-
Learning to Generate Instruction Tuning Datasets for Zero-Shot Task Adaptation
Authors:
Nihal V. Nayak,
Yiyang Nan,
Avi Trost,
Stephen H. Bach
Abstract:
We introduce Bonito, an open-source model for conditional task generation that converts unannotated text into task-specific training datasets for instruction tuning. We aim to enable zero-shot task adaptation of large language models on users' specialized, private data. We train Bonito by fine-tuning a pretrained large language model on a new large-scale dataset with 1.65M examples created by remi…
▽ More
We introduce Bonito, an open-source model for conditional task generation that converts unannotated text into task-specific training datasets for instruction tuning. We aim to enable zero-shot task adaptation of large language models on users' specialized, private data. We train Bonito by fine-tuning a pretrained large language model on a new large-scale dataset with 1.65M examples created by remixing existing instruction tuning datasets into meta-templates. The meta-templates for a dataset produce training examples where the input is the unannotated text and the task attribute and the output consists of the instruction and the response. We use Bonito to generate synthetic tasks for seven datasets from specialized domains with unannotated text across three task types -- yes-no question answering, extractive question answering, and natural language inference -- and adapt language models. We show that Bonito significantly improves the average performance of pretrained and instruction tuned models over the de facto self supervised baseline. For example, adapting Mistral-Instruct-v2 and instruction tuned variants of Mistral and Llama2 with Bonito improves the strong zero-shot performance by 22.1 F1 points whereas the next word prediction objective undoes some of the benefits of instruction tuning and reduces the average performance by 0.8 F1 points. We conduct additional experiments with Bonito to understand the effects of the domain, the size of the training set, and the choice of alternative synthetic task generators. Overall, we show that learning with synthetic instruction tuning datasets is an effective way to adapt language models to new domains. The model, dataset, and code are available at https://github.com/BatsResearch/bonito.
△ Less
Submitted 6 June, 2024; v1 submitted 28 February, 2024;
originally announced February 2024.
-
Probing the small scale structure of the Inter-Galactic Medium with ESPRESSO: spectroscopy of the lensed QSO UM673
Authors:
Stefano Cristiani,
Guido Cupani,
Andrea Trost,
Valentina D'Odorico,
Francesco Guarneri,
Gaspare Lo Curto,
Massimo Meneghetti,
Paolo Di Marcantonio,
João P. Faria,
Jonay I. González Hernández,
Christophe Lovis,
Carlos J. A. P. Martins,
Dinko Milaković,
Paolo Molaro,
Michael T. Murphy,
Nelson J. Nunes,
Francesco Pepe,
Rafael Rebolo,
Nuno C. Santos,
Tobias M. Schmidt,
Sérgio G. Sousa,
Alessandro Sozzetti,
María Rosa Zapatero Osorio
Abstract:
The gravitationally lensed quasar J014516.6-094517 at z=2.719 has been observed with the ESPRESSO instrument at the ESO VLT to obtain high-fidelity spectra of the two images A and B with a resolving power R=70000. At the redshifts under investigation (2.1 < z < 2.7), the Lyman forests along the two sightlines are separated by sub-kiloparsec physical distances and exhibit a strong correlation. We f…
▽ More
The gravitationally lensed quasar J014516.6-094517 at z=2.719 has been observed with the ESPRESSO instrument at the ESO VLT to obtain high-fidelity spectra of the two images A and B with a resolving power R=70000. At the redshifts under investigation (2.1 < z < 2.7), the Lyman forests along the two sightlines are separated by sub-kiloparsec physical distances and exhibit a strong correlation. We find that the two forests are indistinguishable at the present level of signal-to-noise ratio and do not show any global velocity shift, with the cross-correlation peaking at $Δv = 12 \pm 48$ m/s. The distribution of the difference in velocity of individual Lyman-$α$ features is compatible with a null average and a mean absolute deviation of 930 m/s. Significant differences in NHI column density are not detected, putting a limit to the RMS fluctuation in the baryon density on $\leq 1$ proper kpc scales of $Δρ/ ρ< 3$%. On the other hand, metal lines show significant differences both in velocity structure and in column density. A toy model shows that the difference in velocity of the metal features between the two sightlines is compatible with the the motions of the baryonic component associated to dark matter halos of typical mass $M\simeq 2\times 10^{10} M_\odot$, also compatible with the observed incidence of the metal systems. The present observations confirm the feasibility of the Sandage test of the cosmic redshift drift with high-fidelity spectroscopy of the Lyman forest of distant, bright quasars, but also provide an element of caution about the intrinsic noise associated to the usage of metal features for the same purpose.
