Our Work

MD simulation data in HDAC6.

Datasets used within article "Near-complete 3D deformation mapping from Sentinel-1 through extended burst overlap interferometry" (in prep.) The datasets contain products generated from Sentinel-1 dat…

Datasets used within article "Near-complete 3D deformation mapping from Sentinel-1 through extended burst overlap interferometry" (in prep.) The datasets contain products generated from Sentinel-1 data. Data use WGS-84 coordinate system.   dataset_QML.zip: Standard (BOI) and extended (EBOI) burst overlap interferogram over Queen Maud Land (relative orbit 2, ascending track, between 2026-03-06 and 2026-03-07). Note the BOI data are downsampled to 0.001x0.001 deg resolution. Format: GeoTIFF, unit: rad   dataset_tur.zip: Input data and results of decomposition inversion capturing coseismic deformation of 2023 Türkiye earthquake sequence (Kahramanmaraş earthquakes), as described in the article. Input data are from four overlapping frames (014A, 021D, 116A, 123D) as a measurement converted to metres (variable increment) and include unit vectors (variables E, N, U) and estimated precision (variable std [m]). Input data includes: unwrapped interferograms (insar_XXXX.nc), standard burst overlap interferograms (boi_XXXX.nc), extended burst overlap interferograms (eboi_XXXX.nc), azimuth pixel offsets (azi_XXXX.nc) and range pixel offsets (range_XXXX.nc). Decomposition data include outputs combining range pixel offsets, unwrapped interferograms, azimuth pixel offsets (decomposed_azi.nc) and with burst overlap interferograms (decomposed_boi.nc) or extended burst overlap interferograms (decomposed_eboi.nc) where these exist. Format: NetCDF-4, unit: m

This milestone document reports the status of CardioSynth, the multi-centre, multilingual and fully synthetic heart-failure dataset produced by DataTools4Heart (DT4H). It briefly reported the met…

This milestone document reports the status of CardioSynth, the multi-centre, multilingual and fully synthetic heart-failure dataset produced by DataTools4Heart (DT4H). It briefly reported the methods and results of the full CardioSynth report into three questions: why CardioSynth was built, how it is generated, and what it consists of today (September 2026). CardioSynth is generated inside the secure environment of each participating hospital from adult heart-failure hospitalisations extracted through the DT4H Common Data Model. One pipeline produces structured synthetic records with nine families of generative models, evaluates them against the natural variability between two real samples (the noise floor) and a suite of privacy attacks, and passes every candidate file through an automated release gate. A second pipeline writes one clinical note per synthetic patient, in the language of the centre and in the style of clinician-authored documents about fictional patients, so that no real patient text is ever used. Structured and unstructured generation is complete or ongoing at several centres and planned at the others. The first version of CardioSynth (i.e CardioSynth v1.0) includes data generated with two methods (TVAE and MST) for structured data and derived notes from four clinical centres. Following approval from the local officers, CardioSynth will therefore be released in tiers: open documentation, code and evaluation evidence; gate-cleared synthetic data under a data use agreement on the DataTools4Heart platform; and tailored cohorts generated on request inside the hospitals. The route to a fully open release, and the governance decisions it requires, are described in the final section. Full details of the pipeline to build CardioSynth are published in the Deliverable 7.2 (same date of publication).

Around two-fifths of the population living in Saudi Arabia consists of expatriates. However, there is limited research on how these families maintain their heritage languages (HLs) in a digital world …

