-
Molecular relaxation by reverse diffusion with time step prediction
Authors:
Khaled Kahouli,
Stefaan Simon Pierre Hessmann,
Klaus-Robert Müller,
Shinichi Nakajima,
Stefan Gugler,
Niklas Wolf Andreas Gebauer
Abstract:
Molecular relaxation, finding the equilibrium state of a non-equilibrium structure, is an essential component of computational chemistry to understand reactivity. Classical force field methods often rely on insufficient local energy minimization, while neural network force field models require large labeled datasets encompassing both equilibrium and non-equilibrium structures. As a remedy, we prop…
▽ More
Molecular relaxation, finding the equilibrium state of a non-equilibrium structure, is an essential component of computational chemistry to understand reactivity. Classical force field methods often rely on insufficient local energy minimization, while neural network force field models require large labeled datasets encompassing both equilibrium and non-equilibrium structures. As a remedy, we propose MoreRed, molecular relaxation by reverse diffusion, a conceptually novel and purely statistical approach where non-equilibrium structures are treated as noisy instances of their corresponding equilibrium states. To enable the denoising of arbitrarily noisy inputs via a generative diffusion model, we further introduce a novel diffusion time step predictor. Notably, MoreRed learns a simpler pseudo potential energy surface instead of the complex physical potential energy surface. It is trained on a significantly smaller, and thus computationally cheaper, dataset consisting of solely unlabeled equilibrium structures, avoiding the computation of non-equilibrium structures altogether. We compare MoreRed to classical force fields, equivariant neural network force fields trained on a large dataset of equilibrium and non-equilibrium data, as well as a semi-empirical tight-binding model. To assess this quantitatively, we evaluate the root-mean-square deviation between the found equilibrium structures and the reference equilibrium structures as well as their DFT energies.
△ Less
Submitted 16 April, 2024;
originally announced April 2024.
-
Adaptive integration of history variables in constrained mixture models for organ-scale growth and remodeling
Authors:
Amadeus M. Gebauer,
Martin R. Pfaller,
Jason M. Szafron,
Wolfgang A. Wall
Abstract:
In the last decades, many computational models have been developed to predict soft tissue growth and remodeling (G&R). The constrained mixture theory describes fundamental mechanobiological processes in soft tissue G&R and has been widely adopted in cardiovascular models of G&R. However, even after two decades of work, large organ-scale models are rare, mainly due to high computational costs (mode…
▽ More
In the last decades, many computational models have been developed to predict soft tissue growth and remodeling (G&R). The constrained mixture theory describes fundamental mechanobiological processes in soft tissue G&R and has been widely adopted in cardiovascular models of G&R. However, even after two decades of work, large organ-scale models are rare, mainly due to high computational costs (model evaluation and memory consumption), especially in long-range simulations. We propose two strategies to adaptively integrate history variables in constrained mixture models to enable large organ-scale simulations of G&R. Both strategies exploit that the influence of deposited tissue on the current mixture decreases over time through degradation. One strategy is independent of external loading, allowing the estimation of the computational resources ahead of the simulation. The other adapts the history snapshots based on the local mechanobiological environment so that the additional integration errors can be controlled and kept negligibly small, even in G&R scenarios with severe perturbations. We analyze the adaptively integrated constrained mixture model on a tissue patch for a parameter study and show the performance under different G&R scenarios. To confirm that adaptive strategies enable large organ-scale examples, we show simulations of different hypertension conditions with a real-world example of a biventricular heart discretized with a finite element mesh. In our example, adaptive integrations sped up simulations by a factor of three and reduced memory requirements to one-sixth. The reduction of the computational costs gets even more pronounced for simulations over longer periods. Adaptive integration of the history variables allows studying more finely resolved models and longer G&R periods while computational costs are drastically reduced and largely constant in time.
△ Less
Submitted 11 July, 2024; v1 submitted 15 April, 2024;
originally announced April 2024.
