Tootfinder

Subspace-constrained randomized coordinate descent for linear systems with good low-rank matrix approximations
Jackie Lok, Elizaveta Rebrova
https://arxiv.org/abs/2506.09394

Subspace-constrained randomized coordinate descent for linear systems with good low-rank matrix approximations
The randomized coordinate descent (RCD) method is a classical algorithm with simple, lightweight iterations that is widely used for various optimization problems, including the solution of positive semidefinite linear systems. As a linear solver, RCD is particularly effective when the matrix is well-conditioned; however, its convergence rate deteriorates rapidly in the presence of large spectral outliers. In this paper, we introduce the subspace-constrained randomized coordinate descent (SC-RCD…

Was speaking to some journalists recently who didn't realise how long a history the use of radar has in measuring ice thickness over glaciers.
The technique was discovered by pilots, who must have had the frankly terrifying experience of landing on the ice sheet with their radars telling them they still had a few thousand metres of descent to reach the surface...
It's not mentioned in this obituary but I wonder if this guy would have seen it? Perhaps he didn't even have radar to assist in landing?
https://flipboard.com/@newyorktimes/obituaries-bq1u22g8z/-/a-gd4LiLY4RbOMXavSDWOBQQ:a:3195393-/0
newyorktimes@flipboard.com - Conrad Shinn, First Pilot to Land at the South Pole, Dies at 102
https://www.nytimes.com/2025/06/08/obituaries/conrad-shinn-dead.html?utm_source=flipboard&utm_medium=activitypub
Posted into Obituaries @obituaries-newyorktimes

This https://arxiv.org/abs/2409.08469 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_mat…

Improved Finite-Particle Convergence Rates for Stein Variational Gradient Descent
We provide finite-particle convergence rates for the Stein Variational Gradient Descent (SVGD) algorithm in the Kernelized Stein Discrepancy ($\mathsf{KSD}$) and Wasserstein-2 metrics. Our key insight is that the time derivative of the relative entropy between the joint density of $N$ particle locations and the $N$-fold product target measure, starting from a regular initial distribution, splits into a dominant `negative part' proportional to $N$ times the expected $\mathsf{KSD}^2$ and a smalle…

Just finished "Statistically Speaking" by Debbie Johnson. It's coincidentally the second book dealing with adoption that I've just finished, though I suspect in both cases not #OwnVoices, which I also suspect matters somewhat. I was well-absorbed and enjoyed it immensely, but was left again with the reservation that I'm sure it may reflect only that small facet of real life which is pleasing and/or tolerable to a wide audience, and may thus in its own way make things more difficult for those whose realities it does not reflect. I find myself very glad to have also recently read Mama by Nikkya Hargrove, which is autobiographical and which as a result of having more real-world complexity drives its similar point about found family home with more force, to me (to be fair, Johnson's work has a decent amount of real life complexity, for a novel).
#AmReading

Achieving Linear Speedup and Near-Optimal Complexity for Decentralized Optimization over Row-stochastic Networks
Liyuan Liang, Xinyi Chen, Gan Luo, Kun Yuan
https://arxiv.org/abs/2506.04600

Achieving Linear Speedup and Near-Optimal Complexity for Decentralized Optimization over Row-stochastic Networks
A key challenge in decentralized optimization is determining the optimal convergence rate and designing algorithms to achieve it. While this problem has been extensively addressed for doubly-stochastic and column-stochastic mixing matrices, the row-stochastic scenario remains unexplored. This paper bridges this gap by introducing effective metrics to capture the influence of row-stochastic mixing matrices and establishing the first convergence lower bound for decentralized learning over row-sto…

This https://arxiv.org/abs/2503.09532 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csLG_…

SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability
Sparse autoencoders (SAEs) are a popular technique for interpreting language model activations, and there is extensive recent work on improving SAE effectiveness. However, most prior work evaluates progress using unsupervised proxy metrics with unclear practical relevance. We introduce SAEBench, a comprehensive evaluation suite that measures SAE performance across eight diverse metrics, spanning interpretability, feature disentanglement and practical applications like unlearning. To enable syst…

Characterizing Multi-Hunk Patches: Divergence, Proximity, and LLM Repair Challenges
Noor Nashid, Daniel Ding, Keheliya Gallaba, Ahmed E. Hassan, Ali Mesbah
https://arxiv.org/abs/2506.04418

Characterizing Multi-Hunk Patches: Divergence, Proximity, and LLM Repair Challenges
Multi-hunk bugs, where fixes span disjoint regions of code, are common in practice, yet remain underrepresented in automated repair. Existing techniques and benchmarks pre-dominantly target single-hunk scenarios, overlooking the added complexity of coordinating semantically related changes across the codebase. In this work, we characterize HUNK4J, a dataset of multi-hunk patches derived from 372 real-world defects. We propose hunk divergence, a metric that quantifies the variation among edits i…

This https://arxiv.org/abs/2505.03768 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCR_…

From Concept to Measurement: A Survey of How the Blockchain Trilemma Can Be Analyzed
To meet non-functional requirements, practitioners must identify Pareto-optimal configurations of the degree of decentralization, scalability, and security of blockchain systems. Maximizing all of these subconcepts is, however, impossible due to the trade-offs highlighted by the blockchain trilemma. We reviewed analysis approaches to identify constructs and their operationalization through metrics for analyzing the blockchain trilemma subconcepts and to assess the applicability of the operation…

It’s not the mistakes that matter. Apple has made them before. The Newton, MobileMe, the butterfly keyboard. What matters is the posture. A company once defined by joyful provocation—by thinking different—is now defined by its defensiveness. Its leadership acts not like inventors but like stewards of a status quo. They protect margins like relics. They fear dilution. They optimize at the expense of surprise.

Apple’s Diet of Worms
On a crisp April morning in 1521, the twenty-one-year-old Emperor Charles V entered the Bishop’s Palace in Worms. The weight of a continent rested on his adolescent shoulders, and he carried it with the stiff dignity of someone raised to believe history itself had chosen him. He wore black

Linear Discriminant Analysis with Gradient Optimization on Covariance Inverse
Cencheng Shen, Yuexiao Dong
https://arxiv.org/abs/2506.06845 https://

Linear Discriminant Analysis with Gradient Optimization on Covariance Inverse
Linear discriminant analysis (LDA) is a fundamental method in statistical pattern recognition and classification, achieving Bayes optimality under Gaussian assumptions. However, it is well-known that classical LDA may struggle in high-dimensional settings due to instability in covariance estimation. In this work, we propose LDA with gradient optimization (LDA-GO), a new approach that directly optimizes the inverse covariance matrix via gradient descent. The algorithm parametrizes the inverse co…