Around two-fifths of the population living in Saudi Arabia consists of expatriates. However, there is limited research on how these families maintain their heritage languages (HLs) in a digital world with limited institutional support. Maintaining HLs is important for identity, cultural continuity, and a sense of belonging across generations, especially for families living far from home. This qualitative study explores how expatriate parents in Saudi Arabia use digital tools to help their children maintain HLs, using Fishman’s reversing language shift (RLS) framework and family language policy (FLP) theory. Researchers interviewed 36 expatriate parents from 15 different national and linguistic backgrounds and analyzed the data with reflexive thematic analysis in NVivo. The results show that families are moving from exclusively home-based language practices to a mix of digital strategies, such as apps, video calls, and online learning spaces, which help strengthen cross-border connections and increase language exposure. However, these new practices also increase mothers’ workload, as they take on most of the planning, mediation, and emotional support. The study suggests policy and practical steps that fit with Saudi Arabia’s Vision 2030, such as providing subsidized multilingual digital resources and family-focused support programs. The clear research design makes the study easy to replicate, and future research should include lower-income families, children’s views, and long-term studies of digital family language practices.

No description provided.

             Single-cell RNA sequencing was used to analyze transcriptional profiles during erythropoiesis in erythroblasts from normal subjects, mild, and severe &b…

             Single-cell RNA sequencing was used to analyze transcriptional profiles during erythropoiesis in erythroblasts from normal subjects, mild, and severe β-thalassemia/Hb E patients, revealing disease severity- and differentiation stage-associated alterations in erythroid transcriptional states. Erythroblasts from each group were collected on days 8 and 10 of differentiation phase and labeled with lipid-modified oligonucleotide (LMO)-based MULTI-seq barcodes. GEM generation and library preparation were performed according to the protocol provided with the Chromium GEM-X Single Cell 3′ Reagent Kits v4 (10x Genomics).              In this study, two libraries were generated: library 1 contained 16,673 cells from erythroblasts on day 8, whereas library 2 contained 12,754 cells from erythroblasts on day 10. R script used for integration, clustering, differential expression analysis, functional enrichment analysis, and visualization of single-cell RNA sequencing data is available in “Rscript_CS07_integ” file. All sequencing raw and processed data have been uploaded to NCBI’s Gene Expression Omnibus (accession no. GSE348715).

This article analyzes the state of introduction of artificial intelligence technologies in the production infrastructure of enterprises of Navoi region. The study highlights the impact of artificial i…

This article analyzes the state of introduction of artificial intelligence technologies in the production infrastructure of enterprises of Navoi region. The study highlights the impact of artificial intelligence use on productivity, automation levels, rational resource utilization, and product quality. The proposals also include developing digital infrastructure in enterprises, widely introducing modern technologies, training qualified specialists, and increasing the efficiency of using artificial intelligence solutions, as well as scientifically substantiating existing problems and promising development directions.

This is the archival home of the PolyClank Collatz research workbench: an ongoing, open human–LLM collaboration studying the mathematics of the Collatz map. The ordinary map sends an even positi…

