Showing 1–44 of 44 results for author: Hernández, B

AI4Math: A Native Spanish Benchmark for University-Level Mathematical Reasoning in Large Language Models

Authors: Miguel Angel Peñaloza Perez, Bruno Lopez Orozco, Jesus Tadeo Cruz Soto, Michelle Bruno Hernandez, Miguel Angel Alvarado Gonzalez, Sandra Malagon

Abstract: Existing mathematical reasoning benchmarks are predominantly English only or translation-based, which can introduce semantic drift and mask languagespecific reasoning errors. To address this, we present AI4Math, a benchmark of 105 original university level math problems natively authored in Spanish. The dataset spans seven advanced domains (Algebra, Calculus, Geometry, Probability, Number Theory,… ▽ More Existing mathematical reasoning benchmarks are predominantly English only or translation-based, which can introduce semantic drift and mask languagespecific reasoning errors. To address this, we present AI4Math, a benchmark of 105 original university level math problems natively authored in Spanish. The dataset spans seven advanced domains (Algebra, Calculus, Geometry, Probability, Number Theory, Combinatorics, and Logic), and each problem is accompanied by a step by step human solution. We evaluate six large language models GPT 4o, GPT 4o mini, o3 mini, LLaMA 3.3 70B, DeepSeek R1 685B, and DeepSeek V3 685B under four configurations: zero shot and chain of thought, each in Spanish and English. The top models (o3 mini, DeepSeek R1 685B, DeepSeek V3 685B) achieve over 70% accuracy, whereas LLaMA 3.3 70B and GPT-4o mini remain below 40%. Most models show no significant performance drop between languages, with GPT 4o even performing better on Spanish problems in the zero shot setting. Geometry, Combinatorics, and Probability questions remain persistently challenging for all models. These results highlight the need for native-language benchmarks and domain-specific evaluations to reveal reasoning failures not captured by standard metrics. △ Less

Submitted 25 May, 2025; originally announced May 2025.

Comments: 36 pages, 5 figures

MSC Class: 68 ACM Class: I.2

Finest decomposition coarsening of reaction networks of biochemical systems

Authors: Bryan S. Hernandez, Juan Paolo C. Santos, Patrick Vincent N. Lubenia, Eduardo R. Mendoza

Abstract: Biochemical reaction networks are typically modeled by $\dfrac{dx}{dt}=N\cdot K(x)=Y\cdot I_a\cdot K(x)$, with $x$ and $K(x)$ as the concentration and rate vectors, respectively, and $N$, $Y$, and $I_a$ as the stoichiometric, molecularity, and incidence matrices, respectively. Steady states, which describe their long-term behaviors, are determined by solving $N\cdot K(x)=0$, while complex balanced… ▽ More Biochemical reaction networks are typically modeled by $\dfrac{dx}{dt}=N\cdot K(x)=Y\cdot I_a\cdot K(x)$, with $x$ and $K(x)$ as the concentration and rate vectors, respectively, and $N$, $Y$, and $I_a$ as the stoichiometric, molecularity, and incidence matrices, respectively. Steady states, which describe their long-term behaviors, are determined by solving $N\cdot K(x)=0$, while complex balanced steady states are found by solving $I_a \cdot K(x)=0$. To investigate these complex networks, decomposition techniques are important, in particular, for computing steady states. Previously, we identified a widespread property across many networks: the existence of independent and incidence-independent decompositions, characterized by the ability to directly sum the stoichiometric and incidence matrices of the subnetworks, respectively, to match those of the entire network. Here, we discover the ubiquitous property that we call the Finest Decomposition Coarsening (FDC), where the finest independent decomposition (FID) is a coarsening of the finest incidence-independent decomposition (FIID). To support the analysis of this property, we introduce a MATLAB package designed to compute both these decompositions. We then characterize the FDC property and its relationship to structural factors such as the invertibility of the molecularity matrix. We also introduce and characterize the Finest Decompositions Equality (FDE) property, where FIID equals FID. Notably, we show that all deficiency zero networks exhibit the FDE property. Furthermore, we establish important relationships of the FID and FIID with decomposition of the network into its connected components. Our results highlight the prevalence of the coarsening property in reaction networks and deepens the understanding of the algebraic structure and dynamics of biochemical networks. △ Less

Submitted 4 December, 2024; originally announced December 2024.

Deep Learning Predicts Mammographic Breast Density in Clinical Breast Ultrasound Images

Authors: Arianna Bunnell, Dustin Valdez, Thomas K. Wolfgruber, Brandon Quon, Kailee Hung, Brenda Y. Hernandez, Todd B. Seto, Jeffrey Killeen, Marshall Miyoshi, Peter Sadowski, John A. Shepherd

Abstract: Background: Breast density, as derived from mammographic images and defined by the American College of Radiology's Breast Imaging Reporting and Data System (BI-RADS), is one of the strongest risk factors for breast cancer. Breast ultrasound (BUS) is an alternative breast cancer screening modality, particularly useful for early detection in low-resource, rural contexts. The purpose of this study wa… ▽ More Background: Breast density, as derived from mammographic images and defined by the American College of Radiology's Breast Imaging Reporting and Data System (BI-RADS), is one of the strongest risk factors for breast cancer. Breast ultrasound (BUS) is an alternative breast cancer screening modality, particularly useful for early detection in low-resource, rural contexts. The purpose of this study was to explore an artificial intelligence (AI) model to predict BI-RADS mammographic breast density category from clinical, handheld BUS imaging. Methods: All data are sourced from the Hawaii and Pacific Islands Mammography Registry. We compared deep learning methods from BUS imaging, as well as machine learning models from image statistics alone. The use of AI-derived BUS density as a risk factor for breast cancer was then compared to clinical BI-RADS breast density while adjusting for age. The BUS data were split by individual into 70/20/10% groups for training, validation, and testing. Results: 405,120 clinical BUS images from 14.066 women were selected for inclusion in this study, resulting in 9.846 women for training (302,574 images), 2,813 for validation (11,223 images), and 1,406 for testing (4,042 images). On the held-out testing set, the strongest AI model achieves AUROC 0.854 predicting BI-RADS mammographic breast density from BUS imaging and outperforms all shallow machine learning methods based on image statistics. In cancer risk prediction, age-adjusted AI BUS breast density predicted 5-year breast cancer risk with 0.633 AUROC, as compared to 0.637 AUROC from age-adjusted clinical breast density. Conclusions: BI-RADS mammographic breast density can be estimated from BUS imaging with high accuracy using a deep learning model. Furthermore, we demonstrate that AI-derived BUS breast density is predictive of 5-year breast cancer risk in our population. △ Less

Submitted 7 November, 2024; v1 submitted 31 October, 2024; originally announced November 2024.

Multi-Camera Multi-Person Association using Transformer-Based Dense Pixel Correspondence Estimation and Detection-Based Masking

Authors: Daniel Kathein, Byron Hernandez, Henry Medeiros

Abstract: Multi-camera Association (MCA) is the task of identifying objects and individuals across camera views and is an active research topic, given its numerous applications across robotics, surveillance, and agriculture. We investigate a novel multi-camera multi-target association algorithm based on dense pixel correspondence estimation with a Transformer-based architecture and underlying detection-base… ▽ More Multi-camera Association (MCA) is the task of identifying objects and individuals across camera views and is an active research topic, given its numerous applications across robotics, surveillance, and agriculture. We investigate a novel multi-camera multi-target association algorithm based on dense pixel correspondence estimation with a Transformer-based architecture and underlying detection-based masking. After the algorithm generates a set of corresponding keypoints and their respective confidence levels between every pair of detections in the camera views are computed, an affinity matrix is determined containing the probabilities of matches between each pair. Finally, the Hungarian algorithm is applied to generate an optimal assignment matrix with all the predicted associations between the camera views. Our method is evaluated on the WILDTRACK Seven-Camera HD Dataset, a high-resolution dataset containing footage of walking pedestrians as well as precise annotations and camera calibrations. Our results conclude that the algorithm performs exceptionally well associating pedestrians on camera pairs that are positioned close to each other and observe the scene from similar perspectives. On camera pairs with orientations that are drastically different in distance or angle, there is still significant room for improvement. △ Less

Submitted 17 August, 2024; originally announced August 2024.