-
SchNetPack 2.0: A neural network toolbox for atomistic machine learning
Authors:
Kristof T. Schütt,
Stefaan S. P. Hessmann,
Niklas W. A. Gebauer,
Jonas Lederer,
Michael Gastegger
Abstract:
SchNetPack is a versatile neural networks toolbox that addresses both the requirements of method development and application of atomistic machine learning. Version 2.0 comes with an improved data pipeline, modules for equivariant neural networks as well as a PyTorch implementation of molecular dynamics. An optional integration with PyTorch Lightning and the Hydra configuration framework powers a f…
▽ More
SchNetPack is a versatile neural networks toolbox that addresses both the requirements of method development and application of atomistic machine learning. Version 2.0 comes with an improved data pipeline, modules for equivariant neural networks as well as a PyTorch implementation of molecular dynamics. An optional integration with PyTorch Lightning and the Hydra configuration framework powers a flexible command-line interface. This makes SchNetPack 2.0 easily extendable with custom code and ready for complex training task such as generation of 3d molecular structures.
△ Less
Submitted 11 December, 2022;
originally announced December 2022.
-
A homogenized constrained mixture model of cardiac growth and remodeling: Analyzing mechanobiological stability and reversal
Authors:
Amadeus M. Gebauer,
Martin R. Pfaller,
Fabian A. Braeu,
Christian J. Cyron,
Wolfgang A. Wall
Abstract:
Cardiac growth and remodeling (G&R) patterns change ventricular size, shape, and function both globally and locally. Biomechanical, neurohormonal, and genetic stimuli drive these patterns through changes in myocyte dimension and fibrosis. We propose a novel microstructure-motivated model that predicts organ-scale G&R in the heart based on the homogenized constrained mixture theory. Previous models…
▽ More
Cardiac growth and remodeling (G&R) patterns change ventricular size, shape, and function both globally and locally. Biomechanical, neurohormonal, and genetic stimuli drive these patterns through changes in myocyte dimension and fibrosis. We propose a novel microstructure-motivated model that predicts organ-scale G&R in the heart based on the homogenized constrained mixture theory. Previous models, based on the kinematic growth theory, reproduced consequences of G&R in bulk myocardial tissue by prescribing the direction and extent of growth but neglected underlying cellular mechanisms. In our model, the direction and extent of G&R emerge naturally from intra- and extra cellular turnover processes in myocardial tissue constituents and their preferred homeostatic stretch state. We additionally propose a method to obtain a mechanobiologically equilibrated reference configuration. We test our model on an idealized 3D left ventricular geometry and demonstrate that our model aims to maintain tensional homeostasis in hypertension conditions. In a stability map, we identify regions of stable and unstable G&R from an identical parameter set with varying systolic pressures and growth factors. Furthermore, we show the extent of G&R reversal after returning the systolic pressure to baseline following stage 1 and 2 hypertension. A realistic model of organ-scale cardiac G&R has the potential to identify patients at risk of heart failure, enable personalized cardiac therapies, and facilitate the optimal design of medical devices.
△ Less
Submitted 3 May, 2023; v1 submitted 23 March, 2022;
originally announced March 2022.
-
Inverse design of 3d molecular structures with conditional generative neural networks
Authors:
Niklas W. A. Gebauer,
Michael Gastegger,
Stefaan S. P. Hessmann,
Klaus-Robert Müller,
Kristof T. Schütt
Abstract:
The rational design of molecules with desired properties is a long-standing challenge in chemistry. Generative neural networks have emerged as a powerful approach to sample novel molecules from a learned distribution. Here, we propose a conditional generative neural network for 3d molecular structures with specified chemical and structural properties. This approach is agnostic to chemical bonding…
▽ More
The rational design of molecules with desired properties is a long-standing challenge in chemistry. Generative neural networks have emerged as a powerful approach to sample novel molecules from a learned distribution. Here, we propose a conditional generative neural network for 3d molecular structures with specified chemical and structural properties. This approach is agnostic to chemical bonding and enables targeted sampling of novel molecules from conditional distributions, even in domains where reference calculations are sparse. We demonstrate the utility of our method for inverse design by generating molecules with specified motifs or composition, discovering particularly stable molecules, and jointly targeting multiple electronic properties beyond the training regime.
△ Less
Submitted 22 December, 2021; v1 submitted 10 September, 2021;
originally announced September 2021.