This is the archival home of the PolyClank Collatz research workbench: an ongoing, open human–LLM collaboration studying the mathematics of the Collatz map. The ordinary map sends an even positive integer n to n/2 and an odd one to 3n+1. The workbench reconstructs the literature, develops exact comparison maps and new deductions, and makes the arguments available with source files, reproducible calculations and explicitly scoped formal proofs. GitHub workbench · Reading guide · PolyClank discussion and contributions Mathematical programme The wider workbench contains source-faithful exposition of Maxwell C. Siegel’s Hydra and numen constructions; a separate study of clocks and first-passage measures in Terence Tao’s theorem; and independent work on weighted odd-step paths, arithmetic histories, affine transport, finite spectral encodings, integral comparisons and bounded-support constructions. These strands have separate manuscripts and attribution. Exact maps and proofs, rather than a shared vocabulary, establish their connections. First passage, natural density, clocks and height The principal reader in this edition is Conditioning and first-passage fibres in the natural-density argument. It reconstructs the joint Syracuse offset–valuation calculation and the arithmetic first-passage fibres, develops the uniform kernel and timed-target estimates, and compares the arguments of Tao, Allikvere, Mazur and Shaik. It also credits Inselmann’s earlier drift-envelope results. One resulting synthesis controls three clocks and the height of the same orbit prefix. Let d = log(4/3), with natural logarithms. For every β,e > 0 there is A = A(β,e), independent of B,c,X, such that for every X ≥ 2, integer B ≥ 2 and 0 < c < 1/17.232, all but at most C_c X(log B)−c + C_(β,e) X(log(X+2))−e starting integers n ≤ X have a prefix ending at an odd value ≤ B. Its odd-step, shortcut-step and ordinary-step counts are bounded, respectively, by j log(n)/d + A(log(n+2))4/5, for j = 1,2,3, and its largest ordinary value is at most n1+β. A shortcut odd step combines 3n+1 with one halving; a Syracuse return removes all its powers of two. The proof joins a height-controlled passage to a polylogarithmic level with a timed first hit of B on the original deterministic orbit. A lower bound for time already spent leaves only O((log n)4/5) odd returns in the tail. Exact valuation telescoping then controls the two other clocks and the tail height, without assuming independence or restarting a probability law at the landing point. The fixed-target estimate implies passage strictly below every f(n) → ∞ for a natural-density-one set, with clocks independent of f. It does not imply that every orbit reaches 1. Further results include an elementary endpoint-separation estimate, a kernel exponent c < 1/17.232 using the cited Rhin input, refinements of terminal scalar estimates, and exact finite timeout and completed-block height bounds. The reader gives the hypotheses, proofs, dependencies and limitations of each result. These are workbench reconstructions and deductions, not a claim to have corrected Tao’s theorem or independently certified every source manuscript. Sources and verification The principal sources are Tao’s v7 paper, Allikvere’s v2 manuscript, Mazur’s natural-density argument, Shaik’s v3.2.4 manuscript and its pinned source, and Inselmann’s v3 paper. Precise source versions and locators appear in the manuscript and claim records. The archive is a complete snapshot of the public GitHub workbench at commit e56508d9102bf16442ec9a1bc21fb0a7e1f29898, including its earlier readers and research programmes. Start with the separately available 168-page PDF, which is also the record’s preview; its complete LaTeX source is supplied alongside it. The current first-passage package contains six cumulative proof chapters, 33 finite checking scripts and their labelled local reports, figures, and five narrowly scoped Lean source modules with recorded axiom outputs. All 33 finite suites passed in the local workbench after typesetting repairs; two self-contained arithmetic checks were also rerun against the distributed package. Finite tests do not prove the analytic density theorem; the Lean modules certify discrete clock, witness, reflection, dyadic-prefix and completed-height statements, not the whole analytic argument. Historical readers and their separate verification records retain their dates and scope; inclusion in this archive is not a fresh verification of every historical result. The GitHub reading guide provides routes through the collection. Collaboration and reuse The public collaboration name is Kokuno Yumeto. Work was developed with ChatGPT 5.6 Sol and GPT-6 Astra, in Ultra mode in Codex. Literature authors retain their own attribution. To contribute, fork the GitHub repository, let your model read the relevant proofs and primary sources, commit its arguments and reproducible checks, and submit a pull request or link your fork in the PolyClank discussion. Contributions should explain the mathematics clearly enough for another mathematician to follow. This edition continues the Zenodo series formerly titled Affine Packets, Cyclic Orbit Packets, Dyadic Dilation, Lambert–Mahler Series, Hydra/Numen Structure, and Exact Rational Periodic Orbit Enumeration in the Odd-Step Collatz Algebra. Earlier editions remain available in the version history. This is ongoing research, not a completed proof of the Collatz conjecture.

This article analyzes the state of introduction of artificial intelligence technologies in the production infrastructure of enterprises of Navoi region. The study highlights the impact of artificial i…

This article analyzes the state of introduction of artificial intelligence technologies in the production infrastructure of enterprises of Navoi region. The study highlights the impact of artificial intelligence use on productivity, automation levels, rational resource utilization, and product quality. The proposals also include developing digital infrastructure in enterprises, widely introducing modern technologies, training qualified specialists, and increasing the efficiency of using artificial intelligence solutions, as well as scientifically substantiating existing problems and promising development directions.