Aleksandros Sobczyk @asobczyk.bsky.social · 15/02/2026I vibe-coded two variants of the Red-Blue Pebble Game using deepseek, as a fun (and/or educational) browser game. Link to play: asobczyk.github.io/pebble-games/ Slightly more details on LinkedIN www.linkedin.com/posts/aleksa...linkedin.comHoi Zäme 👋. The last few weekends I delved into Red-Blue Pebble Games 🎮 (vibe-)coding. I implemented two minimal web-browser demos for two variants: 🔵 The original one, by Hong & Kung (STOC'81...Hoi Zäme 👋. The last few weekends I delved into Red-Blue Pebble Games 🎮 (vibe-)coding. I implemented two minimal web-browser demos for two variants: 🔵 The original one, by Hong & Kung (STOC'81) 🔴 Our ... 000
Aleksandros Sobczyk @asobczyk.bsky.social · 19/12/2025We have a new preprint with Anastasios Zouzias on prefix sum (scan) circuits. More details and link to the paper on linkedin: www.linkedin.com/posts/aleksa...linkedin.comWe have a new preprint with Anastasios Zouzias on prefix sums... and quantum adders (⁉️). Few months ago, Anastasios Zouzias pinged me with excitement about a Kronecker formula that he discovered to…...We have a new preprint with Anastasios Zouzias on prefix sums... and quantum adders (⁉️). Few months ago, Anastasios Zouzias pinged me with excitement about a Kronecker formula that he discovered to ... 000
Aleksandros Sobczyk @asobczyk.bsky.social · 14/11/2025@jannisblrn.bsky.social wrote a very nice teaser about our Neurips paper Quantum Doubly Stochastic Transformers (spotlight). Our co-authors Filip and Kahn will present it in San Diego, and Jannis in EurIPS. You can find links to the paper, video, and poster below: www.linkedin.com/posts/jannis...linkedin.com#transformers #neurips #eurips #ibmresearch | Jannis Born𝗡𝗲𝘂𝗿𝗜𝗣𝗦 𝘀𝗽𝗼𝘁𝗹𝗶𝗴𝗵𝘁 for our work on "𝗤𝘂𝗮𝗻𝘁𝘂𝗺 𝗗𝗼𝘂𝗯𝗹𝘆 𝗦𝘁𝗼𝗰𝗵𝗮𝘀𝘁𝗶𝗰 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗲𝗿𝘀" 🔦 𝘊𝘢𝘯 𝘱𝘳𝘪𝘯𝘤𝘪𝘱𝘭𝘦𝘴 𝘧𝘳𝘰𝘮 𝘲𝘶𝘢𝘯𝘵𝘶𝘮 𝘤𝘰𝘮𝘱𝘶𝘵𝘪𝘯𝘨 𝘣𝘦 𝘣𝘭𝘦𝘯𝘥𝘦𝘥 𝘪𝘯𝘵𝘰 𝘵𝘩𝘦 𝘮𝘰𝘴𝘵 𝘱𝘰𝘸𝘦𝘳𝘧𝘶𝘭 𝘔𝘓 𝘮𝘰𝘥𝘦𝘭𝘴? 🤔 𝗧𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝗶𝗻 #𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗲𝗿𝘀: Transf... 041
Reposted by Aleksandros SobczykKasper Green Larsen @kasperglarsen.bsky.social · 21/10/2025CS at Aarhus University is hiring up to six professors of any rank and area! Come join my section and do cutting-edge research in TCS, database systems and/or ML/AI, both from a theory and applied side. international.au.dk/about/profil... The application deadline is January 5th, 2026.international.au.dkAarhus University is hiring Assistant, Associate and Full Professors for the Department of Computer Science - Vacancy at Aarhus UniversityVacancy at Computer Science, Dept. of, Aarhus University 063
Reposted by Aleksandros SobczykKimon Fountoulakis @kfountou.bsky.social · 18/10/2025On the Statistical Query Complexity of Learning Semiautomata: a Random Walk Approach Link to the paper: arxiv.org/abs/2510.04115 161
Reposted by Aleksandros SobczykEthan Epperly @ethanepperly.bsky.social · 03/09/2025New paper out with Chris Camaño, Raphael Meyer, and Joel Tropp re-examining sketching algorithms! Included: subspace injections as an alternative to subspace embeddings, the theory and practice of sparse sketching, tensor sketching, and much more! arxiv.org/abs/2508.21189arxiv.orgFaster Linear Algebra Algorithms with Structured Random MatricesTo achieve the greatest possible speed, practitioners regularly implement randomized algorithms for low-rank approximation and least-squares regression with structured dimension reduction maps. Despit... 121
Aleksandros Sobczyk @asobczyk.bsky.social · 31/08/2025We have a new preprint with Almudena Carrera Vazquez, one of the last (and favorite) works that I contributed to before leaving IBM. Arxiv link: arxiv.org/abs/2508.21002 The paper is about quantum algorithms for approximating so-called "spectral gaps" between consecutive pairs of eigenvalues.arxiv.orgSpectral Gaps with Quantum Counting Queries and Oblivious State PreparationApproximating the $k$-th spectral gap $Δ_k=|λ_k-λ_{k+1}|$ and the corresponding midpoint $μ_k=\frac{λ_k+λ_{k+1}}{2}$ of an $N\times N$ Hermitian matrix with eigenvalues $λ_1\geqλ_2\geq\ldots\geqλ_N$, ... 010