Comments: 5 pages, 6 figures

Learning a Clinically-Relevant Concept Bottleneck for Lesion Detection in Breast Ultrasound

Authors: Arianna Bunnell, Yannik Glaser, Dustin Valdez, Thomas Wolfgruber, Aleen Altamirano, Carol Zamora González, Brenda Y. Hernandez, Peter Sadowski, John A. Shepherd

Abstract: Detecting and classifying lesions in breast ultrasound images is a promising application of artificial intelligence (AI) for reducing the burden of cancer in regions with limited access to mammography. Such AI systems are more likely to be useful in a clinical setting if their predictions can be explained to a radiologist. This work proposes an explainable AI model that provides interpretable pred… ▽ More Detecting and classifying lesions in breast ultrasound images is a promising application of artificial intelligence (AI) for reducing the burden of cancer in regions with limited access to mammography. Such AI systems are more likely to be useful in a clinical setting if their predictions can be explained to a radiologist. This work proposes an explainable AI model that provides interpretable predictions using a standard lexicon from the American College of Radiology's Breast Imaging and Reporting Data System (BI-RADS). The model is a deep neural network featuring a concept bottleneck layer in which known BI-RADS features are predicted before making a final cancer classification. This enables radiologists to easily review the predictions of the AI system and potentially fix errors in real time by modifying the concept predictions. In experiments, a model is developed on 8,854 images from 994 women with expert annotations and histological cancer labels. The model outperforms state-of-the-art lesion detection frameworks with 48.9 average precision on the held-out testing set, and for cancer classification, concept intervention is shown to increase performance from 0.876 to 0.885 area under the receiver operating characteristic curve. Training and evaluation code is available at https://github.com/hawaii-ai/bus-cbm. △ Less

Submitted 28 June, 2024; originally announced July 2024.

Comments: Submitted version of manuscript accepted at MICCAI 2024. This preprint has not undergone peer review or any post-submission improvements or corrections

Embedding-based comparison of reaction networks of Wnt signaling

Authors: Bryan S. Hernandez, Patrick Vincent N. Lubenia, Eduardo R. Mendoza

Abstract: This work introduces a new method for comparing two reaction networks of the same or closely related systems through their embedded networks in terms of the shared set of species. Hence, we call this method the Common Species Embedded Networks (CSEN) analysis. Using this approach, we conduct a comparison of existing reaction networks associated with Wnt signaling models (Lee, Schmitz, MacLean, and… ▽ More This work introduces a new method for comparing two reaction networks of the same or closely related systems through their embedded networks in terms of the shared set of species. Hence, we call this method the Common Species Embedded Networks (CSEN) analysis. Using this approach, we conduct a comparison of existing reaction networks associated with Wnt signaling models (Lee, Schmitz, MacLean, and Feinberg) that we have identified. The analysis yields three important results for these Wnt models. First, the CSEN analysis of the Lee (mono-stationary) and Feinberg (multi-stationary) shows a strong similarity, justifying the study of the Feinberg model, which was a modified Lee model constructed to study an important network property called "concordance". It also challenge the absoluteness of discrimination of the models into mono-stationarity versus multi-stationarity, which is a main result of Maclean et al. (PNAS USA 2015). Second, the CSEN analysis provides evidence supporting a strong similarity between the Schmitz and MacLean models, as indicated by the "proximate equivalence" that we have identified. Third, the analysis underscores the absence of a comparable relationship between the Feinberg and MacLean models, highlighting distinctive differences between the two. Thus, our approach could be a useful tool to compare mathematical models of the same or closely related systems. △ Less

Submitted 1 April, 2024; originally announced April 2024.

Positive steady states of a class of power law systems with independent decompositions

Authors: Al Jay Lan J. Alamin, Bryan S. Hernandez

Abstract: Power law systems have been studied extensively due to their wide-ranging applications, particularly in chemistry. In this work, we focus on power law systems that can be decomposed into stoichiometrically independent subsystems. We show that for such systems where the ranks of the augmented matrices containing the kinetic order vectors of the underlying subnetworks sum up to the rank of the augme… ▽ More Power law systems have been studied extensively due to their wide-ranging applications, particularly in chemistry. In this work, we focus on power law systems that can be decomposed into stoichiometrically independent subsystems. We show that for such systems where the ranks of the augmented matrices containing the kinetic order vectors of the underlying subnetworks sum up to the rank of the augmented matrix containing the kinetic order vectors of the entire network, then the existence of the positive steady states of each stoichiometrically independent subsystem is a necessary and sufficient condition for the existence of the positive steady states of the given power law system. We demonstrate the result through illustrative examples. One of which is a network of a carbon cycle model that satisfies the assumption, while the other network fails to meet the assumption. Finally, using the aforementioned result, we present a systematic method for deriving positive steady state parametrizations for the mentioned subclass of power law systems, which is a generalization of our recent method for mass action systems. △ Less

Submitted 21 April, 2024; v1 submitted 1 April, 2024; originally announced April 2024.

DevPhish: Exploring Social Engineering in Software Supply Chain Attacks on Developers

Authors: Hossein Siadati, Sima Jafarikhah, Elif Sahin, Terrence Brent Hernandez, Elijah Lorenzo Tripp, Denis Khryashchev, Amin Kharraz

Abstract: The Software Supply Chain (SSC) has captured considerable attention from attackers seeking to infiltrate systems and undermine organizations. There is evidence indicating that adversaries utilize Social Engineering (SocE) techniques specifically aimed at software developers. That is, they interact with developers at critical steps in the Software Development Life Cycle (SDLC), such as accessing Gi… ▽ More The Software Supply Chain (SSC) has captured considerable attention from attackers seeking to infiltrate systems and undermine organizations. There is evidence indicating that adversaries utilize Social Engineering (SocE) techniques specifically aimed at software developers. That is, they interact with developers at critical steps in the Software Development Life Cycle (SDLC), such as accessing Github repositories, incorporating code dependencies, and obtaining approval for Pull Requests (PR) to introduce malicious code. This paper aims to comprehensively explore the existing and emerging SocE tactics employed by adversaries to trick Software Engineers (SWEs) into delivering malicious software. By analyzing a diverse range of resources, which encompass established academic literature and real-world incidents, the paper systematically presents an overview of these manipulative strategies within the realm of the SSC. Such insights prove highly beneficial for threat modeling and security gap analysis. △ Less

Submitted 26 September, 2024; v1 submitted 28 February, 2024; originally announced February 2024.

Comments: 7 pages, 2 figures

MSC Class: computer security; phishing; software supply chain ACM Class: D.4.6; E.3

Equilibria decomposition-based comparison of reaction networks of Wnt signaling

Authors: Bryan S. Hernandez, Patrick Vincent N. Lubenia, Eduardo R. Mendoza

Abstract: The Wnt signaling pathway plays a critical role in various biochemical processes, including embryonic development, tissue homeostasis, and cancer progression. In this paper, we conduct a comparative analysis of $β$-catenin-dependent Wnt signaling reaction networks, which we refer to as the Feinberg, Schmitz, and MacLean models, based on the previous study by MacLean et al. (PNAS USA 2015). Our ana… ▽ More The Wnt signaling pathway plays a critical role in various biochemical processes, including embryonic development, tissue homeostasis, and cancer progression. In this paper, we conduct a comparative analysis of $β$-catenin-dependent Wnt signaling reaction networks, which we refer to as the Feinberg, Schmitz, and MacLean models, based on the previous study by MacLean et al. (PNAS USA 2015). Our analysis is based on the (unique) finest independent decomposition (FID) of each reaction network and our comparative techniques include equilibria parametrizations (EP) and the newly developed methods of Common Reactions Equilibria (CORE) analysis and Concordance Profile (CP) analysis. Our investigation yields three interesting results concerning the equilibria sets of these models. Firstly, we explore the concept of absolute concentration robustness (ACR), wherein a system exhibits ACR in a specific species if the equilibrium value for that species is the same for any positive equilibrium. Through ACR analysis employing FID and EP, we observe that both the Schmitz and MacLean models lack ACR, whereas the Feinberg model demonstrates ACR in a single species. Second, our analyses using FID and CORE reveal important relationships within the equilibria sets of the augmented Schmitz and MacLean models. Furthermore, FID and CORE identify the lack of a substantial relationship between the equilibria sets of the Feinberg and MacLean models. Hence, these methods detect subtle differences between the Feinberg and MacLean models and also between the Schmitz and MacLean models, which are not evident in the standard reaction network analysis. Finally, based on the concordance levels, CP analysis indicates that the MacLean and Schmitz models are more similar than the MacLean and Feinberg models. △ Less

Submitted 27 March, 2024; v1 submitted 29 October, 2023; originally announced October 2023.

EZ: An Efficient, Charge Conserving Current Deposition Algorithm for Electromagnetic Particle-In-Cell Simulations

Authors: Klaus Steiniger, Rene Widera, Sergei Bastrakov, Michael Bussmann, Sunita Chandrasekaran, Benjamin Hernandez, Kristina Holsapple, Axel Huebl, Guido Juckeland, Jeffrey Kelling, Matt Leinhauser, Richard Pausch, David Rogers, Ulrich Schramm, Jeff Young, Alexander Debus

Abstract: We present EZ, a novel current deposition algorithm for particle-in-cell (PIC) simulations. EZ calculates the current density on the electromagnetic grid due to macro-particle motion within a time step by solving the continuity equation of electrodynamics. Being a charge conserving hybridization of Esirkepov's method and ZigZag, we refer to it as ``EZ'' as shorthand for ``Esirkepov meets ZigZag''.… ▽ More We present EZ, a novel current deposition algorithm for particle-in-cell (PIC) simulations. EZ calculates the current density on the electromagnetic grid due to macro-particle motion within a time step by solving the continuity equation of electrodynamics. Being a charge conserving hybridization of Esirkepov's method and ZigZag, we refer to it as ``EZ'' as shorthand for ``Esirkepov meets ZigZag''. Simulations of a warm, relativistic plasma with PIConGPU show that EZ achieves the same level of charge conservation as the commonly used method by Esirkepov, yet reaches higher performance for macro-particle assignment-functions up to third-order. In addition to a detailed description of the functioning of EZ, reasons for the expected and observed performance increase are given, and guidelines for its implementation aiming at highest performance on GPUs are provided. △ Less

Submitted 18 September, 2023; originally announced September 2023.

