Tootfinder

Uncertainty Prioritized Experience Replay
Rodrigo Carrasco-Davis, Sebastian Lee, Claudia Clopath, Will Dabney
https://arxiv.org/abs/2506.09270 https://

Uncertainty Prioritized Experience Replay
Prioritized experience replay, which improves sample efficiency by selecting relevant transitions to update parameter estimates, is a crucial component of contemporary value-based deep reinforcement learning models. Typically, transitions are prioritized based on their temporal difference error. However, this approach is prone to favoring noisy transitions, even when the value estimation closely approximates the target mean. This phenomenon resembles the noisy TV problem postulated in the explo…

I have published an update to the 'Add Arguments to Google Searches' Firefox extension which permits appending the &udm=14 parameter to the results URL, fully eliminating all AI slop as well as any other hint of machine 'helpfulness' in Google results. The extension will also now install and run on both Firefox and Firefox for Android.

Adaptive Lattice-based Motion Planning
Abhishek Dhar, Sarthak Mishra, Spandan Roy, Daniel Axehill
https://arxiv.org/abs/2508.02350 https://arxiv.org/pdf/25…

Adaptive Lattice-based Motion Planning
This paper proposes an adaptive lattice-based motion planning solution to address the problem of generating feasible trajectories for systems, represented by a linearly parameterizable non-linear model operating within a cluttered environment. The system model is considered to have uncertain model parameters. The key idea here is to utilize input/output data online to update the model set containing the uncertain system parameter, as well as a dynamic estimated parameter of the model, so that t…

Initial curve25519 accelerator refactoring update: added new MULT_AREA_OPT parameter.
0 = existing implementation (best for 7 series)
1 = resource sharing between constant and variable multipliers, but still does 32 16x32 multiplies per clock. Much slower on 7 series due to DSP cascade, but hits the same 71 MHz Fmax on Trion.
Fabric usage on Trion is slightly higher (9528 -> 10613 LE) but multiplier block usage is down from 96 to 66.
Next step will be trying to fig…

Fully Bayesian Sequential Design for Mean Response Surface Prediction of Heteroscedastic Stochastic Simulations
Yuying Huang, Samuel W. K. Wong
https://arxiv.org/abs/2506.09722

Fully Bayesian Sequential Design for Mean Response Surface Prediction of Heteroscedastic Stochastic Simulations
We present a fully Bayesian sequential strategy for predicting the mean response surface of heteroscedastic stochastic simulation functions. Leveraging dual Gaussian processes as the surrogate model and a criterion based on empirical expected integrated mean-square prediction error, our approach sequentially selects informative design points while fully accounting for parameter uncertainty. Sequential importance sampling is employed to efficiently update the posterior distribution of the parame…

Dimension and model reduction approaches for linear Bayesian inverse problems with rank-deficient prior covariances
Josie K\"onig, Elizabeth Qian, Melina A. Freitag
https://arxiv.org/abs/2506.23892

Dimension and model reduction approaches for linear Bayesian inverse problems with rank-deficient prior covariances
Bayesian inverse problems use observed data to update a prior probability distribution for an unknown state or parameter of a scientific system to a posterior distribution conditioned on the data. In many applications, the unknown parameter is high-dimensional, making computation of the posterior expensive due to the need to sample in a high-dimensional space and the need to evaluate an expensive high-dimensional forward model relating the unknown parameter to the data. However, inverse problem…

Investigating the convergence properties of iterative ptychography for atomic-resolution low-dose imaging
Tamazouzt Chennit, Songge Li, Hoelen L. Lalandec Robert, Christoph Hofer, Nadine J. Schrenker, Liberato Manna, Sara Bals, Timothy J. Pennycook, Jo Verbeeck
https://arxiv.org/abs/2507.06756

Investigating the convergence properties of iterative ptychography for atomic-resolution low-dose imaging
This study investigates the convergence properties of a collection of iterative electron ptychography methods, under low electron doses ($<$ 10$^3$ $e^-/A^2$) and gives particular attention to the impact of the user-defined update strengths. We demonstrate that carefully chosen values for this parameter, ideally smaller than those conventionally met in the literature, are essential for achieving accurate reconstructions of the projected electrostatic potential. Using a 4D dataset of a thin hybr…

MLorc: Momentum Low-rank Compression for Large Language Model Adaptation
Wei Shen, Yaxiang Zhang, Minhui Huang, Mengfan Xu, Jiawei Zhang, Cong Shen
https://arxiv.org/abs/2506.01897

