University of Warsaw

About: University of Warsaw is a education organization based out in Warsaw, Poland. It is known for research contribution in the topics: Population & Large Hadron Collider. The organization has 20832 authors who have published 56617 publications receiving 1185084 citations. The organization is also known as: Uniwersytet Warszawski & Warsaw University.

How Good is the Bayes Posterior in Deep Neural Networks Really

Abstract: During the past five years the Bayesian deep learning community has developed increasingly accurate and efficient approximate inference procedures that allow for Bayesian inference in deep neural networks. However, despite this algorithmic progress and the promise of improved uncertainty quantification and sample efficiency there are—as of early 2020—no publicized deployments of Bayesian neural networks in industrial practice. In this work we cast doubt on the current understanding of Bayes posteriors in popular deep neural networks: we demonstrate through careful MCMC sampling that the posterior predictive induced by the Bayes posterior yields systematically worse predictions compared to simpler methods including point estimates obtained from SGD. Furthermore, we demonstrate that predictive performance is improved significantly through the use of a “cold posterior” that overcounts evidence. Such cold posteriors sharply deviate from the Bayesian paradigm but are commonly used as heuristic in Bayesian deep learning papers. We put forward several hypotheses that could explain cold posteriors and evaluate the hypotheses through experiments. Our work questions the goal of accurate posterior approximations in Bayesian deep learning: If the true Bayes posterior is poor, what is the use of more accurate approximations? Instead, we argue that it is timely to focus on understanding the origin of the improved performance of cold posteriors. Code available on GitHub.

Cell Wall Hydrolases in Bacteria: Insight on the Diversity of Cell Wall Amidases, Glycosidases and Peptidases Toward Peptidoglycan.

Abstract: The cell wall (CW) of bacteria is an intricate arrangement of macromolecules, at least constituted of peptidoglycan (PG) but also of (lipo)teichoic acids, various polysaccharides, polyglutamate and/or proteins. During bacterial growth and division, there is a constant balance between CW degradation and biosynthesis. The CW is remodeled by bacterial hydrolases, whose activities are carefully regulated to maintain cell integrity or lead to bacterial death. Each cell wall hydrolase (CWH) has a specific role regarding the PG: (i) cell wall amidase (CWA) cleaves the amide bond between N-acetylmuramic acid and L-alanine residue at the N-terminal of the stem peptide, (ii) cell wall glycosidase (CWG) catalyses the hydrolysis of the glycosidic linkages, whereas (iii) cell wall peptidase (CWP) cleaves amide bonds between amino acids within the PG chain. After an exhaustive overview of all known conserved catalytic domains responsible for CWA, CWG, and CWP activities, this review stresses that the CWHs frequently display a modular architecture combining multiple and/or different catalytic domains, including some lytic transglycosylases as well as CW binding domains. From there, direct physiological and collateral roles of CWHs in bacterial cells are further discussed.

Derivation and testing of pair potentials for protein folding. When is the quasichemical approximation correct

Abstract: Many existing derivations of knowledge-based statistical pair potentials invoke the quasichemical approximation to estimate the expected side-chain contact frequency if there were no amino acid pair-specific interactions. At first glance, the quasichemical approximation that treats the residues in a protein as being disconnected and expresses the side-chain contact probability as being proportional to the product of the mole fractions of the pair of residues would appear to be rather severe. To investigate the validity of this approximation, we introduce two new reference states in which no specific pair interactions between amino acids are allowed, but in which the connectivity of the protein chain is retained. The first estimates the expected number of side-chain contacts by treating the protein as a Gaussian random coil polymer. The second, more realistic reference state includes the effects of chain connectivity, secondary structure, and chain compactness by estimating the expected side-chain contact probability by placing the sequence of interest in each member of a library of structures of comparable compactness to the native conformation. The side-chain contact maps are not allowed to readjust to the sequence of interest, i.e., the side chains cannot repack. This situation would hold rigorously if all amino acids were the same size. Both reference states effectively permit the factorization of the side-chain contact probability into sequence-dependent and structure-dependent terms. Then, because the sequence distribution of amino acids in proteins is random, the quasichemical approximation to each of these reference states is shown to be excellent. Thus, the range of validity of the quasichemical approximation is determined by the magnitude of the side-chain repacking term, which is, at present, unknown. Finally, the performance of these two sets of pair interaction potentials as well as side-chain contact fraction-based interaction scales is assessed by inverse folding tests both without and with allowing for gaps.

Measurement of jet fragmentation in PbPb and pp collisions at root s(NN)=2.76 TeV

Abstract: The jet fragmentation function of inclusive jets with transverse momentum p_T above 100GeV/c in PbPb collisions has been measured using reconstructed charged particles with p_T above 1GeV/c in a cone of radius 0.3 around the jet axis. A data sample of PbPb collisions collected in 2011 at a nucleon-nucleon center-of-mass energy of √sNN = 2.76 TeV corresponding to an integrated luminosity of 150μb^(−1) is used. The results for PbPb collisions as a function of collision centrality and jet transverse momentum are compared to reference distributions based on pp data collected at the same center-of-mass energy in 2013, with an integrated luminosity of 5.3pb^(−1). A centrality-dependent modification of the fragmentation function is found. For the most central collisions, a significant enhancement is observed in the PbPb/pp fragmentation function ratio for charged particles with p_T less than 3GeV/c. This enhancement is observed for all jet p_T bins studied.

Binary planetary nebulae nuclei towards the Galactic bulge - I. Sample discovery, period distribution, and binary fraction

Abstract: Binarity has been hypothesised to play an important, if not ubiquitous, role in the formation of planetary nebulae (PNe). Yet there remains a severe paucity of known binary central stars required to test the binary hypothesis and to place strong constraints on the physics of the common-envelope (CE) phase of binary stellar evolution. Large photometric surveys offer an unrivalled opportunity to efficiently discover many binary central stars. We have combined photometry from the OGLE microlensing survey with the largest sample of PNe towards the Galactic bulge to systematically search for new binaries. A total of 21 periodic binaries were found thereby more than doubling the known sample. The orbital period distribution was found to be best described by CE population synthesis models when no correlation between primary and secondary masses is assumed for the initial mass ratio distribution. A comparison with post-CE white dwarf binaries indicates both distributions are representative of the true post-CE period distribution with most binaries exhibiting periods less than one day. A close binary fraction of 12-21% is derived and is the first robust and independent validation of the previous 10-15% estimate. This suggests that binarity is not a precondition for the formation of PNe and that close binaries do not play a dominant role in the shaping of nebular morphologies. Systematic effects and biases of the survey are discussed with implications for future photometric surveys.