Showing 1–27 of 27 results for author: Käser, S

Outlier-Detection for Reactive Machine Learned Potential Energy Surfaces

Authors: Luis Itza Vazquez-Salazar, Silvan Käser, Markus Meuwly

Abstract: Uncertainty quantification (UQ) to detect samples with large expected errors (outliers) is applied to reactive molecular potential energy surfaces (PESs). Three methods - Ensembles, Deep Evidential Regression (DER), and Gaussian Mixture Models (GMM) - were applied to the H-transfer reaction between ${\it syn-}$Criegee and vinyl hydroxyperoxide. The results indicate that ensemble models provide the… ▽ More Uncertainty quantification (UQ) to detect samples with large expected errors (outliers) is applied to reactive molecular potential energy surfaces (PESs). Three methods - Ensembles, Deep Evidential Regression (DER), and Gaussian Mixture Models (GMM) - were applied to the H-transfer reaction between ${\it syn-}$Criegee and vinyl hydroxyperoxide. The results indicate that ensemble models provide the best results for detecting outliers, followed by GMM. For example, from a pool of 1000 structures with the largest uncertainty, the detection quality for outliers is $\sim 90$ \% and $\sim 50$ \%, respectively, if 25 or 1000 structures with large errors are sought. On the contrary, the limitations of the statistical assumptions of DER greatly impacted its prediction capabilities. Finally, a structure-based indicator was found to be correlated with large average error, which may help to rapidly classify new structures into those that provide an advantage for refining the neural network. △ Less

Submitted 27 February, 2024; originally announced February 2024.

Numerical Accuracy Matters: Applications of Machine Learned Potential Energy Surfaces

Authors: Silvan Käser, Markus Meuwly

Abstract: The role of numerical accuracy in training and evaluating neural network-based potential energy surfaces is examined for different experimental observables. For observables that require third- and fourth-order derivatives of the total energy with respect to Cartesian coordinates single-precision arithmetics as is typically used in ML-based approaches is insufficient and leads to roughness of the u… ▽ More The role of numerical accuracy in training and evaluating neural network-based potential energy surfaces is examined for different experimental observables. For observables that require third- and fourth-order derivatives of the total energy with respect to Cartesian coordinates single-precision arithmetics as is typically used in ML-based approaches is insufficient and leads to roughness of the underlying PES as is explicitly demonstrated. Increasing the numerical accuracy to double-precision yields a smooth PES with higher-order derivatives that are numerically stable and yield meaningful anharmonic frequencies and tunneling splitting as is demonstrated for H$_2$CO and malonaldehyde. For molecular dynamics simulations, which only require first-order derivatives, single-precision arithmetics appears to be sufficient, though. △ Less

Submitted 29 November, 2023; originally announced November 2023.

On the Effect of Aleatoric and Epistemic Errors on the Learnability and Quality of NN-based Potential Energy Surfaces

Authors: S. Goswami, S. Käser, R. J. Bemish, M. Meuwly

Abstract: The effect of noise in the input data for learning potential energy surfaces (PESs) based on neural networks for chemical applications is assessed. Noise in energies and forces can result from aleatoric and epistemic errors in the quantum chemical reference calculations. Statistical (aleatoric) noise arises for example due to the need to set convergence thresholds in the self consistent field (SCF… ▽ More The effect of noise in the input data for learning potential energy surfaces (PESs) based on neural networks for chemical applications is assessed. Noise in energies and forces can result from aleatoric and epistemic errors in the quantum chemical reference calculations. Statistical (aleatoric) noise arises for example due to the need to set convergence thresholds in the self consistent field (SCF) iterations whereas systematic (epistemic) noise is due to, {\it inter alia}, particular choices of basis sets in the calculations. The two molecules considered here as proxies are H$_{2}$CO and HONO which are examples for single- and multi-reference problems, respectively, for geometries around the minimum energy structure. For H$_2$CO it is found that adding noise to energies with magnitudes representative of single-point calculations does not deteriorate the quality of the final PESs whereas increasing the noise level commensurate with electronic structure calculations for more complicated, e.g. metal-containing, systems is expected to have a more notable effect. However, the effect of noise on the forces is more noticeable. On the other hand, for HONO which requires a multi-reference treatment, a clear correlation between model quality and the degree of multi-reference character as measured by the $T_1$ amplitude is found. It is concluded that for chemically "simple" cases the effect of aleatoric and epistemic noise is manageable without evident deterioration of the trained model - although the quality of the forces is important. However, considerably more care needs to be exercised for situations in which multi-reference effects are present. △ Less

Submitted 10 September, 2023; originally announced September 2023.