△ Less
Submitted 8 February, 2024;
originally announced February 2024.
-
Fundamental physics with ESPRESSO: a new determination of the D/H ratio towards PKS1937-101
Authors:
Francesco Guarneri,
Luca Pasquini,
Valentina D'Odorico,
Stefano Cristiani,
Guido Cupani,
Paolo Di Marcantonio,
J. I. González Hernández,
C. J. A. P. Martins,
Alejandro Suárez Mascareño,
Dinko Milaković,
Paolo Molaro,
Michael T. Murphy,
Nelson J. Nunes,
Enric Palle,
Francesco Pepe,
Rafael Rebolo,
Nuno C. Santos,
Ricardo Génova Santos,
Tobias M. Schmidt,
Sérgio G. Sousa,
Alessandro Sozzetti,
Andrea Trost
Abstract:
Primordial abundances of light elements are sensitive to the physics of the early Universe and can directly constrain cosmological quantities, such as the baryon-to-photon ratio $η_{10}$, the baryon density and the number of neutrino families. Deuterium is especially suited for these studies: its primordial abundance is sensitive and monotonically dependent on $η_{10}$, allowing an independent mea…
▽ More
Primordial abundances of light elements are sensitive to the physics of the early Universe and can directly constrain cosmological quantities, such as the baryon-to-photon ratio $η_{10}$, the baryon density and the number of neutrino families. Deuterium is especially suited for these studies: its primordial abundance is sensitive and monotonically dependent on $η_{10}$, allowing an independent measurement of the cosmic baryon density that can be compared, for instance, against the Planck satellite data. The primordial deuterium abundance can be measured in high $H_I$ column density absorption systems towards distant quasars. We report here a new measurement, based on high-resolution ESPRESSO data, of the primordial $D_I$ abundance of a system at redshift $z \sim 3.572$, towards PKS1937-101. Using only ESPRESSO data, we find a D/H ratio of $2.638\pm0.128 \times 10^{-5}$, while including the available UVES data improves the precision, leading to a ratio of $2.608 \pm 0.102 \times 10^{-5}$. The results of this analysis agree with those of the most precise existing measurements. We find that the relatively low column density of this system ($\log{N_{\rm H_I}/ {\rm cm}^{-2}}\sim18 $) introduces modelling uncertainties, which become the main contributor to the error budget.
△ Less
Submitted 8 February, 2024;
originally announced February 2024.
-
Stability, bounded generation and strong boundedness
Authors:
Alexander Trost
Abstract:
We provide bounds linear in the rank for the generalized conjugacy diameters, introduced by Kedra, Libman and Martin, for the special linear and symplectic groups defined over the rings of integers of global fields by way of using certain stability considerations familiar from classical algebraic K-theory. This determines the growth rates of these generalized conjugacy diameters.
We provide bounds linear in the rank for the generalized conjugacy diameters, introduced by Kedra, Libman and Martin, for the special linear and symplectic groups defined over the rings of integers of global fields by way of using certain stability considerations familiar from classical algebraic K-theory. This determines the growth rates of these generalized conjugacy diameters.
△ Less
Submitted 13 October, 2023; v1 submitted 19 May, 2023;
originally announced May 2023.