Journal ref: Computer Physics Communications 291 (2023) 108849

A decomposition-based approach for deriving positive steady states of a class of chemical reaction networks with non-mass-action kinetics

Authors: Bryan S. Hernandez, Patrick Vincent N. Lubenia

Abstract: Steady states are frequently used to investigate the long-term behaviors of (bio)-chemical systems. Recently, there has been a growing interest in network-based approaches due to their efficiency in deriving parametrizations of positive steady states in systems with mass-action kinetics. In this study, we extend this approach to derive positive steady states in networks under non-mass-action kinet… ▽ More Steady states are frequently used to investigate the long-term behaviors of (bio)-chemical systems. Recently, there has been a growing interest in network-based approaches due to their efficiency in deriving parametrizations of positive steady states in systems with mass-action kinetics. In this study, we extend this approach to derive positive steady states in networks under non-mass-action kinetics, specifically mixed kinetics. In a system with mixed kinetics, some reactions {may follow} mass-action kinetics, while others in the same network follow different rate laws, such as quotient rate laws. An example of such complexity is evident in a mathematical model of the insulin signaling pathway in type 2 diabetes. To compute its positive {steady states}, we adapt our existing network decomposition approach, originally designed for mass-action kinetics, to handle networks with non-mass-action kinetics. This approach involves breaking down a given network into smaller, independent subnetworks to derive the positive steady states of each subnetwork separately. These individual steady states are then combined to obtain the positive steady states of the entire network. This strategy makes computations more manageable for complex and large networks. More importantly, this method could separate reactions with purely mass-action kinetics into certain subnetworks from those that follow different rate laws. We also present an illustrative example that provides insights into methods for transforming networks with mixed kinetics into their associated mass-action systems. △ Less

Submitted 1 April, 2024; v1 submitted 23 April, 2023; originally announced April 2023.

Computational translation framework identifies biochemical reaction networks with special topologies and their long-term dynamics

Authors: Hyukpyo Hong, Bryan S. Hernandez, Jinsu Kim, Jae Kyoung Kim

Abstract: Long-term behaviors of biochemical systems are described by steady states in deterministic models and stationary distributions in stochastic models. Obtaining their analytic solutions can be done for limited cases, such as linear or finite-state systems, as it generally requires solving many coupled equations. Interestingly, analytic solutions can be easily obtained when underlying networks have s… ▽ More Long-term behaviors of biochemical systems are described by steady states in deterministic models and stationary distributions in stochastic models. Obtaining their analytic solutions can be done for limited cases, such as linear or finite-state systems, as it generally requires solving many coupled equations. Interestingly, analytic solutions can be easily obtained when underlying networks have special topologies, called weak reversibility (WR) and zero deficiency (ZD), and the kinetic law follows a generalized form of mass-action kinetics. However, such desired topological conditions do not hold for the majority of cases. Thus, translating networks to have WR and ZD while preserving the original dynamics was proposed. Yet, this approach is limited because manually obtaining the desired network translation among the large number of candidates is challenging. Here, we prove necessary conditions for having WR and ZD after translation, and based on these conditions, we develop a user-friendly computational package, TOWARDZ, that automatically and efficiently identifies translated networks with WR and ZD. This allows us to quantitatively examine how likely it is to obtain WR and ZD after translation depending on the number of species and reactions. Importantly, we also describe how our package can be used to analytically derive steady states of deterministic models and stationary distributions of stochastic models. TOWARDZ provides an effective tool to analyze biochemical systems. △ Less

Submitted 2 December, 2022; originally announced December 2022.

Comments: 24 pages, 7 figures

MSC Class: 92B05; 92C42; 34A34; 60J27; 60J28; 60G10

Mean convergence of Fourier-Akhiezer-Chebyshev series

Authors: Manuel Bello Hernández, Alejandro del Campo López

Abstract: We prove mean convergence of the Fourier series in Akhiezer-Chebyshev polynomials in $L^p$, $p>1$, using a weighted inequality for the Hilbert transform in an arc of the unit circle. We prove mean convergence of the Fourier series in Akhiezer-Chebyshev polynomials in $L^p$, $p>1$, using a weighted inequality for the Hilbert transform in an arc of the unit circle. △ Less

Submitted 12 November, 2022; originally announced November 2022.

Comments: 20 pages, summited to a journal

MSC Class: 42C10; 42C05; 32A35

Gold-standard of HER2 breast cancer biopsies using supervised learning based on multiple pathologist annotations

Authors: Benjamín Hernández, Violeta Chang

Abstract: Breast cancer is one of the most common cancer in women around the world. For diagnosis, pathologists evaluate biomarkers such as HER2 protein using immunohistochemistry over tissue extracted by a biopsy. Through microscopic inspection, this assessment estimates the intensity and integrity of the membrane cells' staining and scores the sample as 0, 1+, 2+, or 3+: a subjective decision that depends… ▽ More Breast cancer is one of the most common cancer in women around the world. For diagnosis, pathologists evaluate biomarkers such as HER2 protein using immunohistochemistry over tissue extracted by a biopsy. Through microscopic inspection, this assessment estimates the intensity and integrity of the membrane cells' staining and scores the sample as 0, 1+, 2+, or 3+: a subjective decision that depends on the interpretation of the pathologist. This paper presents the preliminary data analysis of the annotations of three pathologists over the same set of samples obtained using 20x magnification and including $1,252$ non-overlapping biopsy patches. We evaluate the intra- and inter-expert variability achieving substantial and moderate agreement, respectively, according to Fleiss' Kappa coefficient, as a previous stage towards a generation of a HER2 breast cancer biopsy gold-standard using supervised learning from multiple pathologist annotations. △ Less

Submitted 8 November, 2022; originally announced November 2022.

Semi-vortices and cluster-vorticity: new concepts in the Berezinskii-Kosterlitz-Thouless phase transition

Authors: Brandon Gómez Bravo, Bryan D. Juárez Hernández, Wolfgang Bietenholz

Abstract: The Berezinskii-Kosterlitz-Thouless (BKT) essential phase transition in the 2d XY model is revisited. Its mechanism is usually described by the (un)binding of vortex--anti-vortex (V--AV) pairs, which does, however, not provide a clear-cut quantitative criterion for criticality. Known sharp criteria are the divergence of the correlation length and a discontinuity of the helicity modulus. Here we pr… ▽ More The Berezinskii-Kosterlitz-Thouless (BKT) essential phase transition in the 2d XY model is revisited. Its mechanism is usually described by the (un)binding of vortex--anti-vortex (V--AV) pairs, which does, however, not provide a clear-cut quantitative criterion for criticality. Known sharp criteria are the divergence of the correlation length and a discontinuity of the helicity modulus. Here we propose and probe a new criterion: it is based on the concepts of semi-vortices and cluster vorticity, which are formulated in the framework of the multi-cluster algorithm that we use to simulate the 2d XY model. △ Less

Submitted 30 July, 2022; v1 submitted 3 June, 2022; originally announced June 2022.

Comments: 6 pages, LaTex, 7 figures, presented at the "XXXV Reunión Anual de la División de Partículas y Campos", of the "Socieded Mexicana de Física", 2 references added

Journal ref: Supl. Rev. Mex. Fis. 3 (2022) 020724

Proteome-scale Deployment of Protein Structure Prediction Workflows on the Summit Supercomputer

Authors: Mu Gao, Mark Coletti, Russell B. Davidson, Ryan Prout, Subil Abraham, Benjamin Hernandez, Ada Sedova

Abstract: Deep learning has contributed to major advances in the prediction of protein structure from sequence, a fundamental problem in structural bioinformatics. With predictions now approaching the accuracy of crystallographic resolution in some cases, and with accelerators like GPUs and TPUs making inference using large models rapid, fast genome-level structure prediction becomes an obvious aim. Leaders… ▽ More Deep learning has contributed to major advances in the prediction of protein structure from sequence, a fundamental problem in structural bioinformatics. With predictions now approaching the accuracy of crystallographic resolution in some cases, and with accelerators like GPUs and TPUs making inference using large models rapid, fast genome-level structure prediction becomes an obvious aim. Leadership-class computing resources can be used to perform genome-scale protein structure prediction using state-of-the-art deep learning models, providing a wealth of new data for systems biology applications. Here we describe our efforts to efficiently deploy the AlphaFold2 program, for full-proteome structure prediction, at scale on the Oak Ridge Leadership Computing Facility's resources, including the Summit supercomputer. We performed inference to produce the predicted structures for 35,634 protein sequences, corresponding to three prokaryotic proteomes and one plant proteome, using under 4,000 total Summit node hours, equivalent to using the majority of the supercomputer for one hour. We also designed an optimized structure refinement that reduced the time for the relaxation stage of the AlphaFold pipeline by over 10X for longer sequences. We demonstrate the types of analyses that can be performed on proteome-scale collections of sequences, including a search for novel quaternary structures and implications for functional annotation. △ Less

Submitted 24 January, 2022; originally announced January 2022.

