Search

Frameworks and DSLs auto-generating code have traditionally relied on human experts developing them to have in place rigorous methods to assure the legality of the applied code transformations. Machine Learning (ML) is gaining wider adoption as a means to auto-generate code optimised for the hardware target. However, ML solutions, and in particular black-box DNNs, provide no such guarantees on legality. In this paper we propose a library, Tadashi, which leverages the polyhedral model to empower researchers seeking to curate datasets crucial for applying ML in code-generation. Tadashi provides the ability to reliably and practically check the legality of candidate transformations on polyhedral schedules applied on a baseline reference code. We provide a proof that our library guarantees the legality of generated transformations, and demonstrate its lightweight practical cost. Tadashi is available at https://github.com/vatai/tadashi/., Comment: Submitted to CGO

We construct integrable chiral cosmological models with two scalar fields and potentials represented in terms of hyperbolic functions. Using the conformal transformation of the metric and the corresponding models with induced gravity terms, we obtain the general solutions in the spatially flat, open and closed Friedmann universes and the corresponding integrals of motion. The obtained general solutions can be written in terms of the Jacobi elliptic functions of the conformal time., Comment: 21 pages, v3: typos corrected, to appear in Phys. Rev. D

We consider cosmological models with an arbitrary number of scalar fields nonminimally coupled to gravity and construct new integrable cosmological models. In the constructed models, the Ricci scalar is an integral of motion irrespectively of the type of metric. The general solutions of evolution equations in the spatially flat FLRW metric have been found for models with the quartic potentials., Comment: 10 pages; to be published in Physics of Particles and Nuclei

The size and complexity of software applications is increasing at an accelerating pace. Source code repositories (along with their dependencies) require vast amounts of labor to keep them tested, maintained, and up to date. As the discipline now begins to also incorporate automatically generated programs, automation in testing and tuning is required to keep up with the pace - let alone reduce the present level of complexity. While machine learning has been used to understand and generate code in various contexts, machine learning models themselves are trained almost exclusively on static code without inputs, traces, or other execution time information. This lack of training data limits the ability of these models to understand real-world problems in software. In this work we show that inputs, like code, can be generated automatically at scale. Our generated inputs are stateful, and appear to faithfully reproduce the arbitrary data structures and system calls required to rerun a program function. By building our tool within the compiler, it both can be applied to arbitrary programming languages and architectures and can leverage static analysis and transformations for improved performance. Our approach is able to produce valid inputs, including initial memory states, for 90% of the ComPile dataset modules we explored, for a total of 21.4 million executable functions. Further, we find that a single generated input results in an average block coverage of 37%, whereas guided generation of five inputs improves it to 45%.

We study anisotropic solutions for the pure $R^2$ gravity model with a scalar field in the Bianchi I metric. The evolution equations have a singularity at zero value of the Ricci scalar $R$ for anisotropic solutions, whereas these equations are smooth for isotropic solutions. So, there is no anisotropic solution with the Ricci scalar smoothly changing its sign during evolution. We have found anisotropic solutions using the conformal transformation of the metric and the Einstein frame. The general solution in the Einstein frame has been found explicitly. The corresponding solution in the Jordan frame has been constructed in quadratures., Comment: 13 pages, accepted for publication in Phys. Atom. Nuclei

Analyzing proteins from single cells by tandem mass spectrometry (MS) has become technically feasible. While such analysis has the potential to accurately quantify thousands of proteins across thousands of single cells, the accuracy and reproducibility of the results may be undermined by numerous factors affecting experimental design, sample preparation, data acquisition, and data analysis. Broadly accepted community guidelines and standardized metrics will enhance rigor, data quality, and alignment between laboratories. Here we propose best practices, quality controls, and data reporting recommendations to assist in the broad adoption of reliable quantitative workflows for single-cell proteomics., Comment: Supporting website: https://single-cell.net/guidelines

While parallelism remains the main source of performance, architectural implementations and programming models change with each new hardware generation, often leading to costly application re-engineering. Most tools for performance portability require manual and costly application porting to yet another programming model. We propose an alternative approach that automatically translates programs written in one programming model (CUDA), into another (CPU threads) based on Polygeist/MLIR. Our approach includes a representation of parallel constructs that allows conventional compiler transformations to apply transparently and without modification and enables parallelism-specific optimizations. We evaluate our framework by transpiling and optimizing the CUDA Rodinia benchmark suite for a multi-core CPU and achieve a 76% geomean speedup over handwritten OpenMP code. Further, we show how CUDA kernels from PyTorch can efficiently run and scale on the CPU-only Supercomputer Fugaku without user intervention. Our PyTorch compatibility layer making use of transpiled CUDA PyTorch kernels outperforms the PyTorch CPU native backend by 2.7$\times$.

We study several extensions of the Starobinsky model of inflation, which obey all observational constraints on the inflationary parameters, by demanding that both the inflaton scalar potential in the Einstein frame and the $F(R)$ gravity function in the Jordan frame have the explicit dependence upon fields {\it and} parameters in terms of elementary functions. Our models are continuously connected to the original Starobinsky model via changing the parameters. We modify the Starobinsky $(R+R^2)$ model by adding an $R^3$-term, an $R^4$-term, and an $R^{3/2}$-term, respectively, and calculate the scalar potentials, the inflationary observables and the allowed limits on the deformation parameters by using the latest observational bounds. We find that the tensor-to-scalar ratio in the Starobinsky model modified by the $R^{3/2}$-term significantly increases with raising the parameter in front of that term. On the other side, we deform the scalar potential of the Starobinsky model in the Einstein frame in powers of $y=\exp\left(-\sqrt{\frac{2}{3}}\phi/M_{Pl}\right)$, where $\phi$ is the canonical inflaton (scalaron) field, calculate the corresponding $F(R)$ gravity functions in the two new cases, and find the restrictions on the deformation parameters in the lowest orders with respect to the variable $y$ that is physically small during slow-roll inflation., Comment: 30 pages, 10 figures; minor additions, references added; the version to appear in JCAP