-
Traveling Wave Method for Event Localization and Characterization of Power Transmission Lines
Authors:
Marko Hudomalj,
Andrej Trost,
Andrej Čampa
Abstract:
Traveling wave theory is deployed today to improve the monitoring of transmission lines in electrical power grids. Most traveling wave methods require prior knowledge of the wave propagation of the transmission line, which is a major source of error as the value changes during the operation of the line. To improve the localization of events on transmission lines, we propose a new online localizati…
▽ More
Traveling wave theory is deployed today to improve the monitoring of transmission lines in electrical power grids. Most traveling wave methods require prior knowledge of the wave propagation of the transmission line, which is a major source of error as the value changes during the operation of the line. To improve the localization of events on transmission lines, we propose a new online localization method that simultaneously determines the frequency-dependent wave propagation characteristic from the traveling wave measurements of the event. Compared to conventional methods, this is achieved with one additional traveling wave measurement, but the method can still be applied in different measurement setups. We have derived the method based on the complex continuous wavelet transform. The accuracy of the method is evaluated in a simulation with a frequency-dependent transmission line model of the IEEE 39-bus system. The method was developed independently of the type of event and evaluated in test setups considering different lengths of the monitored line, line types and event locations. The localization accuracy is compared with existing online methods and analyzed with regard to the characterization capabilities. The method has a high relative localization accuracy in the range of 0.1\,\% under different test conditions.
△ Less
Submitted 29 November, 2023; v1 submitted 4 April, 2023;
originally announced April 2023.
-
Spectroscopy of QUBRICS quasar candidates: 1672 new redshifts and a Golden Sample for the Sandage Test of the Redshift Drift
Authors:
Stefano Cristiani,
Matteo Porru,
Francesco Guarneri,
Giorgio Calderone,
Konstantina Boutsia,
Andrea Grazian,
Guido Cupani,
Valentina D'Odorico,
Fabio Fontanot,
Carlos J. A. P. Martins,
Catarina M. J. Marques,
Soumak Maitra,
Andrea Trost
Abstract:
The QUBRICS (QUasars as BRIght beacons for Cosmology in the Southern hemisphere) survey aims at constructing a sample of the brightest quasars with z>~2.5, observable with facilities in the Southern Hemisphere. QUBRICS makes use of the available optical and IR wide-field surveys in the South and of Machine Learning techniques to produce thousands of bright quasar candidates of which only a few hun…
▽ More
The QUBRICS (QUasars as BRIght beacons for Cosmology in the Southern hemisphere) survey aims at constructing a sample of the brightest quasars with z>~2.5, observable with facilities in the Southern Hemisphere. QUBRICS makes use of the available optical and IR wide-field surveys in the South and of Machine Learning techniques to produce thousands of bright quasar candidates of which only a few hundred have been confirmed with follow-up spectroscopy. Taking advantage of the recent Gaia Data Release 3, which contains 220 million low-resolution spectra, and of a newly developed spectral energy distribution fitting technique, designed to combine the photometric information with the Gaia spectroscopy, it has been possible to measure 1672 new secure redshifts of QUBRICS candidates, with a typical uncertainty $σ_z = 0.02$. This significant progress of QUBRICS brings it closer to (one of) its primary goals: providing a sample of bright quasars at redshift 2.5 < z < 5 to perform the Sandage test of the cosmological redshift drift. A Golden Sample of seven quasars is presented that makes it possible to carry out this experiment in about 1500 hours of observation in 25 years, using the ANDES spectrograph at the 39m ELT, a significant improvement with respect to previous estimates.
△ Less
Submitted 1 April, 2023;
originally announced April 2023.
-
CUBES: a UV spectrograph for the future
Authors:
S. Covino,
S. Cristiani,
J. M. Alcala',
S. H. P. Alencar,
S. A. Balashev,
B. Barbuy,
N. Bastian,
U. Battino,
L. Bissell,
P. Bristow,
A. Calcines,
G. Calderone,
P. Cambianica,
R. Carini,
B. Carter,
S. Cassisi,
B. V. Castilho,
G. Cescutti,
N. Christlieb,
R. Cirami,
R. Conzelmann,
I. Coretti,
R. Cooke,
G. Cremonese,
K. Cunha
, et al. (64 additional authors not shown)
Abstract:
In spite of the advent of extremely large telescopes in the UV/optical/NIR range, the current generation of 8-10m facilities is likely to remain competitive at ground-UV wavelengths for the foreseeable future. The Cassegrain U-Band Efficient Spectrograph (CUBES) has been designed to provide high-efficiency (>40%) observations in the near UV (305-400 nm requirement, 300-420 nm goal) at a spectral r…
▽ More
In spite of the advent of extremely large telescopes in the UV/optical/NIR range, the current generation of 8-10m facilities is likely to remain competitive at ground-UV wavelengths for the foreseeable future. The Cassegrain U-Band Efficient Spectrograph (CUBES) has been designed to provide high-efficiency (>40%) observations in the near UV (305-400 nm requirement, 300-420 nm goal) at a spectral resolving power of R>20,000, although a lower-resolution, sky-limited mode of R ~ 7,000 is also planned.