PhysNet Meets CHARMM: A Framework for Routine Machine Learning / Molecular Mechanics Simulations

Authors: Kaisheng Song, Silvan Käser, Kai Töpfer, Luis Itza Vazquez-Salazar, Markus Meuwly

Abstract: Full dimensional potential energy surfaces (PESs) based on machine learning (ML) techniques provide means for accurate and efficient molecular simulations in the gas- and condensed-phase for various experimental observables ranging from spectroscopy to reaction dynamics. Here, the MLpot extension with PhysNet as the ML-based model for a PES is introduced into the newly developed pyCHARMM API. To i… ▽ More Full dimensional potential energy surfaces (PESs) based on machine learning (ML) techniques provide means for accurate and efficient molecular simulations in the gas- and condensed-phase for various experimental observables ranging from spectroscopy to reaction dynamics. Here, the MLpot extension with PhysNet as the ML-based model for a PES is introduced into the newly developed pyCHARMM API. To illustrate conceiving, validating, refining and using a typical workflow, para-chloro-phenol is considered as an example. The main focus is on how to approach a concrete problem from a practical perspective and applications to spectroscopic observables and the free energy for the -OH torsion in solution are discussed in detail. For the computed IR spectra in the fingerprint region the computations for para-chloro-phenol in water are in good qualitative agreement with experiment carried out in CCl$_4$. Also, relative intensities are largely consistent with experimental findings. The barrier for rotation of the -OH group increases from $\sim 3.5$ kcal/mol in the gas phase to $\sim 4.1$ kcal/mol from simulations in water due to favourable H-bonding interactions of the -OH group with surrounding water molecules. △ Less

Submitted 25 April, 2023; originally announced April 2023.

Comments: 38 pages, 11 figures

Conformational and state-specific effects in reactions of 2,3-dibromobutadiene with Coulomb-crystallized calcium ions

Authors: Ardita Kilaj, Silvan Käser, Jia Wang, Patrik Straňák, Max Schwilk, Lei Xu, O. Anatole von Lilienfeld, Jochen Küpper, Markus Meuwly, Stefan Willitsch

Abstract: Recent advances in experimental methodology enabled studies of the quantum-state and conformational dependence of chemical reactions under precisely controlled conditions in the gas phase. Here, we generated samples of selected gauche and s-trans 2,3-dibromobutadiene (DBB) by electrostatic deflection in a molecular beam and studied their reaction with Coulomb crystals of laser-cooled… ▽ More Recent advances in experimental methodology enabled studies of the quantum-state and conformational dependence of chemical reactions under precisely controlled conditions in the gas phase. Here, we generated samples of selected gauche and s-trans 2,3-dibromobutadiene (DBB) by electrostatic deflection in a molecular beam and studied their reaction with Coulomb crystals of laser-cooled $\mathrm{Ca^{+}}$ ions in an ion trap. The rate coefficients for the total reaction were found to strongly depend on both the conformation of DBB and the electronic state of $\mathrm{Ca^{+}}$. In the $\mathrm{(4p)~^{2}P_{1/2}}$ and $\mathrm{(3d)~^{2}D_{3/2}}$ excited states of $\mathrm{Ca^{+}}$, the reaction is capture-limited and faster for the gauche conformer due to long-range ion-dipole interactions. In the $\mathrm{(4s)~^{2}S_{1/2}}$ ground state of $\mathrm{Ca^{+}}$, the reaction rate for s-trans DBB still conforms with the capture limit, while that for gauche DBB is strongly suppressed. The experimental observations were analysed with the help of adiabatic capture theory, ab-initio calculations and reactive molecular dynamics simulations on a machine-learned full-dimensional potential energy surface of the system. The theory yields near-quantitative agreement for s-trans-DBB, but overestimates the reactivity of the gauche-conformer compared to the experiment. The present study points to the important role of molecular geometry even in strongly reactive exothermic systems and illustrates striking differences in the reactivity of individual conformers in gas-phase ion-molecule reactions. △ Less

Submitted 21 March, 2023; originally announced March 2023.