-
Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules
Authors:
Niklas W. A. Gebauer,
Michael Gastegger,
Kristof T. Schütt
Abstract:
Deep learning has proven to yield fast and accurate predictions of quantum-chemical properties to accelerate the discovery of novel molecules and materials. As an exhaustive exploration of the vast chemical space is still infeasible, we require generative models that guide our search towards systems with desired properties. While graph-based models have previously been proposed, they are restricte…
▽ More
Deep learning has proven to yield fast and accurate predictions of quantum-chemical properties to accelerate the discovery of novel molecules and materials. As an exhaustive exploration of the vast chemical space is still infeasible, we require generative models that guide our search towards systems with desired properties. While graph-based models have previously been proposed, they are restricted by a lack of spatial information such that they are unable to recognize spatial isomerism and non-bonded interactions. Here, we introduce a generative neural network for 3d point sets that respects the rotational invariance of the targeted structures. We apply it to the generation of molecules and demonstrate its ability to approximate the distribution of equilibrium structures using spatial metrics as well as established measures from chemoinformatics. As our model is able to capture the complex relationship between 3d geometry and electronic properties, we bias the distribution of the generator towards molecules with a small HOMO-LUMO gap - an important property for the design of organic solar cells.
△ Less
Submitted 9 January, 2020; v1 submitted 2 June, 2019;
originally announced June 2019.
-
Generating equilibrium molecules with deep neural networks
Authors:
Niklas W. A. Gebauer,
Michael Gastegger,
Kristof T. Schütt
Abstract:
Discovery of atomistic systems with desirable properties is a major challenge in chemistry and material science. Here we introduce a novel, autoregressive, convolutional deep neural network architecture that generates molecular equilibrium structures by sequentially placing atoms in three-dimensional space. The model estimates the joint probability over molecular configurations with tractable cond…
▽ More
Discovery of atomistic systems with desirable properties is a major challenge in chemistry and material science. Here we introduce a novel, autoregressive, convolutional deep neural network architecture that generates molecular equilibrium structures by sequentially placing atoms in three-dimensional space. The model estimates the joint probability over molecular configurations with tractable conditional probabilities which only depend on distances between atoms and their nuclear charges. It combines concepts from state-of-the-art atomistic neural networks with auto-regressive generative models for images and speech. We demonstrate that the architecture is capable of generating molecules close to equilibrium for constitutional isomers of C$_7$O$_2$H$_{10}$.
△ Less
Submitted 26 October, 2018;
originally announced October 2018.
-
Deep underground rotation measurements: GINGERino ring laser gyroscope in Gran Sasso
Authors:
Jacopo Belfi,
Nicolo' Beverini,
Filippo Bosi,
Giorgio Carelli,
Davide Cuccato,
Gaetano De Luca,
Angela Di Virgilio,
Andre' Gebauer,
Enrico Maccioni,
Antonello Ortolan,
Alberto Porzio,
Gilberto Saccorotti,
Andreino Simonelli,
Giuseppe Terreni
Abstract:
GINGERino is a large frame laser gyroscope investigating the ground motion in the most inner part of the underground international laboratory of the Gran Sasso, in central Italy. It consists of a square ring laser with a $3.6$ m side. Several days of continuous measurements have been collected, with the apparatus running unattended. The power spectral density in the seismic bandwidth is at the lev…
▽ More
GINGERino is a large frame laser gyroscope investigating the ground motion in the most inner part of the underground international laboratory of the Gran Sasso, in central Italy. It consists of a square ring laser with a $3.6$ m side. Several days of continuous measurements have been collected, with the apparatus running unattended. The power spectral density in the seismic bandwidth is at the level of $10^{-10} \rm{(rad/s)/\sqrt{Hz}}$. A maximum resolution of $30\,\rm{prad/s}$ is obtained with an integration time of few hundred seconds. The ring laser routinely detects seismic rotations induced by both regional earthquakes and teleseisms. A broadband seismic station is installed on the same structure of the gyroscope. First analysis of the correlation between the rotational and the translational signal are presented.
△ Less
Submitted 9 February, 2017;
originally announced February 2017.