Positive equilibria of power law kinetics on networks with independent linkage classes

Authors: Bryan S. Hernandez, Eduardo R. Mendoza

Abstract: Studies about the set of positive equilibria ($E_+$) of kinetic systems have been focused on mass action, and not that much on power law kinetic (PLK) systems, even for PL-RDK systems (PLK systems where two reactions with identical reactant complexes have the same kinetic order vectors). For mass action, reactions with different reactants have different kinetic order rows. A PL-RDK system satisfyi… ▽ More Studies about the set of positive equilibria ($E_+$) of kinetic systems have been focused on mass action, and not that much on power law kinetic (PLK) systems, even for PL-RDK systems (PLK systems where two reactions with identical reactant complexes have the same kinetic order vectors). For mass action, reactions with different reactants have different kinetic order rows. A PL-RDK system satisfying this property is called factor span surjective (PL-FSK). In this work, we show that a cycle terminal PL-FSK system with $E_+\ne \varnothing$ and has independent linkage classes (ILC) is a poly-PLP system, i.e., $E_+$ is the disjoint union of log-parametrized sets. The key insight for the extension is that factor span surjectivity induces an isomorphic digraph structure on the kinetic complexes. The result also completes, for ILC networks, the structural analysis of the original complex balanced generalized mass action systems (GMAS) by Müller and Regensburger. We also identify a large set of PL-RDK systems where non-emptiness of $E_+$ is a necessary and sufficient condition for non-emptiness of each set of positive equilibria for each linkage class. These results extend those of Boros on mass action systems with ILC. We conclude this paper with two applications of our results. Firstly, we consider absolute complex balancing (ACB), i.e., the property that each positive equilibrium is complex balanced, in poly-PLP systems. Finally, we use the new results to study absolute concentration robustness (ACR) in these systems. In particular, we obtain a species hyperplane containment criterion to determine ACR in the system species. △ Less

Submitted 5 April, 2022; v1 submitted 13 November, 2021; originally announced November 2021.

MSC Class: 92C45; 80A30; 92C42

Independent, Incidence Independent and Weakly Reversible Decompositions of Chemical Reaction Networks

Authors: Bryan S. Hernandez, Deza A. Amistas, Ralph John L. De la Cruz, Lauro L. Fontanil, Aurelio A. de los Reyes V, Eduardo R. Mendoza

Abstract: Chemical reaction networks (CRNs) are directed graphs with reactant or product complexes as vertices, and reactions as arcs. A CRN is weakly reversible if each of its connected components is strongly connected. Weakly reversible networks can be considered as the most important class of reaction networks. Now, the stoichiometric subspace of a network is the linear span of the reaction vectors (i.e.… ▽ More Chemical reaction networks (CRNs) are directed graphs with reactant or product complexes as vertices, and reactions as arcs. A CRN is weakly reversible if each of its connected components is strongly connected. Weakly reversible networks can be considered as the most important class of reaction networks. Now, the stoichiometric subspace of a network is the linear span of the reaction vectors (i.e., difference between the product and the reactant complexes). A decomposition of a CRN is independent (incidence independent) if the direct sum of the stoichiometric subspaces (incidence maps) of the subnetworks equals the stoichiometric subspace (incidence map) of the whole network. Decompositions can be used to study relationships between steady states of the whole system (induced from partitioning the reaction set of the underlying network) and those of its subsystems. In this work, we revisit our novel method of finding independent decomposition, and use it to expand applicability on (vector) components of steady states. We also explore CRNs with embedded deficiency zero independent subnetworks. In addition, we establish a method for finding incidence independent decomposition of a CRN. We determine all the forms of independent and incidence independent decompositions of a network, and provide the number of such decompositions. Lastly, for weakly reversible networks, we determine that incidence independence is a sufficient condition for weak reversibility of a decomposition, and we identify subclasses of weakly reversible networks where any independent decomposition is weakly reversible. △ Less

Submitted 23 August, 2021; v1 submitted 12 August, 2021; originally announced August 2021.

First-Generation Inference Accelerator Deployment at Facebook

Authors: Michael Anderson, Benny Chen, Stephen Chen, Summer Deng, Jordan Fix, Michael Gschwind, Aravind Kalaiah, Changkyu Kim, Jaewon Lee, Jason Liang, Haixin Liu, Yinghai Lu, Jack Montgomery, Arun Moorthy, Satish Nadathur, Sam Naghshineh, Avinash Nayak, Jongsoo Park, Chris Petersen, Martin Schatz, Narayanan Sundaram, Bangsheng Tang, Peter Tang, Amy Yang, Jiecao Yu , et al. (90 additional authors not shown)

Abstract: In this paper, we provide a deep dive into the deployment of inference accelerators at Facebook. Many of our ML workloads have unique characteristics, such as sparse memory accesses, large model sizes, as well as high compute, memory and network bandwidth requirements. We co-designed a high-performance, energy-efficient inference accelerator platform based on these requirements. We describe the in… ▽ More In this paper, we provide a deep dive into the deployment of inference accelerators at Facebook. Many of our ML workloads have unique characteristics, such as sparse memory accesses, large model sizes, as well as high compute, memory and network bandwidth requirements. We co-designed a high-performance, energy-efficient inference accelerator platform based on these requirements. We describe the inference accelerator platform ecosystem we developed and deployed at Facebook: both hardware, through Open Compute Platform (OCP), and software framework and tooling, through Pytorch/Caffe2/Glow. A characteristic of this ecosystem from the start is its openness to enable a variety of AI accelerators from different vendors. This platform, with six low-power accelerator cards alongside a single-socket host CPU, allows us to serve models of high complexity that cannot be easily or efficiently run on CPUs. We describe various performance optimizations, at both platform and accelerator level, which enables this platform to serve production traffic at Facebook. We also share deployment challenges, lessons learned during performance optimization, as well as provide guidance for future inference hardware co-design. △ Less

Submitted 4 August, 2021; v1 submitted 8 July, 2021; originally announced July 2021.

Independent Decompositions of Chemical Reaction Networks

Authors: Bryan S. Hernandez, Ralph John L. De la Cruz

Abstract: A chemical reaction network (CRN) is composed of reactions that can be seen as interactions among entities called species, which exist within the system. Endowed with kinetics, CRN has a corresponding set of ordinary differential equations (ODEs). In Chemical Reaction Network Theory, we are interested with connections between the structure of the CRN and qualitative properties of the corresponding… ▽ More A chemical reaction network (CRN) is composed of reactions that can be seen as interactions among entities called species, which exist within the system. Endowed with kinetics, CRN has a corresponding set of ordinary differential equations (ODEs). In Chemical Reaction Network Theory, we are interested with connections between the structure of the CRN and qualitative properties of the corresponding ODEs. One of the results in Decomposition Theory of CRNs is that the intersection of the sets of positive steady states of the subsystems is equal to the set of positive steady states of the whole system, if the decomposition is independent. Hence, computational approach using independent decompositions can be used as an efficient tool in studying large systems. In this work, we provide a necessary and sufficient condition for the existence of a nontrivial independent decomposition of a CRN, which leads to a novel step-by-step method to obtain such decomposition, if it exists. We also illustrate these results using real-life examples. In particular, we show that a CRN of a popular model of anaerobic yeast fermentation pathway has a nontrivial independent decomposition, while a particular biological system, which is a metabolic network with one positive feedforward and a negative feedback has none. Finally, we analyze properties of steady states of reaction networks of specific influenza virus models. △ Less

Submitted 19 April, 2021; originally announced April 2021.

Weakly Reversible CF-Decompositions of Chemical Kinetic Systems

Authors: Bryan S. Hernandez, Eduardo R. Mendoza

Abstract: This paper studies chemical kinetic systems which decompose into weakly reversible complex factorizable (CF) systems. Among power law kinetic systems, CF systems (denoted as PL-RDK systems) are those where branching reactions of a reactant complex have identical rows in the kinetic order matrix. Mass action and generalized mass action systems (GMAS) are well-known examples. Schmitz's global carbon… ▽ More This paper studies chemical kinetic systems which decompose into weakly reversible complex factorizable (CF) systems. Among power law kinetic systems, CF systems (denoted as PL-RDK systems) are those where branching reactions of a reactant complex have identical rows in the kinetic order matrix. Mass action and generalized mass action systems (GMAS) are well-known examples. Schmitz's global carbon cycle model is a previously studied non-complex factorizable (NF) power law system (denoted as PL-NDK). We derive novel conditions for the existence of weakly reversible CF-decompositions and present an algorithm for verifying these conditions. We discuss methods for identifying independent decompositions, i.e., those where the stoichiometric subspaces of the subnetworks form a direct sum, as such decompositions relate positive equilibria sets of the subnetworks to that of the whole network. We then use the results to determine the positive equilibria sets of PL-NDK systems which admit an independent weakly reversible decomposition into PL-RDK systems of PLP type, i.e., the positive equilibria are log-parametrized, which is a broad generalization of a Deficiency Zero Theorem of Fortun et al. (2019). △ Less

Submitted 5 March, 2022; v1 submitted 16 April, 2021; originally announced April 2021.