Transfer-Learned Potential Energy Surfaces: Towards Microsecond-Scale Molecular Dynamics Simulations in the Gas Phase at CCSD(T) Quality

Authors: Silvan Käser, Markus Meuwly

Abstract: The rise of machine learning has greatly influenced the field of computational chemistry, and that of atomistic molecular dynamics simulations in particular. One of its most exciting prospects is the development of accurate, full-dimensional potential energy surfaces (PESs) for molecules and clusters, which, however, often require thousands to tens of thousands of ab initio data points restricting… ▽ More The rise of machine learning has greatly influenced the field of computational chemistry, and that of atomistic molecular dynamics simulations in particular. One of its most exciting prospects is the development of accurate, full-dimensional potential energy surfaces (PESs) for molecules and clusters, which, however, often require thousands to tens of thousands of ab initio data points restricting the community to medium sized molecules and/or lower levels of theory (e.g. DFT). Transfer learning, which improves a global PES from a lower to a higher level of theory, offers a data efficient alternative requiring only a fraction of the high level data (on the order of 100 are found to be sufficient for malonaldehyde). The present work demonstrates that even with Hartree-Fock theory and a double-zeta basis set as the lower level model, transfer learning yields CCSD(T)-level quality for H-transfer barrier energies, harmonic frequencies and H-transfer tunneling splittings. Most importantly, finite-temperature molecular dynamics simulations on the sub-microsecond time scale in the gas phase are possible and the infrared spectra determined from the transfer learned PESs are in good agreement with experiment. It is concluded that routine, long-time atomistic simulations on PESs fulfilling CCSD(T)-standards become possible. △ Less

Submitted 21 March, 2023; originally announced March 2023.

Distinct spin and orbital dynamics in Sr$_{2}$RuO$_{4}$

Authors: H. Suzuki, L. Wang, J. Bertinshaw, H. U. R. Strand, S. Käser, M. Krautloher, Z. Yang, N. Wentzell, O. Parcollet, F. Jerzembeck, N. Kikugawa, A. P. Mackenzie, A. Georges, P. Hansmann, H. Gretarsson, B. Keimer

Abstract: The unconventional superconductor Sr$_2$RuO$_4$ has long served as a benchmark for theories of correlated-electron materials. The determination of the superconducting pairing mechanism requires detailed experimental information on collective bosonic excitations as potential mediators of Cooper pairing. We have used Ru $L_3$-edge resonant inelastic x-ray scattering to obtain comprehensive maps of t… ▽ More The unconventional superconductor Sr$_2$RuO$_4$ has long served as a benchmark for theories of correlated-electron materials. The determination of the superconducting pairing mechanism requires detailed experimental information on collective bosonic excitations as potential mediators of Cooper pairing. We have used Ru $L_3$-edge resonant inelastic x-ray scattering to obtain comprehensive maps of the electronic excitations of Sr$_2$RuO$_4$ over the entire Brillouin zone. We observe multiple branches of dispersive spin and orbital excitations associated with distinctly different energy scales. The spin and orbital dynamical response functions calculated within the dynamical mean-field theory are in excellent agreement with the experimental data. Our results highlight the Hund metal nature of Sr$_{2}$RuO$_{4}$ and provide key information for the understanding of its unconventional superconductivity. △ Less

Submitted 30 November, 2022; originally announced December 2022.

Journal ref: Nat. Commun. 14, 7042 (2023)

Novel ion imaging concept based on time-of-flight measurements with low gain avalanche detectors

Authors: Felix Ulrich-Pur, Thomas Bergauer, Albert Hirtl, Christian Irmler, Stefanie Kaser, Florian Pitters, Simon Rit

Abstract: Treatment planning in ion beam therapy requires accurate knowledge of the relative stop** power (RSP) distribution within the patient. Currently, RSP maps are obtained via conventional x-ray computed tomography (CT) by converting the measured attenuation coefficients of photons into RSP values for ions. Alternatively, to avoid conversion errors that are inherent to this method, ion computed tomo… ▽ More Treatment planning in ion beam therapy requires accurate knowledge of the relative stop** power (RSP) distribution within the patient. Currently, RSP maps are obtained via conventional x-ray computed tomography (CT) by converting the measured attenuation coefficients of photons into RSP values for ions. Alternatively, to avoid conversion errors that are inherent to this method, ion computed tomography (iCT) can be used since it allows determining the RSP directly. In typical iCT systems, which usually consist of a tracking system and a separate residual energy detector, the RSP is obtained by measuring the particle trajectory and the corresponding water equivalent path length (WEPL) of single ions travelling through the patient. Within this work, we explore a novel iCT approach which does not require a residual energy detector. Instead, the WEPL is estimated indirectly by determining the change in time of flight (TOF) due to the energy loss along the ion's path. For this purpose, we have created a Geant4 model of a TOF-iCT system based on low gain avalanche detectors (LGADs), which are fast 4D-tracking detectors that can measure the time of arrival and position of individual particles with high spatial and time precision. To assess the performance of this TOF-iCT concept, we determined the RSP resolution and accuracy for different system settings using the Catphan CTP404 sensitometry phantom. Within the set of investigated system parameters, the lower limit of the RSP accuracy was found at 0.91%, demonstrating the proof-of-principle of this novel TOF-iCT concept. The main advantage of using this approach is that it could potentially facilitate clinical integration due to its compact design, which, however, requires experimental verification and an improvement of the current WEPL calibration procedure. △ Less

Submitted 27 September, 2022; originally announced September 2022.

