Dense retrieval has shown great success in passage ranking in English. However, its effectiveness in document retrieval for non-English languages remains unexplored due to the limitation in training resources. In this work, we explore different trans
fer techniques for document ranking from English annotations to multiple non-English languages. Our experiments on the test collections in six languages (Chinese, Arabic, French, Hindi, Bengali, Spanish) from diverse language families reveal that zero-shot model-based transfer using mBERT improves the search quality in non-English mono-lingual retrieval. Also, we find that weakly-supervised target language transfer yields competitive performances against the generation-based target language transfer that requires external translators and query generators.
Baikal-GVD is a km$^3$-scale neutrino telescope being constructed in Lake Baikal. Muon and partially tau (anti)neutrino interactions near the detector through the W$^{pm}$-boson exchange are accompanied by muon tracks. Reconstructed direction of the
track is arguably the most precise probe of the neutrino direction attainable in Cerenkov neutrino telescopes. Muon reconstruction techniques adopted by Baikal-GVD are discussed in the present report. Performance of the muon reconstruction is studied using realistic Monte Carlo simulation of the detector. The algorithms are applied to real data from Baikal-GVD and the results are compared with simulations. The performance of the neutrino selection based on a boosted decision tree classifier is discussed.
The Baikal Gigaton Volume Detector (Baikal-GVD) is a km$^3$-scale neutrino detector currently under construction in Lake Baikal, Russia. The detector currently consists of 2304 optical modules arranged on 64 vertical strings. Further extension of the
array is planned for March 2022. The data from the partially complete array have been analyzed using a $chi^2$-based track reconstruction algorithm. After suppression of the downward-going atmospheric muon background, a flux of upward-going neutrino events is observed, dominated by the atmospheric neutrinos. The observed flux is in good agreement with Monte Carlo predictions.
The Baikal-GVD neutrino telescope currently consists of 8 clusters of 288 optical modules (photodetectors). One cluster comprises 8 strings, each of which is subdivided into 3 sections of 12 optical modules. This paper presents the methods of time sy
nchronization between the different GVD components (optical modules, sections, clusters) and estimations of time synchronization accuracy.
We report transport measurements on Josephson junctions consisting of Bi2Te3 topological insulator (TI) thin films contacted by superconducting Nb electrodes. For a device with junction length L = 134 nm, the critical supercurrent Ic can be modulated
by an electrical gate which tunes the carrier type and density of the TI film. Ic can reach a minimum when the TI is near the charge neutrality regime with the Fermi energy lying close to the Dirac point of the surface state. In the p-type regime the Josephson current can be well described by a short ballistic junction model. In the n-type regime the junction is ballistic at 0.7 K < T < 3.8 K while for T < 0.7 K the diffusive bulk modes emerge and contribute a larger Ic than the ballistic model. We attribute the lack of diffusive bulk modes in the p-type regime to the formation of p-n junctions. Our work provides new clues for search of Majorana zero mode in TI-based superconducting devices.
Neutrino astronomy offers a novel view of the non-thermal Universe and is complementary to other astronomical disciplines. The field has seen rapid progress in recent years, including the first detection of astrophysical neutrinos in the TeV-PeV ener
gy range by IceCube and the first identified extragalactic neutrino source (TXS 0506+056). Further discoveries are aimed for with new cubic-kilometer telescopes in the Northern Hemisphere: Baikal-GVD, in Lake Baikal, and KM3NeT-ARCA, in the Mediterranean sea. The construction of Baikal-GVD proceeds as planned; the detector currently includes over 2000 optical modules arranged on 56 strings, providing an effective volume of 0.35 km$^3$. We review the scientific case for Baikal-GVD, the construction plan, and first results from the partially built array.
The semantics of a text is manifested not only by what is read, but also by what is not read. In this article, we will study how the implicit not read information such as end-of-paragraph (eop) and end-of-sequence (eos) affect the quality of text gen
eration. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the eop in the fine-tuning stage. Experimental results on English story generation show that eop can lead to higher BLEU score and lower eos perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without eop or eos during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with eop.
We demonstrate the effective use of randomized methods for linear algebra to perform network-based analysis of complex vortical flows. Network theoretic approaches can reveal the connectivity structures among a set of vortical elements and analyze th
eir collective dynamics. These approaches have recently been generalized to analyze high-dimensional turbulent flows, for which network computations can become prohibitively expensive. In this work, we propose efficient methods to approximate network quantities, such as the leading eigendecomposition of the adjacency matrix, using randomized methods. Specifically, we use the Nystrom method to approximate the leading eigenvalues and eigenvectors, achieving significant computational savings and reduced memory requirements. The effectiveness of the proposed technique is demonstrated on two high-dimensional flow fields: two-dimensional flow past an airfoil and two-dimensional turbulence. We find that quasi-uniform column sampling outperforms uniform column sampling, while both feature the same computational complexity.
We present the results of a new analysis of data taken in 1998-2002 for a search for high-energy extraterrestrial neutrinos. The analysis is based on a full reconstruction of high-energy cascade parameters: vertex coordinates, energy and arrival dire
ction. Upper limits on the diffuse fluxes of all neutrino flavors, predicted by several models of AGN-like neutrino sources are derived. For an ${bf E^{-2}}$ behavior of the neutrino spectrum, our limit is ${bf E^{2} F_{ u}(E) < 2.9 times 10^{-7}}$ cm${bf ^{-2}}$ s${bf ^{-1}}$ sr${bf ^{-1}}$ GeV over a neutrino energy range ${bf 2 times 10^4 div 2 times 10^7}$ GeV. This limit is by a factor of 2.8 more stringent than a limit obtained with a previous analysis.
A prototype string for the future km3-scale Baikal neutrino telescope has been deployed in April, 2008 and is fully integrated into the NT200+ telescope. All basic string elements - optical modules (with 12/13 hemispherical photomultipliers), 200MHz
FADC readout and calibration system - have been redesigned following experience with NT200+. First results of in-situ operation of this prototype string are presented.