Datacentric analysis to reduce pedestrians accidents: A case study in Colombia

Authors: Michael Puentes, Diana Novoa, John Delgado Nivia, Carlos Barrios Hernández, Oscar Carrillo, Frédéric Le Mouël

Abstract: Since 2012, in a case-study in Bucaramanga-Colombia, 179 pedestrians died in car accidents, and another 2873 pedestrians were injured. Each day, at least one passerby is involved in a tragedy. Knowing the causes to decrease accidents is crucial, and using system-dynamics to reproduce the collisions' events is critical to prevent further accidents. This work implements simulations to save lives by… ▽ More Since 2012, in a case-study in Bucaramanga-Colombia, 179 pedestrians died in car accidents, and another 2873 pedestrians were injured. Each day, at least one passerby is involved in a tragedy. Knowing the causes to decrease accidents is crucial, and using system-dynamics to reproduce the collisions' events is critical to prevent further accidents. This work implements simulations to save lives by reducing the city's accidental rate and suggesting new safety policies to implement. Simulation's inputs are video recordings in some areas of the city. Deep Learning analysis of the images results in the segmentation of the different objects in the scene, and an interaction model identifies the primary reasons which prevail in the pedestrians or vehicles' behaviours. The first and most efficient safety policy to implement-validated by our simulations-would be to build speed bumps in specific places before the crossings reducing the accident rate by 80%. △ Less

Submitted 2 April, 2021; originally announced April 2021.

Journal ref: International Conference on Sustainable Smart Cities and Territories (SSCt2021), Apr 2021, Doha, Qatar

Nuclear shape transitions and elastic magnetic electron scattering

Authors: B. Hernandez, P. Sarriguren, O. Moreno, E. Moya de Guerra, D. N. Kadrev, A. N. Antonov

Abstract: Backward elastic electron scattering from odd-A nuclear targets is characterized by magnetic form factors containing precise information on the nuclear structure. We study the sensitivity of the magnetic form factors to structural effects related to the evolution and shape transitions in both isotopic and isotonic chains. Calculations of magnetic form factors are performed in the plane-wave Born a… ▽ More Backward elastic electron scattering from odd-A nuclear targets is characterized by magnetic form factors containing precise information on the nuclear structure. We study the sensitivity of the magnetic form factors to structural effects related to the evolution and shape transitions in both isotopic and isotonic chains. Calculations of magnetic form factors are performed in the plane-wave Born approximation. The nuclear structure is obtained from a deformed self-consistent mean-field calculation based on a Skyrme HF+BCS formalism. Collective effects are included in the cranking approximation, whereas nucleon-nucleon correlations are taken into account in the coherent density fluctuation model. The evolution of the magnetic form factors is found to exhibit signatures of shape transitions that show up in selected isotopic and isotonic chains involving both stable and unstable nuclei. Several cases are identified as suitable candidates for showing such fingerprints of shape transitions. A new generation of electron scattering experiments involving electron-radioactive beam colliders will be available in the near future, leading to a renewed interest in this field. △ Less

Submitted 10 December, 2020; v1 submitted 9 December, 2020; originally announced December 2020.

Comments: 21 pages, 10 figures, accepted for publication in Physical Review C. arXiv admin note: text overlap with arXiv:1903.09600

Journal ref: Phys. Rev. C 103, 014303 (2021)

Evolution of Retrograde Orbiters in an AGN Disk

Authors: Amy Secunda, Betsy Hernandez, Jeremy Goodman, Nathan W. C. Leigh, Barry McKernan, K. E. Saavik Ford, Jose I. Adorno

Abstract: AGN disks have been proposed as promising locations for the mergers of stellar mass black hole binaries (BBHs). Much recent work has been done on this merger channel, but the majority focuses on stellar mass black holes (BHs) orbiting in the prograde direction. Little work has been done to examine the impact of retrograde orbiters (ROs) on the formation and mergers of BBHs in AGN disks. Quantifyin… ▽ More AGN disks have been proposed as promising locations for the mergers of stellar mass black hole binaries (BBHs). Much recent work has been done on this merger channel, but the majority focuses on stellar mass black holes (BHs) orbiting in the prograde direction. Little work has been done to examine the impact of retrograde orbiters (ROs) on the formation and mergers of BBHs in AGN disks. Quantifying the retrograde contribution is important, since roughly half of all orbiters should initially be on retrograde orbits when the disk forms. We perform an analytic calculation of the evolution of ROs in an AGN disk. Because this evolution could cause the orbits of ROs to cross those of prograde BBHs, we derive the collision rate between a given RO and a given BBH orbiting in the prograde direction. In the examples given here, ROs in the inner region of the disk experience a rapid decrease in the semimajor axis of their orbits while also becoming highly eccentric in less than a million years. This rapid orbital evolution could lead to extreme mass ratio inspirals detectable by the Laser Interferometer Space Antenna. The collision rates of our example ROs with prograde BBHs in the migration trap depend strongly on the volume of the inner radiation-pressure-dominated region which depends on the mass of the supermassive black hole (SMBH). Rates are lowest for larger mass SMBHs, which dominate the AGN merger channel, suggesting that merger rates for this channel may not be significantly altered by ROs. △ Less

Submitted 4 February, 2021; v1 submitted 8 September, 2020; originally announced September 2020.

Comments: 12 pages, 3 figures, accepted to ApJL

Positive Equilibria of Hill-Type Kinetic Systems

Authors: Bryan S. Hernandez, Eduardo R. Mendoza

Abstract: This work introduces a novel approach to study properties of positive equilibria of a chemical reaction network $\mathscr{N}$ endowed with Hill-type kinetics $K$, called a Hill-type kinetic (HTK) system $\left(\mathscr{N},K\right)$, including their multiplicity and concentration robustness in a species. We associate a unique positive linear combination of power-law kinetic systems called poly-PL k… ▽ More This work introduces a novel approach to study properties of positive equilibria of a chemical reaction network $\mathscr{N}$ endowed with Hill-type kinetics $K$, called a Hill-type kinetic (HTK) system $\left(\mathscr{N},K\right)$, including their multiplicity and concentration robustness in a species. We associate a unique positive linear combination of power-law kinetic systems called poly-PL kinetic (PYK) system $\left( {\mathscr{N},{K_\text{PY}}} \right)$ to the given HTK system. The associated system has the key property that its equilibria sets coincide with those of the Hill-type system, i.e., ${E_ + }\left( {\mathscr{N},K} \right) = {E_ + }\left( {\mathscr{N},{K_\text{PY}}} \right)$ and ${Z_ + }\left( {\mathscr{N},K} \right) = {Z_ + }\left( {\mathscr{N},{K_\text{PY}}} \right)$. This allows us to identify two novel subsets of the Hill-type kinetics, called PL-equilibrated and PL-complex balanced kinetics, to which recent results on absolute concentration robustness (ACR) of species and complex balancing at positive equilibria of power-law (PL) kinetic systems can be applied. Our main results also include the Shinar-Feinberg ACR Theorem for PL-equilibrated HT-RDK systems (i.e., subset of complex factorizable HTK systems), which establishes a foundation for the analysis of ACR in HTK systems, and the extension of the results of Müller and Regensburger on generalized mass action systems to PL-complex balanced HT-RDK systems. In addition, we derive the theory of balanced concentration robustness (BCR) in an analogous manner to ACR for PL-equilibrated systems. Finally, we provide further extensions of our results to a more general class of kinetics, which includes quotients of poly-PL functions. △ Less

Submitted 5 September, 2020; originally announced September 2020.

Adding Seemingly Uninformative Labels Helps in Low Data Regimes

Authors: Christos Matsoukas, Albert Bou I Hernandez, Yue Liu, Karin Dembrower, Gisele Miranda, Emir Konuk, Johan Fredin Haslum, Athanasios Zouzos, Peter Lindholm, Fredrik Strand, Kevin Smith

Abstract: Evidence suggests that networks trained on large datasets generalize well not solely because of the numerous training examples, but also class diversity which encourages learning of enriched features. This raises the question of whether this remains true when data is scarce - is there an advantage to learning with additional labels in low-data regimes? In this work, we consider a task that require… ▽ More Evidence suggests that networks trained on large datasets generalize well not solely because of the numerous training examples, but also class diversity which encourages learning of enriched features. This raises the question of whether this remains true when data is scarce - is there an advantage to learning with additional labels in low-data regimes? In this work, we consider a task that requires difficult-to-obtain expert annotations: tumor segmentation in mammography images. We show that, in low-data settings, performance can be improved by complementing the expert annotations with seemingly uninformative labels from non-expert annotators, turning the task into a multi-class problem. We reveal that these gains increase when less expert data is available, and uncover several interesting properties through further studies. We demonstrate our findings on CSAW-S, a new dataset that we introduce here, and confirm them on two public datasets. △ Less

Submitted 11 August, 2020; v1 submitted 20 July, 2020; originally announced August 2020.