CUBES will offer new possibilities in many fields of astrophysics, providing access to key lines of stellar spectra: a tremendous diversity of iron-peak and heavy elements, lighter elements (in particular Beryllium) and light-element molecules (CO, CN, OH), as well as Balmer lines and the Balmer jump (particularly important for young stellar objects). The UV range is also critical in extragalactic studies: the circumgalactic medium of distant galaxies, the contribution of different types of sources to the cosmic UV background, the measurement of H2 and primordial Deuterium in a regime of relatively transparent intergalactic medium, and follow-up of explosive transients.
The CUBES project completed a Phase A conceptual design in June 2021 and has now entered the Phase B dedicated to detailed design and construction. First science operations are planned for 2028. In this paper, we briefly describe the CUBES project development and goals, the main science cases, the instrument design and the project organization and management.
△ Less
Submitted 24 December, 2022;
originally announced December 2022.
-
The CUBES Science Case
Authors:
Chris Evans,
Stefano Cristiani,
Cyrielle Opitom,
Gabriele Cescutti,
Valentina D'Odorico,
Juan Manuel Alcalá,
Silvia H. P. Alencar,
Sergei Balashev,
Beatriz Barbuy,
Nate Bastian,
Umberto Battino,
Pamela Cambianica,
Roberta Carini,
Brad Carter,
Santi Cassisi,
Bruno Vaz Castilho,
Norbert Christlieb,
Ryan Cooke,
Stefano Covino,
Gabriele Cremonese,
Katia Cunha,
André R. da Silva,
Valerio D'Elia,
Annalisa De Cia,
Gayandhi De Silva
, et al. (29 additional authors not shown)
Abstract:
We introduce the scientific motivations for the development of the Cassegrain U-Band Efficient Spectrograph (CUBES) that is now in construction for the Very Large Telescope. The assembled cases span a broad range of contemporary topics across Solar System, Galactic and extragalactic astronomy, where observations are limited by the performance of current ground-based spectrographs shortwards of 400…
▽ More
We introduce the scientific motivations for the development of the Cassegrain U-Band Efficient Spectrograph (CUBES) that is now in construction for the Very Large Telescope. The assembled cases span a broad range of contemporary topics across Solar System, Galactic and extragalactic astronomy, where observations are limited by the performance of current ground-based spectrographs shortwards of 400nm. A brief background to each case is presented and specific technical requirements on the instrument design that flow-down from each case are identified. These were used as inputs to the CUBES design, that will provide a factor of ten gain in efficiency for astronomical spectroscopy over 300-405nm, at resolving powers of R~24,000 and ~7,000. We include performance estimates that demonstrate the ability of CUBES to observe sources that are up to three magnitudes fainter than currently possible at ground-ultraviolet wavelengths, and we place its predicted performance in the context of existing facillities.
△ Less
Submitted 30 September, 2022; v1 submitted 2 August, 2022;
originally announced August 2022.
-
Elementary bounded generation for ${\rm SL}_n$ for global function fields and $n\geq 3$
Authors:
Alexander Alois Trost
Abstract:
This paper shows that the group ${\rm SL}_n(R)$ is boundedly elementary generated for $n\geq 3$ and $R$ the ring of algebraic integers in a global function field. Contrary to previous statements for number fields and earlier statements for global function fields, the bounds proven in this preprint for elementary bounded generation are independent of the underlying global function field and only de…
▽ More
This paper shows that the group ${\rm SL}_n(R)$ is boundedly elementary generated for $n\geq 3$ and $R$ the ring of algebraic integers in a global function field. Contrary to previous statements for number fields and earlier statements for global function fields, the bounds proven in this preprint for elementary bounded generation are independent of the underlying global function field and only depend on the integer $n.$ Combining our main result with earlier results, we further establish that elementary bounded generation always has bounds independent from the global field in question, only depending on $n.$
△ Less
Submitted 28 June, 2022;
originally announced June 2022.