We consider modified gravity cosmological models that can be transformed into two-field chiral cosmological models by the conformal metric transformation. For the $R^2$ gravity model with an additional scalar field and the corresponding two-field model with the cosmological constant and nonstandard kinetic part of the action, the general solutions have been obtained in the spatially flat FLRW metric. We analyze the correspondence of the cosmic time solutions obtained and different possible evolutions of the Hubble parameters in the Einstein and Jordan frames., Comment: 11 pages, 3 figures, v2: minor corrections, references are added, to appear in EPJC

Motivation: Accurate estimation of false discovery rate (FDR) of spectral identification is a central problem in mass spectrometry-based proteomics. Over the past two decades, target decoy approaches (TDAs) and decoy-free approaches (DFAs), have been widely used to estimate FDR. TDAs use a database of decoy species to faithfully model score distributions of incorrect peptide-spectrum matches (PSMs). DFAs, on the other hand, fit two-component mixture models to learn the parameters of correct and incorrect PSM score distributions. While conceptually straightforward, both approaches lead to problems in practice, particularly in experiments that push instrumentation to the limit and generate low fragmentation-efficiency and low signal-to-noise-ratio spectra. Results: We introduce a new decoy-free framework for FDR estimation that generalizes present DFAs while exploiting more search data in a manner similar to TDAs. Our approach relies on multi-component mixtures, in which score distributions corresponding to the correct PSMs, best incorrect PSMs, and second-best incorrect PSMs are modeled by the skew normal family. We derive EM algorithms to estimate parameters of these distributions from the scores of best and second-best PSMs associated with each experimental spectrum. We evaluate our models on multiple proteomics datasets and a HeLa cell digest case study consisting of more than a million spectra in total. We provide evidence of improved performance over existing DFAs and improved stability and speed over TDAs without any performance degradation. We propose that the new strategy has the potential to extend beyond peptide identification and reduce the need for TDA on all analytical platforms.

We construct the $F(R)$ gravity models with exact particular solutions using the conformal transformation and the superpotential method for the corresponding models in the Einstein frame. The functions $F(R)$ are obtained explicitly. We consider exact solutions for the obtained $R^2$ gravity model with the cosmological constant in detail., Comment: 10 pages, references added

Massive stellar content of stellar complexes and associations in M 33 is studied combining deep UBV photometry from the Local Group Survey (Massey et al. 2006) and JHK photometry from the 2MASS. Two basic populations (incl. OB stars and red supergiants) are distinguished and their application for reconstruction of the star formation process in this galaxy are discussed., Comment: Proc. VI Serbian-Bulgarian Astronomical Conference, Belgrade 7-11 May 2008, Publ. Astr. Soc. "Rudjer Boskovic", No. 9, 2009, 449-456

A method for identification of stellar complexes in M33 is applied. Several OB associations form a stellar complex with mean size of 0.3 - 1 kpc. We apply a correlation technique to compare different stellar populations in M33. Our results confirm the existence of a strong correlation between OB stars, HII regions and WR stars, which trace the regions of massive star formation. There is a good correlation between red supergiants (RSGs) and WR stars in the spiral arms of M33. This can be expected since the progenitors of WR stars are massive OB stars or RSGs with masses greater than 20 M_sun. The massive RSGs and WR stars probably originate from nearby sites of star formation. We consider this fact as a ground for identification of stellar complexes in M33. The presented method for identification of stellar complexes can be applied to other nearby galaxies., Comment: 5 pages (in PostScript format), 1 figure. To appear in the Publications of the Astronomical Observatory of Belgrade

Three new obscured Milky Way clusters were detected as surface density peaks in the 2MASS point source catalog during our on-going search for hidden globular clusters and massive Arches-like star clusters. One more cluster was discovered serendipitously during a visual inspection of the candidates. The first deep J, H, and Ks imaging of the cluster [IBP2002] CC01 is presented. We estimated a cluster age of ~1-3 Myr, distance modulus of (m-M)0=12.56+-0.08 mag (D=3.5 Kpc), and extinction of AV~7.7 mag. We also derived the initial mass function slope for the cluster: Gamma=-2.23+-0.16. The integration over the initial mass function yielded a total cluster mass M_{total}<=1800+-200Msol. CC01 appears to be a regular, not very massive star cluster, whose formation has probably been induced by the shock front from the nearby HII region Sh2-228., Comment: 8 pages, 9 figures, accepted in A&A

A systematic search of the 2MASS point source catalog, covering 47% of the sky, was carried out aiming to reveal any hidden globular clusters in our Galaxy. Eight new star clusters were discovered by a search algorithm based on finding peaks in the apparent stellar surface density, and a visual inspection of their vicinities yielded additional two. They all are concentrated toward the Galactic plane and are hidden behind up to Av=20 mag which accounts for their late discovery. The majority of new clusters are associated with H II regions or unidentified IRAS sources suggesting that they are young, probably similar to Arches or open clusters. Only one candidate has morphology similar to a globular cluster and the verification of its nature will require deeper observations with higher angular resolution than the 2MASS data., Comment: 4 pages, 3 figures (1 is a jpeg file), acccepted for publication in Astronomy and Astrophysics