-
First Results of GINGERino, a deep underground ringlaser
Authors:
J. Belfi,
N. Beverini,
F. Bosi,
G. Carelli,
D. Cuccato,
G. De Luca,
A. Di Virgilio,
A. Gebauer,
E. Maccioni,
A. Ortolan,
A. Porzio,
R. Santagata,
A. Simonelli,
G. Terreni
Abstract:
Large ring-laser gyroscopes are capable of measuring angular rotations with a precision well below fractions of $prad/s$, not far from $10^{-14}$ $rad/s$, the accuracy required for General Relativity tests, this is what the GINGER (Gyroscope-IN-GEneral-Relativity) experiment is aiming for. These features do not guarantee the possibility of measuring the General Relativity Lense--Thirring effect, t…
▽ More
Large ring-laser gyroscopes are capable of measuring angular rotations with a precision well below fractions of $prad/s$, not far from $10^{-14}$ $rad/s$, the accuracy required for General Relativity tests, this is what the GINGER (Gyroscope-IN-GEneral-Relativity) experiment is aiming for. These features do not guarantee the possibility of measuring the General Relativity Lense--Thirring effect, that manifests itself as a tiny ($\approx 10^{-9} \times Ω_E$) perturbation of the Earth rotation rate. An underground location being in principle less affected by external local disturbances represents a good candidate for housing such a challenging experiment. GINGERino is a test apparatus to investigate the residual local disturbances in the most inner part of the underground international laboratory of the GranSasso (LNGS). It consists of a square ring laser with a $3.6$ m side. The instrument has been tailored to be the larger allowed by the particular location inside the laboratory. Its main objective is to measure the very low frequency rotational motions, in order to prove that LNGS is a suitable location for very low noise measurements and, possibly, General Relativity tests. Aside this main goal, GINGERino will provide unique data for geodesy and geophysics. Its installation has been completed during 2015. Since then, several long set of data have been collected, and the apparatus has been continuously running unattended for more than one week. The typical power spectrum sensitivity was a few $ 10^{-10} rad/s/\sqrt(Hz)$, with integration time not longer than tens of seconds. Improvements of the apparatus are ongoing in order to improve the integration time.
△ Less
Submitted 29 March, 2016; v1 submitted 12 January, 2016;
originally announced January 2016.
-
Towards a Holographic Model of the QCD Phase Diagram
Authors:
Nick Evans,
Astrid Gebauer,
Keun-Young Kim,
Maria Magou
Abstract:
We describe the temperature-chemical potential phase diagrams of holographic models of a range of strongly coupled gauge theories that display chiral symmetry breaking/restoration transitions. The models are based on the D3/probe-D7 system but with a phenomenologically chosen running coupling/dilaton profile. We realize chiral phase transitions with either temperature or density that are first or…
▽ More
We describe the temperature-chemical potential phase diagrams of holographic models of a range of strongly coupled gauge theories that display chiral symmetry breaking/restoration transitions. The models are based on the D3/probe-D7 system but with a phenomenologically chosen running coupling/dilaton profile. We realize chiral phase transitions with either temperature or density that are first or second order by changing the dilaton profile. Although the models are only caricatures of QCD they show that holographic models can capture many aspects of the QCD phase diagram and hint at the dependence on the running coupling.
△ Less
Submitted 4 April, 2012; v1 submitted 12 September, 2011;
originally announced September 2011.
-
Measuring Gravito-magnetic Effects by Multi Ring-Laser Gyroscope
Authors:
F. Bosi,
G. Cella,
A. Di Virgilio,
A. Ortolan,
A. Porzio,
S. Solimeno,
M. Cerdonio,
J. P. Zendri,
M. Allegrini,
J. Belfi,
N. Beverini,
B. Bouhadef,
l G. Carelli,
I. Ferrante,
E. Maccioni,
R. Passaquieti,
F. Stefani,
M. L. Ruggiero,
A. Tartaglia,
K. U. Schreiber,
A. Gebauer,
J-P. R. Wells
Abstract:
We propose an under-ground experiment to detect the general relativistic effects due to the curvature of space-time around the Earth (de Sitter effect) and to rotation of the planet (dragging of the inertial frames or Lense-Thirring effect). It is based on the comparison between the IERS value of the Earth rotation vector and corresponding measurements obtained by a tri-axial laser detector of rot…
▽ More
We propose an under-ground experiment to detect the general relativistic effects due to the curvature of space-time around the Earth (de Sitter effect) and to rotation of the planet (dragging of the inertial frames or Lense-Thirring effect). It is based on the comparison between the IERS value of the Earth rotation vector and corresponding measurements obtained by a tri-axial laser detector of rotation. The proposed detector consists of six large ring-lasers arranged along three orthogonal axes.