-
The CUBES Instrument Model and Simulation Tools. Their role in the project Phase A study
Authors:
Matteo Genoni,
Marco Landoni,
Guido Cupani,
Mariagrazia Franchini,
Roberto Cirami,
Alessio Zanutta,
Chris Evans,
Paolo Di Marcantonio,
Stefano Cristiani,
Andrea Trost,
Sonia Zorba
Abstract:
We present the simulation tools developed to aid the design phase of the Cassegrain U-Band Efficient Spectrograph (CUBES) for the Very Large Telescope (VLT), exploring aspects of the system design and evaluating the performance for different design configurations. CUBES aims to be the 'ultimate' ultraviolet (UV) instrument at the European Southern Observatory (ESO) in terms of throughput, with the…
▽ More
We present the simulation tools developed to aid the design phase of the Cassegrain U-Band Efficient Spectrograph (CUBES) for the Very Large Telescope (VLT), exploring aspects of the system design and evaluating the performance for different design configurations. CUBES aims to be the 'ultimate' ultraviolet (UV) instrument at the European Southern Observatory (ESO) in terms of throughput, with the goal to cover the bluest part of the spectrum accessible from the ground (300 nm to 400 nm) with the highest possible efficiency. Here we introduce the End-to-End (E2E) and the Exposure Time Calculator (ETC) tools. The E2E simulator has been developed with different versions to meet the needs of different users, including a version that can be accessed for use by the broader scientific community using a Jupyter notebook. The E2E tool was used by the system team to help define the Phase A baseline design of the instrument, as well as in scientific evaluation of a possible low-resolution mode. The ETC is a web-based tool through which the science community are able to test a range of science cases for CUBES, demonstrating its potential to push the limiting magnitude for the detection of specific UV-features, such as abundance estimates of beryllium in main sequence stars.
△ Less
Submitted 29 March, 2022;
originally announced March 2022.
-
CUBES Phase A design overview -- The Cassegrain U-Band Efficient Spectrograph for the Very Large Telescope
Authors:
Alessio Zanutta,
Stefano Cristiani,
David Atkinson,
Veronica Baldini,
Andrea Balestra,
Beatriz Barbuy,
Vanessa Bawden P. Macanhan,
Ariadna Calcines,
Giorgio Calderone,
Scott Case,
Bruno V. Castilho,
Gabriele Cescutti,
Roberto Cirami,
Igor Coretti,
Stefano Covino,
Guido Cupani,
Vincenzo De Caprio,
Hans Dekker,
Paolo Di Marcantonio,
Valentina D'Odorico,
Heitor Ernandes,
Chris Evans,
Tobias Feger,
Carmen Feiz,
Mariagrazia Franchini
, et al. (29 additional authors not shown)
Abstract:
We present the baseline conceptual design of the Cassegrain U-Band Efficient Spectrograph (CUBES) for the Very Large Telescope. CUBES will provide unprecedented sensitivity for spectroscopy on a 8 - 10 m class telescope in the ground ultraviolet (UV), spanning a bandwidth of > 100 nm that starts at 300 nm, the shortest wavelength accessible from the ground. The design has been optimized for end-to…
▽ More
We present the baseline conceptual design of the Cassegrain U-Band Efficient Spectrograph (CUBES) for the Very Large Telescope. CUBES will provide unprecedented sensitivity for spectroscopy on a 8 - 10 m class telescope in the ground ultraviolet (UV), spanning a bandwidth of > 100 nm that starts at 300 nm, the shortest wavelength accessible from the ground. The design has been optimized for end-to-end efficiency and provides a spectral resolving power of R > 20000, that will unlock a broad range of new topics across solar system, Galactic and extraglactic astronomy. The design also features a second, lower-resolution (R \sim 7000) mode and has the option of a fiberlink to the UVES instrument for simultaneous observations at longer wavelengths. Here we present the optical, mechanical and software design of the various subsystems of the instrument after the Phase A study of the project. We discuss the expected performances for the layout choices and highlight some of the performance trade-offs considered to best meet the instrument top-level requirements. We also introduce the model-based system engineering approach used to organize and manage the project activities and interfaces, in the context that it is increasingly necessary to integrate such tools in the development of complex astronomical projects.
