SearcharxivSearch

arXiv subjects

Yuto Kojima

Publications and source records attributed to Yuto Kojima.

3 recordsLinked to original sources

Time-of-flight force sensing below the quantum zero-point fluctuation

Sensing weak forces through observing a mechanical motion near or below its quantum zero-point fluctuation has been desired in diverse areas. While mechanical oscillators have played a crucial role in such studies, their application to free-fall-type sensing has been elusive, in particular in the quantum regime. Here, we demonstrate sensing a static force of the order of 10 zeptonewtons with a levitated nanomechanical oscillator below the zero-point fluctuation through the rapid modulation of its confining potential. We prepare a squeezed state with a reduced velocity uncertainty by abruptly decreasing the potential. Subsequently, we detect the exerted static force through time-of-flight measurements, where we release the nanoparticle from the potential and measure the displacement during a free fall. Furthermore, time-of-flight measurements allow us to perform quantum state tomography of the squeezed state, from which we reconstruct its Wigner quasiprobability distribution and evaluate the Fisher information for the position measurement to quantify the achievable force sensitivity of our protocol. Our results demonstrate that modulating the trap stiffness serves as a crucial technique for quantum-limited force sensing and paves the way to utilize a levitated nanoparticle as a promising sensing platform beyond the quantum limit with a capability of quantum state tomography.

quant-ph

LoRA-TTT: Low-Rank Test-Time Training for Vision-Language Models

The rapid advancements in vision-language models (VLMs), such as CLIP, have intensified the need to address distribution shifts between training and testing datasets. Although prior Test-Time Training (TTT) techniques for VLMs have demonstrated robust performance, they predominantly rely on tuning text prompts, a process that demands substantial computational resources and is heavily dependent on entropy-based loss. In this paper, we propose LoRA-TTT, a novel TTT method that leverages Low-Rank Adaptation (LoRA), applied exclusively to the image encoder of VLMs. By introducing LoRA and updating only its parameters during test time, our method offers a simple yet effective TTT approach, retaining the model's initial generalization capability while achieving substantial performance gains with minimal memory and runtime overhead. Additionally, we introduce a highly efficient reconstruction loss tailored for TTT. Our method can adapt to diverse domains by combining these two losses, without increasing memory consumption or runtime. Extensive experiments on two benchmarks, covering 15 datasets, demonstrate that our method improves the zero-shot top-1 accuracy of CLIP-ViT-B/16 by an average of 5.79% on the OOD benchmark and 1.36% on the fine-grained benchmark, efficiently surpassing test-time prompt tuning, without relying on any external models or cache.

cs.CV

Luminosity Function of Faint Sporadic Meteors measured with a Wide-Field CMOS mosaic camera Tomo-e PM

Imaging observations of faint meteors were carried out on April 11 and 14, 2016 with a wide-field CMOS mosaic camera, Tomo-e PM, mounted on the 105-cm Schmidt telescope at Kiso Observatory, the University of Tokyo. Tomo-e PM, which is a prototype model of Tomo-e Gozen, can monitor a sky of ${\sim}1.98\,\mathrm{deg^2}$ at 2\,Hz. The numbers of detected meteors are 1514 and 706 on April 11 and 14, respectively. The detected meteors are attributed to sporadic meteors. Their absolute magnitudes range from $+4$ to $+10\,\mathrm{mag}$ in the $V$-band, corresponding to about $8.3{\times}10^{-2}$ to $3.3{\times}10^{-4}\,\mathrm{g}$ in mass. The present magnitude distributions we obtained are well explained by a single power-law luminosity function with a slope parameter $r = 3.1{\pm}0.4$ and a meteor rate $\log_{10}N_0 = -5.5{\pm}0.5$. The results demonstrate a high performance of telescopic observations with a wide-field video camera to constrain the luminosity function of faint meteors. The performance of Tomo-e Gozen is about two times higher than that of Tomo-e PM. A survey with Tomo-e Gozen will provide a more robust measurement of the luminosity function.

astro-ph.EP