACEAPEX's absolute-offset design provides exactly this: because the match layer resolves every back-reference to an absolute position at encode time, and the entropy layer is applied per block, an arbitrary block can be decoded through both layers using one coordinate, bit-perfect, in isolation. We prove this with a three-phase verification that closes the empty-buffer trap. The seek of one 16KB block through ANS-entropy and match completes in 0.334ms. We verify the full entropy+match pipeline end-to-end on four data profiles and characterize the hardware ceiling the format reaches: the absolute-offset structure unrolls to as many as 25,344 independent parsers on one H100, which sequential LZ77 cannot do. We state explicitly what is not claimed: this is a round-trip correctness proof, not a disk-archive format; throughput figures are match-phase; and the unified-seek result is demonstrated for two layers, with three-layer generalization left as a hypothesis. Code and the verification harness are in the project repository.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript presents ACEAPEX, a compression scheme using absolute-offset coordinates in the match layer combined with per-block entropy coding. It claims this enables unified position-invariant random access through both layers via a single coordinate, yielding bit-perfect isolated decoding of an arbitrary block. The claim is supported by a three-phase verification asserted to close the empty-buffer trap, with reported seek time of 0.334 ms for a 16 KB block and potential for up to 25,344 independent parsers on one H100; the work explicitly disclaims being a disk-archive format or claiming three-layer generalization.

Significance. If the verification is sound, the result would provide a concrete mechanism for dependency-free random access across two compression layers, enabling hardware parallelism that sequential LZ77-style methods cannot achieve. The explicit statement of scope limitations and the availability of code plus verification harness in the repository are positive contributions to reproducibility.

major comments (1)
  1. [Abstract] Abstract (three-phase verification paragraph): the central claim that absolute-offset resolution at encode time plus per-block entropy produces no residual inter-block dependencies (enabling isolated decode from one coordinate) rests on the unexamined assertion that the verification closes the empty-buffer trap; the manuscript supplies no concrete description of whether the encoder is restricted to intra-block matches only or how cross-block absolute references are resolved without supplying external decompressed data.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful review and for highlighting the need for greater clarity in the abstract. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (three-phase verification paragraph): the central claim that absolute-offset resolution at encode time plus per-block entropy produces no residual inter-block dependencies (enabling isolated decode from one coordinate) rests on the unexamined assertion that the verification closes the empty-buffer trap; the manuscript supplies no concrete description of whether the encoder is restricted to intra-block matches only or how cross-block absolute references are resolved without supplying external decompressed data.

    Authors: We agree that the abstract does not supply a concrete description of the encoder's match policy or the resolution of potential cross-block references. The manuscript's core design resolves every back-reference to an absolute coordinate at encode time and applies entropy coding independently per block; the three-phase verification is asserted to confirm that this produces no residual dependencies, allowing isolated bit-perfect decoding from a single coordinate. However, the abstract does not explicitly state the encoder restriction or the verification steps that close the empty-buffer trap. We will revise the abstract (and add a short clarifying paragraph in the main text) to state that the encoder is restricted to intra-block matches only, thereby eliminating any requirement for external decompressed data, and to outline the three verification phases at a high level. This change will be incorporated in the revised manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity: structural proof of absolute-offset design is self-contained

full rationale

The paper presents a design claim that absolute-offset resolution at encode time plus per-block entropy enables isolated two-layer decoding, verified via three-phase check for the empty-buffer trap. No equations, fitted parameters, or predictions are shown that reduce the result to its own inputs by construction. No self-citation load-bearing steps, uniqueness theorems, or ansatz smuggling appear in the provided text. The result is a claimed structural property of the format rather than a renamed empirical pattern or statistically forced output, making the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities; the central claim rests on the absolute-offset encoding property described.

pith-pipeline@v0.9.1-grok · 5782 in / 1126 out tokens · 13821 ms · 2026-06-25T22:47:49.306073+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

8 extracted references · 6 canonical work pages · 3 internal anchors

  1. [1]

    ACEAPEX: Parallel LZ77 Decoding via Encode-Time Absolute Offset Resolution

    Y. Shavidze, “ACEAPEX: Parallel LZ77 Decod- ing via Encode-Time Absolute Offset Resolution,” arXiv:2606.04268, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  2. [2]

    Compressed-Resident Genomics: Full-Pipeline Device-Resident GPU LZ77 Decode with Position-Invariant Random Access

    Y. Shavidze, “Compressed-Resident Genomics: Full-Pipeline Device-Resident GPU LZ77 De- code with Position-Invariant Random Access,” arXiv:2606.18900, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  3. [3]

    Recoil: Parallel rANS Decoding with Decoder-Adaptive Scalability,

    T. Lin et al., “Recoil: Parallel rANS Decoding with Decoder-Adaptive Scalability,” arXiv:2306.12141, 2023

    work page arXiv 2023

  4. [4]

    A Modular Genomic Compressor,

    K. Szewczyk and S. Rahmann, “A Modular Genomic Compressor,” arXiv:2603.15390, 2026

    work page arXiv 2026

  5. [5]

    DietGPU: GPU-based Lossless Com- pression for Numerical Data,

    J. Johnson, “DietGPU: GPU-based Lossless Com- pression for Numerical Data,” open-source, 2022

    2022

  6. [6]

    Interleaved entropy coders

    F. Giesen, “Interleaved entropy coders,” arXiv:1402.3392, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  7. [7]

    LZ77-like Compression with Fast Random Access,

    S. Kreft and G. Navarro, “LZ77-like Compression with Fast Random Access,” inProc. Data Compres- sion Conference (DCC), 2010, pp. 239–248

    2010

  8. [8]

    Understanding Entropy Coding With Asymmetric Numeral Systems (ANS),

    R. Bamler, “Understanding Entropy Coding With Asymmetric Numeral Systems (ANS),” arXiv:2201.01741, 2022. 5

    work page arXiv 2022