△ Less
Submitted 29 March, 2022;
originally announced March 2022.
-
The dichotomy property of ${\rm SL}_2(R)$-A short note
Authors:
Alexander Alois Trost
Abstract:
A recent paper by Polterovich, Shalom and Shem-Tov has shown that non-discrete, conjugation invariant norms on arithmetic Chevalley groups of higher rank give rise to very restricted topologies. Namely, such topologies always have profinite norm-completions. In this note, we sketch an argument showing that this also holds for ${\rm SL}_2(R)$ for $R$ a ring of algebraic integers with infinitely man…
▽ More
A recent paper by Polterovich, Shalom and Shem-Tov has shown that non-discrete, conjugation invariant norms on arithmetic Chevalley groups of higher rank give rise to very restricted topologies. Namely, such topologies always have profinite norm-completions. In this note, we sketch an argument showing that this also holds for ${\rm SL}_2(R)$ for $R$ a ring of algebraic integers with infinitely many units.
△ Less
Submitted 16 February, 2022;
originally announced February 2022.
-
Comparative Verification of the Digital Library of Mathematical Functions and Computer Algebra Systems
Authors:
André Greiner-Petter,
Howard S. Cohl,
Abdou Youssef,
Moritz Schubotz,
Avi Trost,
Rajen Dey,
Akiko Aizawa,
Bela Gipp
Abstract:
Digital mathematical libraries assemble the knowledge of years of mathematical research. Numerous disciplines (e.g., physics, engineering, pure and applied mathematics) rely heavily on compendia gathered findings. Likewise, modern research applications rely more and more on computational solutions, which are often calculated and verified by computer algebra systems. Hence, the correctness, accurac…
▽ More
Digital mathematical libraries assemble the knowledge of years of mathematical research. Numerous disciplines (e.g., physics, engineering, pure and applied mathematics) rely heavily on compendia gathered findings. Likewise, modern research applications rely more and more on computational solutions, which are often calculated and verified by computer algebra systems. Hence, the correctness, accuracy, and reliability of both digital mathematical libraries and computer algebra systems is a crucial attribute for modern research.
In this paper, we present a novel approach to verify a digital mathematical library and two computer algebra systems with one another by converting mathematical expressions from one system to the other. We use our previously eveloped conversion tool (referred to as LaCASt) to translate formulae from the NIST Digital Library of Mathematical Functions to the computer algebra systems Maple and Mathematica. The contributions of our presented work are as follows: (1) we present the most comprehensive verification of computer algebra systems and digital mathematical libraries with one another; (2) we significantly enhance the performance of the underlying translator in terms of coverage and accuracy; and (3) we provide open access to translations for Maple and Mathematica of the formulae in the NIST Digital Library of Mathematical Functions.
△ Less
Submitted 31 March, 2022; v1 submitted 24 January, 2022;
originally announced January 2022.
-
Bounded generation by root elements for Chevalley groups defined over rings of integers of function fields with an application in strong boundedness
Authors:
Alexander A. Trost
Abstract:
Bounded generation by root elements is a property which has been widely studied for various types of linear algebraic groups defined over rings of integers in algebraic number fields. However, when considering global function fields, there are not many results beyond the treatment of special cases due to Nica and Queen. In this paper, we use model theoretic methods due to Carter, Keller and Paige…
▽ More
Bounded generation by root elements is a property which has been widely studied for various types of linear algebraic groups defined over rings of integers in algebraic number fields. However, when considering global function fields, there are not many results beyond the treatment of special cases due to Nica and Queen. In this paper, we use model theoretic methods due to Carter, Keller and Paige written up by Morris to prove bounded generation by root elements for simply connected, split Chevalley groups defined over the ring of all integers in a global function field. We further apply this bounded generation result together with results from a previous paper by the author to derive that the aforementioned Chevalley groups satisfy the strong boundedness property introduced by Kedra, Libman and Martin.
△ Less
Submitted 27 August, 2021;
originally announced August 2021.
