home | projects | research | videos | CV | GitHub

Research and publications

peer-reviewed publications

Multiscale chromatin dynamics and high entropy in plant iPSC ancestorsJournal of Cell Science 137(20), 2024.
Kinga Rutowicz, Joel Lüthi, Reinoud de Groot, René Holtackers, Yauhen Yakimovich, Diana M. Pazmiño, Olivier Gandrillon, Lucas Pelkmans, Célia Baroux.
doi:10.1242/jcs.261703 · data

Infectio: a Generic Framework for Computational Simulation of Virus Transmission between CellsmSphere 1(1), 2016.
Artur Yakimovich, Yauhen Yakimovich, Michael Schmid, Jason Mercer, Ivo F. Sbalzarini, Urs F. Greber. The first two authors contributed equally.
doi:10.1128/mSphere.00078-15

Computer vision for image-based transcriptomicsMethods 85, 44–53, 2015.
Thomas Stoeger, Nico Battich, Markus D. Herrmann, Yauhen Yakimovich, Lucas Pelkmans.
doi:10.1016/j.ymeth.2015.05.016

preprints and independent research

CXOR: A One-Bit Context Transform over GF(2) and Its Sixteen Update Rules — 2026.
A one-bit context transform over GF(2): previous source bits select a one-bit table entry and the algorithm emits the prediction residual. The paper classifies the sixteen possible update rules by invertibility and adaptation horizon.
doi:10.5281/zenodo.22727918 · source and reference implementation

Rarefied Data Compression Method — 2008, published 2026.
A lossless compression method based on modular encoding for sources with known alphabet redundancy. It derives limits on the achievable compression ratio and adds a variant in which the characteristic vector of missing symbols is further compressed by enumerative encoding. Written while a teaching assistant at the Belarusian State University of Informatics and Radioelectronics; unpublished until deposited on Zenodo.
doi:10.5281/zenodo.22729977 · PDF

drafts

Work in progress, put online early. Expect rough edges and revisions.

Making Sense of LLM Outputs — draft, 2026.
A model-relative framework for semantic invariance in language. It connects Normalized Information Distance and Normalized Compression Distance to neural entropy coding, separates entropy compression from semantic compression, and treats meaning operationally as behavioral invariance under transformation: a family of semantic utilities, each inducing its own operator and quotient space, replaces a single universal abstraction hierarchy.
draft (PDF) · source · reference implementation


home | projects | research | videos | CV | GitHub
last modified: September 12, 2026