Comments: ICML 2020

A Computational Approach to Multistationarity in Poly-PL Kinetic Systems

Authors: Daryl M. Magpantay, Bryan S. Hernandez, Aurelio A. de los Reyes V, Eduardo R. Mendoza, Ederlina G. Nocon

Abstract: One important question that interests those who work in chemical reaction network theory (CRNT) is this: Does the system obtained from a reaction network admit a positive equilibrium and if it does, can there be more than one within a stoichiometric class? The higher deficiency algorithm (HDA) of Ji and Feinberg provided a method of determining the multistationarity capacity of a CRN with mass act… ▽ More One important question that interests those who work in chemical reaction network theory (CRNT) is this: Does the system obtained from a reaction network admit a positive equilibrium and if it does, can there be more than one within a stoichiometric class? The higher deficiency algorithm (HDA) of Ji and Feinberg provided a method of determining the multistationarity capacity of a CRN with mass action kinetics (MAK). An extension of this, called Multistationarity Algorithm (MSA), recently came into the scene tackling CRNs with power law kinetics (PLK), a kinetic system which is more general (having MAK systems as a special case). For this paper, we provide a computational approach to study the multistationarity feature of reaction networks endowed with kinetics which are non-negative linear combinations of power law functions called poly-PL kinetics (PYK). The idea is to use MSA and combine it with a transformation called STAR-MSC (i.e., $S$-invariant Termwise Addition of Reactions via Maximal Stoichiometric Coefficients) producing PLKs that are dynamically equivalent to PYKs. This leads us to being able to determinine the multistationarity capacity of a much larger class of kinetic systems. We show that if the transformed dynamically equivalent PLK system is multistationary for a stoichiometric class for a set of particular rate constants, then so is its original corresponding PYK system. Moreover, the monostationarity property of the transformed PLK system also implies the monostationarity property of the original PYK system. △ Less

Submitted 19 August, 2020; v1 submitted 12 July, 2020; originally announced July 2020.

On the Independence of Fundamental Decompositions of Power-Law Kinetic Systems

Authors: Bryan S. Hernandez

Abstract: The fundamental decomposition of a chemical reaction network (CRN) is induced by partitioning the reaction set into "fundamental classes". It was the basis of the Higher Deficiency Algorithm for mass action systems of Ji and Feinberg, and the Multistationarity Algorithm for power-law kinetic systems of Hernandez et al. In addition to our previous work, we provide important properties of the indepe… ▽ More The fundamental decomposition of a chemical reaction network (CRN) is induced by partitioning the reaction set into "fundamental classes". It was the basis of the Higher Deficiency Algorithm for mass action systems of Ji and Feinberg, and the Multistationarity Algorithm for power-law kinetic systems of Hernandez et al. In addition to our previous work, we provide important properties of the independence (i.e., the network's stoichiometric subspace is the direct sum of the subnetworks' stoichiometric subspaces) and the incidence-independence (i.e., the image of the network's incidence map is the direct sum of the incidence maps' images of the subnetworks) of these decompositions. Feinberg established the essential relationship between independent decompositions and the set of positive equilibria of a network, which we call the Feinberg Decomposition Theorem (FDT). Moreover, Farinas et al. recently documented its version for incidence-independence. Fundamental decomposition divides the network into subnetworks of deficiency either 0 or 1 only. Hence, available results for lower deficiency networks, such as the Deficiency Zero Theorem (DZT), can be used. These justify the study of independent fundamental decompositions. A MATLAB program which (i) computes the subnetworks of a CRN under the fundamental decomposition and (ii) is useful for determining whether the decomposition is independent and incidence-independent is also created. Finally, we provide the following solution for determining multistationarity of CRNs with the following steps: (1) the use of the program, (2) the application of available results for CRNs with deficiency 0 or 1 (e.g., DZT), and (3) the use of FDT. We illustrate the solution by showing that the generalization of a subnetwork of Schmitz's carbon cycle model by Hernandez et al., endowed with mass action kinetics, has no capacity for multistationarity. △ Less

Submitted 16 February, 2020; v1 submitted 14 January, 2020; originally announced January 2020.

Comments: arXiv admin note: substantial text overlap with arXiv:1908.04593

Journal ref: MATCH Commun. Math. Comput. Chem. 84(1) (2020) 57-84

Fundamental Decompositions and Multistationarity of Power-Law Kinetic Systems

Authors: Bryan S. Hernandez, Eduardo R. Mendoza, Aurelio A. de los Reyes V

Abstract: The fundamental decomposition of a chemical reaction network (also called its "$\mathscr{F}$-decomposition") is the set of subnetworks generated by the partition of its set of reactions into the "fundamental classes" introduced by Ji and Feinberg in 2011 as the basis of their "higher deficiency algorithm" for mass action systems. The first part of this paper studies the properties of the… ▽ More The fundamental decomposition of a chemical reaction network (also called its "$\mathscr{F}$-decomposition") is the set of subnetworks generated by the partition of its set of reactions into the "fundamental classes" introduced by Ji and Feinberg in 2011 as the basis of their "higher deficiency algorithm" for mass action systems. The first part of this paper studies the properties of the $\mathscr{F}$-decomposition, in particular, its independence (i.e., the network's stoichiometric subspace is the direct sum of the subnetworks' stoichiometric subspaces) and its incidence-independence (i.e., the image of the network's incidence map is the direct sum of the incidence maps' images of the subnetworks). We derive necessary and sufficient conditions for these properties and identify network classes where the $\mathscr{F}$-decomposition coincides with other known decompositions. The second part of the paper applies the above-mentioned results to improve the Multistationarity Algorithm for power-law kinetic systems (MSA), a general computational approach that we introduced in previous work. We show that for systems with non-reactant determined interactions but with an independent $\mathscr{F}$-decomposition, the transformation to a dynamically equivalent system with reactant-determined interactions -- required in the original MSA -- is not necessary. We illustrate this improvement with the subnetwork of Schmitz's carbon cycle model recently analyzed by Fortun et al. △ Less

Submitted 18 February, 2020; v1 submitted 13 August, 2019; originally announced August 2019.

Journal ref: MATCH Commun. Math. Comput. Chem. 83(2) (2020) 403-434

AI vs Humans for the diagnosis of sleep apnea

Authors: Valentin Thorey, Albert Bou Hernandez, Pierrick J. Arnal, Emmanuel H. During

Abstract: Polysomnography (PSG) is the gold standard for diagnosing sleep obstructive apnea (OSA). It allows monitoring of breathing events throughout the night. The detection of these events is usually done by trained sleep experts. However, this task is tedious, highly time-consuming and subject to important inter-scorer variability. In this study, we adapted our state-of-the-art deep learning method for… ▽ More Polysomnography (PSG) is the gold standard for diagnosing sleep obstructive apnea (OSA). It allows monitoring of breathing events throughout the night. The detection of these events is usually done by trained sleep experts. However, this task is tedious, highly time-consuming and subject to important inter-scorer variability. In this study, we adapted our state-of-the-art deep learning method for sleep event detection, DOSED, to the detection of sleep breathing events in PSG for the diagnosis of OSA. We used a dataset of 52 PSG recordings with apnea-hypopnea event scoring from 5 trained sleep experts. We assessed the performance of the automatic approach and compared it to the inter-scorer performance for both the diagnosis of OSA severity and, at the microscale, for the detection of single breathing events. We observed that human sleep experts reached an average accuracy of 75\% while the automatic approach reached 81\% for sleep apnea severity diagnosis. The F1 score for individual event detection was 0.55 for experts and 0.57 for the automatic approach, on average. These results demonstrate that the automatic approach can perform at a sleep expert level for the diagnosis of OSA. △ Less

Submitted 20 June, 2019; originally announced June 2019.