-
Strong boundedness of ${\rm SL}_2(R)$ for rings of S-algebraic integers with infinitely many units
Authors:
Alexander Alois Trost
Abstract:
A group is called strongly bounded, if the speed with which it is generated by finitely many conjugacy classes has a positive, lower bound only dependent on the number of the conjugacy classes in question rather than the actual conjugacy classes. Earlier papers by Kedra, Libman and Martin and myself have shown that this is a property common to split Chevalley groups defined using an irreducible ro…
▽ More
A group is called strongly bounded, if the speed with which it is generated by finitely many conjugacy classes has a positive, lower bound only dependent on the number of the conjugacy classes in question rather than the actual conjugacy classes. Earlier papers by Kedra, Libman and Martin and myself have shown that this is a property common to split Chevalley groups defined using an irreducible root system of rank at least $2$ and the ring of all S-algebraic integers and that the situation is dependent on the number theory of $R$ for ${\rm Sp}_4$ and $G_2.$ In this paper, we will show that ${\rm SL}_2(R)$ is also strongly bounded for $R$ the ring of all S-algebraic integers in a number field $K$ with $R$ having infinitely many units and will give a complete account of the existence of small conjugacy classes generating ${\rm SL}_2(R)$ in terms of the prime factorization of the rational primes $2$ and $3$ in $R.$
△ Less
Submitted 17 July, 2021; v1 submitted 23 May, 2021;
originally announced May 2021.
-
Bounded generation for congruence subgroups of ${\rm Sp}_4(R)$
Authors:
Alexander Alois Trost
Abstract:
This paper describes a bounded generation result concerning the minimal natural number $K$ such that for $Q(C_2,2R):=\{A\varepsilon_φ(2x)A^{-1}|x\in R,A\in{\rm Sp}_4(R),φ\in C_2\}$, one has $N_{C_2,2R}=\{X_1\cdots X_K|\forall 1\leq i\leq K:X_i\in Q(C_2,2R)\}$ for rings of algebraic integers $R$ and the principal congruence subgroup $N_{C_2,2R}$ in ${\rm Sp}_4(R).$ This gives an explicit version of…
▽ More
This paper describes a bounded generation result concerning the minimal natural number $K$ such that for $Q(C_2,2R):=\{A\varepsilon_φ(2x)A^{-1}|x\in R,A\in{\rm Sp}_4(R),φ\in C_2\}$, one has $N_{C_2,2R}=\{X_1\cdots X_K|\forall 1\leq i\leq K:X_i\in Q(C_2,2R)\}$ for rings of algebraic integers $R$ and the principal congruence subgroup $N_{C_2,2R}$ in ${\rm Sp}_4(R).$ This gives an explicit version of an abstract bounded generation result of a similar type as presented by Morris. Furthermore, the result presented does not depend on several number-theoretic quantities unlike Morris' result. Using this bounded generation result, we further give explicit bounds for the strong boundedness of ${\rm Sp}_4(R)$ for certain examples of rings $R,$ thereby giving explicit versions of results in an earlier paper. We further give a classification of normally generating subsets of ${\rm Sp}_4(R)$ for $R$ a ring of algebraic integers.
△ Less
Submitted 6 January, 2021;
originally announced January 2021.
-
Explicit strong boundedness for higher rank symplectic groups
Authors:
Alexander Trost
Abstract:
This paper gives an explicit argument to show strong boundedness for ${\rm Sp}_{2n}(R)$ for $R$ a ring of S-algebraic integers or a semi-local ring. This gives a quantitative version of a related abstract result in a previous paper of the author. The results presented further generalize older results regarding strong boundedness by Kedra, Libman and Martin and Morris from ${\rm SL}_n$ to…
▽ More
This paper gives an explicit argument to show strong boundedness for ${\rm Sp}_{2n}(R)$ for $R$ a ring of S-algebraic integers or a semi-local ring. This gives a quantitative version of a related abstract result in a previous paper of the author. The results presented further generalize older results regarding strong boundedness by Kedra, Libman and Martin and Morris from ${\rm SL}_n$ to ${\rm Sp}_{2n}$. Further, the presented results solve the question of the asymptotic of strong boundedness for ${\rm Sp}_{2n}(R)$ for $R$ semi-local case with an argument that immediately generalizes to all other split Chevalley groups.