In about two years of data taking, the 1% sensitivity required for the measurement of the Lense-Thirring drag can be reached with square rings of 6 $m$ side, assuming a shot noise limited sensitivity ($ 20 prad/s/\sqrt{Hz}$). The multi-gyros system, composed of rings whose planes are perpendicular to one or the other of three orthogonal axes, can be built in several ways. Here, we consider cubic and octahedron structures. The symmetries of the proposed configurations provide mathematical relations that can be used to study the stability of the scale factors, the relative orientations or the ring-laser planes, very important to get rid of systematics in long-term measurements, which are required in order to determine the relativistic effects.
△ Less
Submitted 24 June, 2011;
originally announced June 2011.
-
E, B, μ, T Phase Structure of the D3/D7 Holographic Dual
Authors:
Nick Evans,
Astrid Gebauer,
Keun-Young Kim
Abstract:
The large N_c N=4 gauge theory with quenched N=2 quark matter displays chiral symmetry breaking in the presence of a magnetic field. We previously studied the temperature and chemical potential phase structure of this theory in the grand canonical ensemble - here we, in addition, include the effect of an electric field which acts to counter chiral symmetry breaking by disassociating mesons. We com…
▽ More
The large N_c N=4 gauge theory with quenched N=2 quark matter displays chiral symmetry breaking in the presence of a magnetic field. We previously studied the temperature and chemical potential phase structure of this theory in the grand canonical ensemble - here we, in addition, include the effect of an electric field which acts to counter chiral symmetry breaking by disassociating mesons. We compute using the gravity dual based on the D3/probe-D7 brane system. The theory displays two transition at one of which chiral symmetry is restored. At the other transition density switches on, the mesons of the theory become unstable and a current forms, making it a conductor-insulator transition. Through the temperature, electric field, chemical potential volume (at fixed magnetic field parallel to the electric field) these transitions can coincide or separate at critical points, and be first order or second order. We map out this full phase structure which provides varied computable examples relevant to strongly coupled gauge theories and potentially condensed matter systems.
△ Less
Submitted 29 March, 2011;
originally announced March 2011.
-
A laser gyroscope system to detect the Gravito-Magnetic effect on Earth
Authors:
A. Di Virgilio,
K. U. Schreiber,
A. Gebauer,
J-P. R. Wells,
A. Tartaglia,
J. Belfi,
N. Beverini,
A. Ortolan
Abstract:
Large scale square ring laser gyros with a length of four meters on each side are approaching a sensitivity of 1x10^-11 rad/s/sqrt(Hz). This is about the regime required to measure the gravitomagnetic effect (Lense Thirring) of the Earth. For an ensemble of linearly independent gyros each measurement signal depends upon the orientation of each single axis gyro with respect to the rotational axis o…
▽ More
Large scale square ring laser gyros with a length of four meters on each side are approaching a sensitivity of 1x10^-11 rad/s/sqrt(Hz). This is about the regime required to measure the gravitomagnetic effect (Lense Thirring) of the Earth. For an ensemble of linearly independent gyros each measurement signal depends upon the orientation of each single axis gyro with respect to the rotational axis of the Earth. Therefore at least 3 gyros are necessary to reconstruct the complete angular orientation of the apparatus. In general, the setup consists of several laser gyroscopes (we would prefer more than 3 for sufficient redundancy), rigidly referenced to each other. Adding more gyros for one plane of observation provides a cross-check against intra-system biases and furthermore has the advantage of improving the signal to noise ratio by the square root of the number of gyros. In this paper we analyze a system of two pairs of identical gyros (twins) with a slightly different orientation with respect to the Earth axis. The twin gyro configuration has several interesting properties. The relative angle can be controlled and provides a useful null measurement. A quadruple twin system could reach a 1% sensitivity after 3:2 years of data, provided each square ring has 6 m length on a side, the system is shot noise limited and there is no source for 1/f- noise.