Comments: copyright 2019 IEEE. Accepted for publication in 41st International Engineering in Medicine and Biology Conference (EMBC), July 23-27, 2019

A Computational Approach to Multistationarity of Power-Law Kinetic Systems

Authors: Bryan S. Hernandez, Eduardo R. Mendoza, Aurelio A. de los Reyes V

Abstract: This paper presents a computational solution to determine if a chemical reaction network endowed with power-law kinetics (PLK system) has the capacity for multistationarity, i.e., whether there exist positive rate constants such that the corresponding differential equations admit multiple positive steady states within a stoichiometric class. The approach, which is called the "Multistationarity Alg… ▽ More This paper presents a computational solution to determine if a chemical reaction network endowed with power-law kinetics (PLK system) has the capacity for multistationarity, i.e., whether there exist positive rate constants such that the corresponding differential equations admit multiple positive steady states within a stoichiometric class. The approach, which is called the "Multistationarity Algorithm for PLK systems" (MSA), combines (i) the extension of the "higher deficiency algorithm" of Ji and Feinberg for mass action to PLK systems with reactant-determined interactions, and (ii) a method that transforms any PLK system to a dynamically equivalent one with reactant-determined interactions. Using this algorithm, we obtain two new results: the monostationarity of a popular model of anaerobic yeast fermentation pathway, and the multistationarity of a global carbon cycle model with climate engineering, both in the generalized mass action format of biochemical systems theory. We also provide examples of the broader scope of our approach for deficiency one PLK systems in comparison to the extension of Feinberg's "deficiency one algorithm" to such systems. △ Less

Submitted 18 February, 2020; v1 submitted 6 February, 2019; originally announced February 2019.

Journal ref: J Math Chem 58 (2020) 56-87

Stabilizing predictive control with persistence of excitation for constrained linear systems

Authors: Bernardo A. Hernandez, Paul A. Trodden

Abstract: A new adaptive predictive controller for constrained linear systems is presented. The main feature of the proposed controller is the partition of the input in two components. The first part is used to persistently excite the system, in order to guarantee accurate and convergent parameter estimates in a deterministic framework. An MPC-inspired receding horizon optimization problem is developed to a… ▽ More A new adaptive predictive controller for constrained linear systems is presented. The main feature of the proposed controller is the partition of the input in two components. The first part is used to persistently excite the system, in order to guarantee accurate and convergent parameter estimates in a deterministic framework. An MPC-inspired receding horizon optimization problem is developed to achieve the required excitation in a manner that is optimal for the plant. The remaining control action is employed by a conventional tube MPC controller to regulate the plant in the presence of parametric uncertainty and the excitation generated for estimation purposes. Constraint satisfaction, robust exponential stability, and convergence of the estimates are guaranteed under design conditions mildly more demanding than that of standard MPC implementations. △ Less

Submitted 20 April, 2018; originally announced April 2018.

Comments: Manuscript submitted for consideration at an Elsevier journal on September 2017

Decentralized, Robust and Efficient Services for an Autonomous and Real-time Urban Crisis Management

Authors: Frédéric Le Mouël, Carlos Barrios Hernández, Oscar Carrillo, Gabriel Pedraza

Abstract: The globalization of trade and the organization of work are currently causing a large migratory flow towards the cities. This growth of cities requires new urban planning where digital tools take a preponderant place to capture data and understand and decide in face of changes. These tools however hardly resist to natural disasters, terrorism, accidents, etc. Based on the expertise of the CITI lab… ▽ More The globalization of trade and the organization of work are currently causing a large migratory flow towards the cities. This growth of cities requires new urban planning where digital tools take a preponderant place to capture data and understand and decide in face of changes. These tools however hardly resist to natural disasters, terrorism, accidents, etc. Based on the expertise of the CITI laboratory of INSA Lyon and SC3 of the Industrial University of Santander, we propose to create the ALERT project - Autonomous Liable Emergency service in Real Time - with decentralized, reliable and efficient services, physically close to the citizens, taking decisions locally, in a relevant manner without risk of disconnection with a central authority. These information gathering and decision-making will involve the population with participatory and social approaches. △ Less

Submitted 6 March, 2017; originally announced March 2017.

Comments: in French. Colloque international interdisciplinaire Colombie -- France " La Ville-Région durable comme projet: défis actuels. Regards croisés et perspectives ", Mar 2017, Bogota, Colombie

Distributed MPC: Guaranteeing Global Stabilizability from Locally Designed Tubes

Authors: Bernardo Hernandez, Pablo Baldivieso, Paul Trodden

Abstract: This paper studies a fundamental relation that exists between stabilizability assumptions usually employed in distributed model predictive control implementations, and the corresponding notions of invariance implicit in such controllers. The relation is made explicit in the form of a theorem that presents sufficient conditions for global stabilizability. It is shown that constraint admissibility o… ▽ More This paper studies a fundamental relation that exists between stabilizability assumptions usually employed in distributed model predictive control implementations, and the corresponding notions of invariance implicit in such controllers. The relation is made explicit in the form of a theorem that presents sufficient conditions for global stabilizability. It is shown that constraint admissibility of local robust controllers is sufficient for the global closed-loop system to be stable, and how these controllers are related to more complex forms of control such as tube-based distributed model predictive control implementations. △ Less

Submitted 2 November, 2016; originally announced November 2016.

Comments: Submitted to the IFAC World Congress 2017 (Toulouse)

Nested Distributed Model Predictive Control

Authors: Pablo R Baldivieso Monasterios, Bernardo Hernandez, Paul A Trodden

Abstract: We propose a distributed model predictive control approach for linear time-invariant systems coupled via dynamics. The proposed approach uses the tube MPC concept for robustness to handle the disturbances induced by mutual interactions between subsystems; however, the main novelty here is to replace the conventional linear disturbance rejection controller with a second MPC controller, as is done i… ▽ More We propose a distributed model predictive control approach for linear time-invariant systems coupled via dynamics. The proposed approach uses the tube MPC concept for robustness to handle the disturbances induced by mutual interactions between subsystems; however, the main novelty here is to replace the conventional linear disturbance rejection controller with a second MPC controller, as is done in tube-based nonlinear MPC. In the distributed setting, this has the advantages that the disturbance rejection controller is able to consider the plans of neighbours, and the reliance on explicit robust invariant sets is removed. △ Less

Submitted 28 October, 2016; originally announced October 2016.

Comments: 7 pages, 1 figure, submitted to IFAC 2017 conference

Distributed Model Predictive Control Using a Chain of Tubes

Authors: Bernardo Hernandez, Paul Trodden

Abstract: A new distributed MPC algorithm for the regulation of dynamically coupled subsystems is presented in this paper. The current control action is computed via two robust controllers working in a nested fashion. The inner controller builds a nominal reference trajectory from a decentralized perspective. The outer controller uses this information to take into account the effects of the coupling and gen… ▽ More A new distributed MPC algorithm for the regulation of dynamically coupled subsystems is presented in this paper. The current control action is computed via two robust controllers working in a nested fashion. The inner controller builds a nominal reference trajectory from a decentralized perspective. The outer controller uses this information to take into account the effects of the coupling and generate a distributed control action. The tube-based approach to robustness is employed. A supplementary constraint is included in the outer optimization problem to provide recursive feasibility of the overall controller △ Less

Submitted 27 March, 2017; v1 submitted 7 March, 2016; originally announced March 2016.

Comments: Accepted for presentation at the UKACC CONTROL 2016 conference (Belfast, UK)

Bayesian Additive Regression Trees using Bayesian Model Averaging

Authors: Belinda Hernández, Adrian E. Raftery, Stephen R. Pennington, Andrew C. Parnell

Abstract: Bayesian Additive Regression Trees (BART) is a statistical sum of trees model. It can be considered a Bayesian version of machine learning tree ensemble methods where the individual trees are the base learners. However for data sets where the number of variables $p$ is large (e.g. $p>5,000$) the algorithm can become prohibitively expensive, computationally. Another method which is popular for hi… ▽ More Bayesian Additive Regression Trees (BART) is a statistical sum of trees model. It can be considered a Bayesian version of machine learning tree ensemble methods where the individual trees are the base learners. However for data sets where the number of variables $p$ is large (e.g. $p>5,000$) the algorithm can become prohibitively expensive, computationally. Another method which is popular for high dimensional data is random forests, a machine learning algorithm which grows trees using a greedy search for the best split points. However, as it is not a statistical model, it cannot produce probabilistic estimates or predictions. We propose an alternative algorithm for BART called BART-BMA, which uses Bayesian Model Averaging and a greedy search algorithm to produce a model which is much more efficient than BART for datasets with large $p$. BART-BMA incorporates elements of both BART and random forests to offer a model-based algorithm which can deal with high-dimensional data. We have found that BART-BMA can be run in a reasonable time on a standard laptop for the "small $n$ large $p$" scenario which is common in many areas of bioinformatics. We showcase this method using simulated data and data from two real proteomic experiments; one to distinguish between patients with cardiovascular disease and controls and another to classify agressive from non-agressive prostate cancer. We compare our results to their main competitors. Open source code written in R and Rcpp to run BART-BMA can be found at: https://github.com/BelindaHernandez/BART-BMA.git △ Less

Submitted 8 July, 2015; v1 submitted 1 July, 2015; originally announced July 2015.