Aleksandros Sobczyk @asobczyk.bsky.social · 06/07/2025In 1) we study a model to analyze algorithms for AI accelerators, using small matrix multiplications as primitive operations. In 2) we provide an in-depth analysis of a Red-Blue Pebble Game extension, allowing partial computations (see also Albert-Jan's post www.linkedin.com/posts/anyzel... )linkedin.comThe Red-Blue Pebble Game by Hong and Kung is the canonical model for determining an algorithm's I/O complexity-- how many times data must be read or written to complete its computations. | Albert-...The Red-Blue Pebble Game by Hong and Kung is the canonical model for determining an algorithm's I/O complexity-- how many times data must be read or written to complete its computations. A computa... 010
Aleksandros Sobczyk @asobczyk.bsky.social · 06/07/2025We have couple of new works to share: 1) arxiv.org/abs/2506.23906 with Giuseppe Sorentino & Anastasios Zouzias, 2) arxiv.org/abs/2506.10854 with Pál András Papp & Albert-Jan Yzelman, to appear in SPAA '25. From tomorrow I'll be in Aarhus for ICALP until 11/7, ping me if you want to chat!arxiv.orgSegmented Operations using Matrix MultiplicationsSpecialized computational units that perform small matrix multiplications as primitive operations are typically present in modern accelerators. However, these units are often underutilized for many fu... 110
Aleksandros Sobczyk @asobczyk.bsky.social · 19/06/2025Anyone have an idea how accurate are the Google scholar metrics? The following highly-cited paper seems to be missing from the metrics of *both* STOC and JACM: dl.acm.org/doi/abs/10.1...dl.acm.orgLow-Rank Approximation and Regression in Input Sparsity Time | Journal of the ACMWe design a new distribution over m × n matrices S so that, for any fixed n × d matrix A of rank r, with probability at least 9/10, ∥SAx∥2 = (1 ± ε)∥Ax∥2 simultaneously for all x ∈ Rd. Here, m is boun... 000
Reposted by Aleksandros SobczykKimon Fountoulakis @kfountou.bsky.social · 26/05/2025New paper: Learning to Add, Multiply, and Execute Algorithmic Instructions Exactly with Neural Networks 151
Reposted by Aleksandros SobczykKasper Green Larsen @kasperglarsen.bsky.social · 30/04/2025I have an opening for a post doc position in machine learning theory, with a deadline of June 1st. Please share and apply: international.au.dk/about/profil...international.au.dkPost doc position in theory of machine learning at Department of Computer Science, Aarhus University - Vacancy at Aarhus UniversityVacancy at Computer Science, Dept. of, Aarhus University 0129
Aleksandros Sobczyk @asobczyk.bsky.social · 14/04/2025With my first Bluesky post, I am very pleased to share that my last PhD paper "Deterministic complexity analysis of Hermitian eigenproblems" has been accepted in ICALP 2025. A preprint is available on Arxiv: arxiv.org/abs/2410.21550 A bit more info on linkedin: www.linkedin.com/posts/aleksa...arxiv.orgDeterministic complexity analysis of Hermitian eigenproblemsIn this work we revisit the arithmetic and bit complexity of Hermitian eigenproblems. We first provide an analysis for the divide-and-conquer tridiagonal eigensolver of Gu and Eisenstat [GE95] in the ... 042
Aleksandros Sobczyk @asobczyk.bsky.social · 09/04/2025Cool work Philippe and the team! Was curious what is the baseline software comparison with DFT, is it quantum espresso? (Sorry if i missed it!) Happy to chat offline if any of you guys are up 120
Reposted by Aleksandros SobczykDaniel Probst @skepteis.bsky.social · 14/03/2025We're offering a fully funded PhD at the intersection of ML/AI and the natural sciences with a focus on sustainability and chemistry. You'll work at WUR in the Netherlands, ranked #3 in environ. sciences, #1 in agricultural science, #38 in life sciences (QS). Apply here: www.wur.nl/nl/vacature/... 03325
Aleksandros Sobczyk @asobczyk.bsky.social · 25/02/2025The team of Albert-Jan Yzelman has postdoc openings in Zurich, in case it is interesting! www.linkedin.com/posts/anyzel...linkedin.comAlbert-Jan Yzelman on LinkedIn: High-Performance Optimisation (Postdoctoral Researcher) - Huawei Research…Passionate about enabling the use of massive compute power via simple high-level abstractions? We are hiring three Postdoctoral Researchers in High Performance… 010