MLorc: Momentum Low-rank Compression for Large Language Model Adaptation
With increasing size of large language models (LLMs), full-parameter fine-tuning imposes substantial memory demands. To alleviate this, we propose a novel memory-efficient training paradigm called Momentum Low-rank compression (MLorc). By directly compressing and reconstructing momentum rather than gradients, MLorc avoids imposing a fixed-rank constraint on weight update matrices and better preserves the training dynamics of full-parameter fine-tuning, in contrast to existing low-rank approache…

Computationally efficient Gauss-Newton reinforcement learning for model predictive control
Dean Brandner, Sebastien Gros, Sergio Lucia
https://arxiv.org/abs/2508.02441 https://

Computationally efficient Gauss-Newton reinforcement learning for model predictive control
Model predictive control (MPC) is widely used in process control due to its interpretability and ability to handle constraints. As a parametric policy in reinforcement learning (RL), MPC offers strong initial performance and low data requirements compared to black-box policies like neural networks. However, most RL methods rely on first-order updates, which scale well to large parameter spaces but converge at most linearly, making them inefficient when each policy update requires solving an opt…

This https://arxiv.org/abs/2505.03834 has been replaced.
initial toot: https://mastoxiv.page/@a…

Update: Progress toward fusion energy breakeven and gain as measured against the Lawson criteria
This paper is an update to our earlier paper ''Progress toward fusion energy breakeven and gain as measured against the Lawson criterion'' [Phys. Plasmas 29, 062103 (2022)]. Plots of Lawson parameter and triple product vs. ion temperature and triple product vs. date achieved are updated with recently published experimental results. A new plot of scientific energy gain vs. date achieved is included. Additionally, notes on new experimental results, clarifications, and a correction are included.

The Hyper-Kamiokande experiment: input to the update of the European Strategy for Particle Physics
Kamiokande Collaboration
https://arxiv.org/abs/2506.16641

The Hyper-Kamiokande experiment: input to the update of the European Strategy for Particle Physics
Hyper-Kamiokande is a large infrastructure for particle and astroparticle physics being built in Japan and aiming to start operations by the end of 2027 whose objective is to address the most important questions in science today, for instance how the universe began and evolved. It aims to measure with the highest precision the leptonic Charge-Parity violation parameter using both a Mega-Watt intense neutrino beam and high-statistics atmospheric neutrino samples. The combination of these sampl…

This https://arxiv.org/abs/2412.17629 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csNE_…

Learn from Global Correlations: Enhancing Evolutionary Algorithm via Spectral GNN
Evolutionary algorithms (EAs) simulate natural selection but have two main limitations: (1) they rarely update individuals based on global correlations, limiting comprehensive learning; (2) they struggle with balancing exploration and exploitation, where excessive exploitation causes premature convergence, and excessive exploration slows down the search. Moreover, EAs often depend on manual parameter settings, which can disrupt the exploration-exploitation balance.To address these issues, we pr…

Inexact Levenberg-Marquardt methods under H\"{o}lder metric subregularity
Bas Symoens, Morteza Rahimi, Masoud Ahookhosh
https://arxiv.org/abs/2507.16461

Inexact Levenberg-Marquardt methods under Hölder metric subregularity
This paper investigates two inexact Levenberg-Marquardt (LM) methods for solving systems of nonlinear equations. Both approaches compute approximate search directions by solving the LM linear system inexactly, subject to specific residual-based conditions. The first method uses an adaptive scheme to update the LM parameter, and we establish its local superlinear convergence under Hölder metric subregularity and local Hölder continuity of the gradient. The second method combines an inexact LM …

This https://arxiv.org/abs/2412.17629 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csNE_…

Learn from Global Correlations: Enhancing Evolutionary Algorithm via Spectral GNN
Evolutionary algorithms (EAs) simulate natural selection but have two main limitations: (1) they rarely update individuals based on global correlations, limiting comprehensive learning; (2) they struggle with balancing exploration and exploitation, where excessive exploitation causes premature convergence, and excessive exploration slows down the search. Moreover, EAs often depend on manual parameter settings, which can disrupt the exploration-exploitation balance.To address these issues, we pr…

Reinforcement Learning Fine-Tunes a Sparse Subnetwork in Large Language Models
Andrii Balashov
https://arxiv.org/abs/2507.17107 https://

Reinforcement Learning Fine-Tunes a Sparse Subnetwork in Large Language Models
Reinforcement learning (RL) is a key post-pretraining step for aligning large language models (LLMs) with complex tasks and human preferences. While it is often assumed that RL fine-tuning requires updating most of a model's parameters, we challenge this assumption with a surprising finding: RL fine-tuning consistently modifies only a small subnetwork (typically 5-30% of weights), leaving most parameters unchanged. We call this phenomenon RL-induced parameter update sparsity. It arises naturall…