△ Less
Submitted 12 July, 2010;
originally announced July 2010.
-
Phase diagram of the D3/D5 system in a magnetic field and a BKT transition
Authors:
Nick Evans,
Astrid Gebauer,
Keun-Young Kim,
Maria Magou
Abstract:
We study the full temperature and chemical potential dependence of the D3/D5 2+1 dimensional theory in the presence of a magnetic field. The theory displays separate transitions associated with chiral symmetry breaking and melting of the bound states. We display the phase diagram which has areas with first and second order transitions meeting at two critical points similar to that of the D3/D7 sys…
▽ More
We study the full temperature and chemical potential dependence of the D3/D5 2+1 dimensional theory in the presence of a magnetic field. The theory displays separate transitions associated with chiral symmetry breaking and melting of the bound states. We display the phase diagram which has areas with first and second order transitions meeting at two critical points similar to that of the D3/D7 system. In addition there is the recently reported BKT transition at zero temperature leading to distinct structure at low temperatures.
△ Less
Submitted 7 March, 2011; v1 submitted 13 March, 2010;
originally announced March 2010.
-
Holographic Description of the Phase Diagram of a Chiral Symmetry Breaking Gauge Theory
Authors:
Nick Evans,
Astrid Gebauer,
Keun-Young Kim,
Maria Magou
Abstract:
The large N_c N=4 gauge theory with quenched N=2 quark matter in the presence of a magnetic field displays chiral symmetry breaking. We study the temperature and chemical potential dependence of this theory using its gravity dual (based on the D3/D7 brane system). With massless quarks, at zero chemical potential, the theory displays a first order thermal transition where chiral symmetry is restore…
▽ More
The large N_c N=4 gauge theory with quenched N=2 quark matter in the presence of a magnetic field displays chiral symmetry breaking. We study the temperature and chemical potential dependence of this theory using its gravity dual (based on the D3/D7 brane system). With massless quarks, at zero chemical potential, the theory displays a first order thermal transition where chiral symmetry is restored and simultaneously the mesons of the theory melt. At zero temperature, these transitions with chemical potential are second order and occur at different chemical potential values. Between the three there are two tri-critical points, the positions of which we identify. At finite quark mass the second order transition for chiral symmetry becomes a cross over and there is a critical point at the end of the first order transition, while the meson melting transition remains similar to the massless quark case. We track the movement of the critical points as the mass is raised relative to the magnetic field.
△ Less
Submitted 11 June, 2010; v1 submitted 9 February, 2010;
originally announced February 2010.
-
Holographic integral equations and walking technicolour
Authors:
Raul Alvares,
Nick Evans,
Astrid Gebauer,
George James Weatherill
Abstract:
We study chiral symmetry breaking in the holographic D3-D7 system in a simple model with an arbitrary running coupling. We derive equations for the D7 embedding and show there is a light pion. In particular we present simple integral equations, involving just the running coupling and the quark self energy, for the quark condensate and the pion decay constant. We compare these to the Pagels-Stoka…
▽ More
We study chiral symmetry breaking in the holographic D3-D7 system in a simple model with an arbitrary running coupling. We derive equations for the D7 embedding and show there is a light pion. In particular we present simple integral equations, involving just the running coupling and the quark self energy, for the quark condensate and the pion decay constant. We compare these to the Pagels-Stokar or constituent quark model equivalent. We discuss the implications for walking Technicolour theories. We also perform a similar analysis in the four dimensional field theory whose dual is the non-supersymmetric D3-D5 system and propose that it represents a walking theory in which the quark condensate has dimension 2 + sqrt{3}.
△ Less
Submitted 10 November, 2009; v1 submitted 16 October, 2009;
originally announced October 2009.