△ Less
Submitted 22 December, 2020;
originally announced December 2020.
-
Strong boundedness of simply connected split Chevalley groups defined over rings
Authors:
Alexander Alois Trost
Abstract:
This paper is concerned with the diameter of certain word norms on S-arithmetic split Chevalley groups. Such groups are well known to be boundedly generated by root elements. We prove that word metrics given by conjugacy classes on S-arithmetic split Chevalley groups have an upper bound only depending on the number of conjugacy classes. This property, called strong boundedness, was introduced by K…
▽ More
This paper is concerned with the diameter of certain word norms on S-arithmetic split Chevalley groups. Such groups are well known to be boundedly generated by root elements. We prove that word metrics given by conjugacy classes on S-arithmetic split Chevalley groups have an upper bound only depending on the number of conjugacy classes. This property, called strong boundedness, was introduced by Kedra, Libmann and Martin and proven for ${\rm SL}_n(R)$, assuming R is a principal ideal domain and $n\geq 3$. We also provide examples of normal generating sets for S-arithmetic split Chevalley groups proving our bounds are sharp in an appropriate sense and give a complete account of obstructions to the existence of small normally generating sets of ${\rm Sp}_4(R)$ and $G_2(R)$. For instance, we prove that ${\rm Sp}_4(\mathbb{Z}[\frac{1+\sqrt{-7}}{2}])$ cannot be generated by a single conjugacy class.
△ Less
Submitted 10 April, 2020;
originally announced April 2020.
-
Qualitative counting closed geodesics
Authors:
Bastien Karlhofer,
Jarek Kędra,
Michał Marcinkowski,
Alexander Trost
Abstract:
We investigate the geometry of word metrics on fundamental groups of manifolds associated with the generating sets consisting of elements represented by closed geodesics. We ask whether the diameter of such a metric is finite or infinite. The first answer we interpret as an abundance of closed geodesics, while the second one as their scarcity. We discuss examples for both cases.
We investigate the geometry of word metrics on fundamental groups of manifolds associated with the generating sets consisting of elements represented by closed geodesics. We ask whether the diameter of such a metric is finite or infinite. The first answer we interpret as an abundance of closed geodesics, while the second one as their scarcity. We discuss examples for both cases.
△ Less
Submitted 25 June, 2021; v1 submitted 25 April, 2019;
originally announced April 2019.
-
On the security relevance of weights in deep learning
Authors:
Kathrin Grosse,
Thomas A. Trost,
Marius Mosbach,
Michael Backes,
Dietrich Klakow
Abstract:
Recently, a weight-based attack on stochastic gradient descent inducing overfitting has been proposed. We show that the threat is broader: A task-independent permutation on the initial weights suffices to limit the achieved accuracy to for example 50% on the Fashion MNIST dataset from initially more than $90$%. These findings are confirmed on MNIST and CIFAR. We formally confirm that the attack su…
▽ More
Recently, a weight-based attack on stochastic gradient descent inducing overfitting has been proposed. We show that the threat is broader: A task-independent permutation on the initial weights suffices to limit the achieved accuracy to for example 50% on the Fashion MNIST dataset from initially more than $90$%. These findings are confirmed on MNIST and CIFAR. We formally confirm that the attack succeeds with high likelihood and does not depend on the data. Empirically, weight statistics and loss appear unsuspicious, making it hard to detect the attack if the user is not aware. Our paper is thus a call for action to acknowledge the importance of the initial weights in deep learning.
△ Less
Submitted 29 November, 2020; v1 submitted 8 February, 2019;
originally announced February 2019.
-
Finite index subgroups in Chevalley groups are bounded: an addendum to "On bi-invariant word metrics"
Authors:
Światosław R. Gal,
Jarek Kędra,
Alexander A. Trost
Abstract:
We prove that finite index subgroups in S-arithmetic Chevalley groups are bounded.
We prove that finite index subgroups in S-arithmetic Chevalley groups are bounded.
△ Less
Submitted 14 July, 2019; v1 submitted 20 August, 2018;
originally announced August 2018.