Propagation of time-truncated Airy-type pulses in media with quadratic and cubic dispersion

Authors: José Angel Borda Hernández, Michel Zamboni-Rached, Amr Shaarawi, Ioannis M. Besieris

Abstract: In this paper, we describe analytically the propagation of Airy-type pulses truncated by a finite-time aperture when second and third order dispersion effects are considered. The mathematical method presented here, based on the superposition of exponentially truncated Airy pulses, is very effective, allowing us to avoid the use of time-consuming numerical simulations. We analyze the behavior of th… ▽ More In this paper, we describe analytically the propagation of Airy-type pulses truncated by a finite-time aperture when second and third order dispersion effects are considered. The mathematical method presented here, based on the superposition of exponentially truncated Airy pulses, is very effective, allowing us to avoid the use of time-consuming numerical simulations. We analyze the behavior of the time truncated Ideal-Airy pulse and also the interesting case of a time truncated Airy pulse with a "defect" in its initial profile, which reveals the self-healing property of this kind of pulse solution. △ Less

Submitted 7 June, 2015; originally announced June 2015.

Comments: 9 pages. 5 figures

Persistently Exciting Tube MPC

Authors: Bernardo Hernandez, Paul Trodden

Abstract: This paper presents a new approach to deal with the dual problem of system identification and regulation. The main feature consists of breaking the control input to the system into a regulator part and a persistently exciting part. The former is used to regulate the plant using a robust MPC formulation, in which the latter is treated as a bounded additive disturbance. The identification process is… ▽ More This paper presents a new approach to deal with the dual problem of system identification and regulation. The main feature consists of breaking the control input to the system into a regulator part and a persistently exciting part. The former is used to regulate the plant using a robust MPC formulation, in which the latter is treated as a bounded additive disturbance. The identification process is executed by a simple recursive least squares algorithm. In order to guarantee sufficient excitation for the identification, an additional non-convex constraint is enforced over the persistently exciting part. △ Less

Submitted 27 March, 2017; v1 submitted 21 May, 2015; originally announced May 2015.

Comments: Accepted for presentation at the American Control Conference 2016 (Boston, MA)

ACM Class: I.2.8; J.2

Demonstration of distinct semiconducting transport characteristics of monolayer graphene functionalized via plasma activation of substrate surfaces

Authors: Po-Hsiang Wang, Fu-Yu Shih, Shao-Yu Chen, Alvin B. Hernandez, Po-Hsun Ho, Lo-Yueh Chang, Chia-Hao Chen, Hsiang-Chih Chiu, Chun-Wei Chen, Wei-Hua Wang

Abstract: We report semiconducting behavior of monolayer graphene enabled through plasma activation of substrate surfaces. The graphene devices are fabricated by mechanical exfoliation onto pre-processed SiO2/Si substrates. Contrary to pristine graphene, these graphene samples exhibit a transport gap as well as nonlinear transfer characteristics, a large on/off ratio of 600 at cryogenic temperatures, and an… ▽ More We report semiconducting behavior of monolayer graphene enabled through plasma activation of substrate surfaces. The graphene devices are fabricated by mechanical exfoliation onto pre-processed SiO2/Si substrates. Contrary to pristine graphene, these graphene samples exhibit a transport gap as well as nonlinear transfer characteristics, a large on/off ratio of 600 at cryogenic temperatures, and an insulating-like temperature dependence. Raman spectroscopic characterization shows evidence of sp3 hybridization of C atoms in the samples of graphene on activated SiO2/Si substrates. We analyze the hopping transport at low temperatures, and weak localization observed from magnetotransport measurements, suggesting a correlation between carrier localization and the sp3-type defects in the functionalized graphene. The present study demonstrates the functionalization of graphene using a novel substrate surface-activation method for future graphene-based applications. △ Less

Submitted 19 May, 2015; originally announced May 2015.

Matrix Codes as Ideals for Grassmannian Codes and their Weight Properties

Authors: Bryan Hernandez, Virgilio Sison

Abstract: A systematic way of constructing Grassmannian codes endowed with the subspace distance as lifts of matrix codes over the prime field $GF(p)$ is introduced. The matrix codes are $GF(p)$-subspaces of the ring $M_2(GF(p))$ of $2 \times 2$ matrices over $GF(p)$ on which the rank metric is applied, and are generated as one-sided proper principal ideals by idempotent elements of $M_2(GF(p))$. Furthermor… ▽ More A systematic way of constructing Grassmannian codes endowed with the subspace distance as lifts of matrix codes over the prime field $GF(p)$ is introduced. The matrix codes are $GF(p)$-subspaces of the ring $M_2(GF(p))$ of $2 \times 2$ matrices over $GF(p)$ on which the rank metric is applied, and are generated as one-sided proper principal ideals by idempotent elements of $M_2(GF(p))$. Furthermore a weight function on the non-commutative matrix ring $M_2(GF(p))$, $q$ a power of $p$, is studied in terms of the egalitarian and homogeneous conditions. The rank weight distribution of $M_2(GF(q))$ is completely determined by the general linear group $GL(2,q)$. Finally a weight function on subspace codes is analogously defined and its egalitarian property is examined. △ Less

Submitted 20 February, 2015; originally announced February 2015.

MSC Class: 94B05

Grassmannian Codes as Lifts of Matrix Codes Derived as Images of Linear Block Codes over Finite Fields

Authors: Bryan Hernandez, Virgilio Sison

Abstract: Let $p$ be a prime such that $p \equiv 2$ or $3$ mod $5$. Linear block codes over the non-commutative matrix ring of $2 \times 2$ matrices over the prime field $GF(p)$ endowed with the Bachoc weight are derived as isometric images of linear block codes over the Galois field $GF(p^2)$ endowed with the Hamming metric. When seen as rank metric codes, this family of matrix codes satisfies the Singleto… ▽ More Let $p$ be a prime such that $p \equiv 2$ or $3$ mod $5$. Linear block codes over the non-commutative matrix ring of $2 \times 2$ matrices over the prime field $GF(p)$ endowed with the Bachoc weight are derived as isometric images of linear block codes over the Galois field $GF(p^2)$ endowed with the Hamming metric. When seen as rank metric codes, this family of matrix codes satisfies the Singleton bound and thus are maximum rank distance codes, which are then lifted to form a special class of subspace codes, the Grassmannian codes, that meet the anticode bound. These so-called anticode-optimal Grassmannian codes are associated in some way with complete graphs. New examples of these maximum rank distance codes and anticode-optimal Grassmannian codes are given. △ Less

Submitted 14 February, 2015; originally announced February 2015.

MSC Class: 94B05; 94B65

Meromorphic continuation of functions and arbitrary distribution of interpolation points

Authors: Manuel Bello Hernández, Bernardo de la Calle Ysern

Abstract: We characterize the region of meromorphic continuation of an analytic function $f$ in terms of the geometric rate of convergence on a compact set of sequences of multi-point rational interpolants of $f$. The rational approximants have a bounded number of poles and the distribution of interpolation points is arbitrary. We characterize the region of meromorphic continuation of an analytic function $f$ in terms of the geometric rate of convergence on a compact set of sequences of multi-point rational interpolants of $f$. The rational approximants have a bounded number of poles and the distribution of interpolation points is arbitrary. △ Less

Submitted 23 November, 2012; originally announced November 2012.

Comments: 18 pages

MSC Class: 41A21; 41A25; 41A27

Cryogenic characterization of the Planck sorption cooler system flight model

Authors: G. Morgante, D. Pearson, F. Melot, P. Stassi, L. Terenzi, P. Wilson, B. Hernandez, L. Wade, A. Gregorio, M. Bersanelli, C. Butler, N. Mandolesi

Abstract: This paper is part of the Prelaunch status LFI papers published on JINST: http://www.iop.org/EJ/journal/-page=extra.proc5/1748-0221 Two continuous closed-cycle hydrogen Joule-Thomson (J-T) sorption coolers have been fabricated and assembled by the Jet Propulsion Laboratory (JPL) for the European Space Agency (ESA) Planck mission. Each refrigerator has been designed to provide a total of ~ 1W o… ▽ More This paper is part of the Prelaunch status LFI papers published on JINST: http://www.iop.org/EJ/journal/-page=extra.proc5/1748-0221 Two continuous closed-cycle hydrogen Joule-Thomson (J-T) sorption coolers have been fabricated and assembled by the Jet Propulsion Laboratory (JPL) for the European Space Agency (ESA) Planck mission. Each refrigerator has been designed to provide a total of ~ 1W of cooling power at two instrument interfaces: they directly cool the Planck Low Frequency Instrument (LFI) around 20K while providing a pre-cooling stage for a 4 K J-T mechanical refrigerator for the High Frequency Instrument (HFI). After sub-system level validation at JPL, the cryocoolers have been delivered to ESA in 2005. In this paper we present the results of the cryogenic qualification and test campaigns of the Nominal Unit on the flight model spacecraft performed at the CSL (Centre Spatial de Liege) facilities in 2008. Test results in terms of input power, cooling power, temperature, and temperature fluctuations over the flight allowable ranges for these interfaces are reported and analyzed with respect to mission requirements. △ Less

Submitted 26 January, 2010; originally announced January 2010.

Comments: This is an author-created, un-copyedited version of an article accepted for publication in Journal of Instrumentation. IOP Publishing Ltd is not responsible for any errors or omissions in this version of the manuscript or any version derived from it. The definitive publisher authenticated version is available online at 10.1088/1748-0221/4/12/T12016

Journal ref: G Morgante et al 2009 JINST 4 T12016