Tootfinder

Boosting Large-scale Parallel Training Efficiency with C4: A Communication-Driven Approach
Jianbo Dong, Bin Luo, Jun Zhang, Pengcheng Zhang, Fei Feng, Yikai Zhu, Ang Liu, Zian Chen, Yi Shi, Hairong Jiao, Gang Lu, Yu Guan, Ennan Zhai, Wencong Xiao, Hanyu Zhao, Man Yuan, Siran Yang, Xiang Li, Jiamang Wang, Rui Men, Jianwei Zhang, Huang Zhong, Dennis Cai, Yuan Xie, Binzhang Fu

Boosting Large-scale Parallel Training Efficiency with C4: A Communication-Driven Approach
The emergence of Large Language Models (LLMs) has necessitated the adoption of parallel training techniques, involving the deployment of thousands of GPUs to train a single model. Unfortunately, we have found that the efficiency of current parallel training is often suboptimal, largely due to the following two main issues. Firstly, hardware failures are inevitable, leading to interruptions in the training tasks. The inability to quickly identify the faulty components results in a substantial wa…

Stochastic full waveform inversion with deep generative prior for uncertainty quantification
Yuke Xie, Herv\'e Chauris, Nicolas Desassis
https://arxiv.org/abs/2406.04859

Stochastic full waveform inversion with deep generative prior for uncertainty quantification
To obtain high-resolution images of subsurface structures from seismic data, seismic imaging techniques such as Full Waveform Inversion (FWI) serve as crucial tools. However, FWI involves solving a nonlinear and often non-unique inverse problem, presenting challenges such as local minima trapping and inadequate handling of inherent uncertainties. In addressing these challenges, we propose leveraging deep generative models as the prior distribution of geophysical parameters for stochastic Bayesi…

Someone stumbled over this over on X.com/Twitter.
Yes, that is me from earlier in the year.
https://x.com/knutsi/status/1758813454381588664

Knut Åbjørsbråten (@knutsi) on X
@Smoothedan Slår ikke han her med enhjulssykkel på holka på Ring 2 i Oslo

Wait, are these the same investor shareholders that said yes to giving the moron $56BB dollars?
If I was a Tesla investor, I would not be a Tesla investor.
https://arstechnica.com/tech-policy/2024/06/tesla-i…

The Nodal Sets of Solutions to Parabolic Equations
Yiqi Huang, Wenshuai Jiang
https://arxiv.org/abs/2406.05877 https://arxiv.org/pdf/…

The Nodal Sets of Solutions to Parabolic Equations
In this paper, we study the parabolic equations $\partial_t u=\partial_j\left(a^{ij}(x,t)\partial_iu\right)+b^j(x,t)\partial_ju+c(x,t)u$ in a domain of $\mathbb{R}^n$ under the condition that $a^{ij}$ are Lipschitz continuous. Consider the nodal set $Z_t=\{x: u(x,t)=0\}$ at a time $t$-slice. Simple examples show that the singular set $\mathcal{S}_t=\{x: u(x,t)=|\nabla_x u|(x,t)=0\}$ may coincide with nodal set. This makes the methods used in the study of nodal sets for elliptic equati…

MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning
Hanqing Wang, Zeguan Xiao, Yixia Li, Shuo Wang, Guanhua Chen, Yun Chen
https://arxiv.org/abs/2406.09044

MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning
Efficient finetuning of large language models (LLMs) aims to adapt the LLMs with reduced computation and memory cost. Previous LoRA-based approaches initialize the low-rank matrices with gaussian distribution and zero values, while keeping the original weight matrices frozen. However, the trainable model parameters optimized in an unguided subspace might have interference with the well-learned subspace of the pretrained weight matrix. In this paper, we propose MiLoRA, a simple yet effective LLM…

The Nodal Sets of Solutions to Parabolic Equations
Yiqi Huang, Wenshuai Jiang
https://arxiv.org/abs/2406.05877 https://arxiv.org/pdf/…

The Nodal Sets of Solutions to Parabolic Equations
In this paper, we study the parabolic equations $\partial_t u=\partial_j\left(a^{ij}(x,t)\partial_iu\right)+b^j(x,t)\partial_ju+c(x,t)u$ in a domain of $\mathbb{R}^n$ under the condition that $a^{ij}$ are Lipschitz continuous. Consider the nodal set $Z_t=\{x: u(x,t)=0\}$ at a time $t$-slice. Simple examples show that the singular set $\mathcal{S}_t=\{x: u(x,t)=|\nabla_x u|(x,t)=0\}$ may coincide with nodal set. This makes the methods used in the study of nodal sets for elliptic equati…

Equidistribution of Kloosterman sums over function fields
Lei Fu, Yuk-Kam Lau, Wen-Ching Winnie Li, Ping Xi
https://arxiv.org/abs/2406.10106 https://

Equidistribution of Kloosterman sums over function fields
We prove the Sato--Tate distribution of Kloosterman sums over function fields with explicit error terms, when the places vary in arithmetic progressions or short intervals. A joint Sato--Tate distribution of two ``different" exponential sums is also proved.

Spin excitations in Nd1-xSrxNiO2 and YBa2Cu3O7-delta: the influence of Hubbard U
Francesco Rosa, Leonardo Martinelli, Guillaume Krieger, Lucio Braicovich, Nicholas B. Brookes, Giacomo Merzoni, Marco Moretti Sala, Flora Yakhou-Harris, Riccardo Arpaia, Daniele Preziosi, Marco Salluzzo, Maciej Fidrysiak, Giacomo Ghiringhelli
https://

Spin excitations in Nd1-xSrxNiO2 and YBa2Cu3O7-delta: the influence of Hubbard U
We use Resonant Inelastic X-ray Scattering (RIXS) to compare the doping dependence of magnetic excitations of an Infinite-Layer nickelate to those of a prototypical superconducting cuprate. The polarization analysis of RIXS spectra establishes the dominant spin-flip nature of the mid-infrared peak in both cases. Hole doping leads to opposite behavior of the magnetic energy in the two materials. By fitting the data with an original Hubbard-based model for dynamic susceptibility, we find that t i…

Feasibility of the observation of $\eta^{\prime}$ mesic nuclei in the semi-exclusive $^{12}$C($p, dp$) reaction
Natsumi Ikeno, Yuko Higashi, Hiroyuki Fujioka, Kenta Itahashi, Ryohei Sekiya, Yoshiki K. Tanaka, Junko Yamagata-Sekihara, Volker Metag, Mariana Nanova, Satoru Hirenzaki
https://arxiv.org/abs/2406.06058

Feasibility of the observation of $η^{\prime}$ mesic nuclei in the semi-exclusive $^{12}$C($p, dp$) reaction
We study theoretically the feasibility of the semi-exclusive $^{12}$C($p,dp$)$X$ reaction for the observation of $η^\prime$ mesic nuclei using the transport model JAM. The semi-exclusive measurements of the ($p,d$) reaction with protons from $η^\prime$ non-mesonic two-body absorption ($η^\prime NN \to NN$) are found to be critically important for the observation of the $η^\prime$ bound states. The Green's function method is used to calculate the momentum spectrum of forward going deuterons …