Neural Network Potentials for Chemistry: Concepts, Applications and Prospects

Authors: Silvan Käser, Luis Itza Vazquez-Salazar, Markus Meuwly, Kai Töpfer

Abstract: Artificial Neural Networks (ANN) are already heavily involved in methods and applications for frequent tasks in the field of computational chemistry such as representation of potential energy surfaces (PES) and spectroscopic predictions. This perspective provides an overview of the foundations of neural network-based full-dimensional potential energy surfaces, their architectures, underlying conce… ▽ More Artificial Neural Networks (ANN) are already heavily involved in methods and applications for frequent tasks in the field of computational chemistry such as representation of potential energy surfaces (PES) and spectroscopic predictions. This perspective provides an overview of the foundations of neural network-based full-dimensional potential energy surfaces, their architectures, underlying concepts, their representation and applications to chemical systems. Methods for data generation and training procedures for PES construction are discussed and means for error assessment and refinement through transfer learning are presented. A selection of recent results illustrates the latest improvements regarding accuracy of PES representations and system size limitations in dynamics simulations, but also NN application enabling direct prediction of physical results without dynamics simulations. The aim is to provide an overview for the current state-of-the-art NN approaches in computational chemistry and also to point out the current challenges in enhancing reliability and applicability of NN methods on larger scale. △ Less

Submitted 22 December, 2022; v1 submitted 23 September, 2022; originally announced September 2022.

Transfer learning for affordable and high quality tunneling splittings from instanton calculations

Authors: Silvan Käser, Jeremy O. Richardson, Markus Meuwly

Abstract: The combination of transfer learning (TL) a low level potential energy surface (PES) to a higher level of electronic structure theory together with ring-polymer instanton (RPI) theory is explored and applied to malonaldehyde. The RPI approach provides a semiclassical approximation of the tunneling splitting and depends sensitively on the accuracy of the PES. With second order Møller-Plesset pertur… ▽ More The combination of transfer learning (TL) a low level potential energy surface (PES) to a higher level of electronic structure theory together with ring-polymer instanton (RPI) theory is explored and applied to malonaldehyde. The RPI approach provides a semiclassical approximation of the tunneling splitting and depends sensitively on the accuracy of the PES. With second order Møller-Plesset perturbation theory (MP2) as the low-level (LL) model and energies and forces from coupled cluster singles, doubles and perturbative triples (CCSD(T)) as the high-level (HL) model, it is demonstrated that CCSD(T) information from only 25 to 50 judiciously selected structures along and around the instanton path suffice to reach HL-accuracy for the tunneling splitting. In addition, the global quality of the HL-PES is demonstrated through a mean average error of 0.3 kcal/mol for energies up to 40 kcal/mol above the minimum energy structure (a factor of 2 higher than the energies employed during TL) and $< 2 $ cm$^{-1}$ for harmonic frequencies compared with computationally challenging normal mode calculations at the CCSD(T) level. △ Less

Submitted 2 August, 2022; originally announced August 2022.

Hydration Dynamics and IR Spectroscopy of 4-Fluorophenol

Authors: Seyedeh Maryam Salehi, Silvan Käser, Kai Töpfer, Polydefkis Diamantis, Rolf Pfister, Peter Hamm, Ursula Röthlisberger, Markus Meuwly

Abstract: Halogenated groups are relevant in pharmaceutical applications and potentially useful spectroscopic probes for infrared spectroscopy. In this work, the structural dynamics and infrared spectroscopy of $para$-fluorophenol (F-PhOH) and phenol (PhOH) is investigated in the gas phase and in water using a combination of experiment and molecular dynamics (MD) simulations. The gas phase and solvent dynam… ▽ More Halogenated groups are relevant in pharmaceutical applications and potentially useful spectroscopic probes for infrared spectroscopy. In this work, the structural dynamics and infrared spectroscopy of $para$-fluorophenol (F-PhOH) and phenol (PhOH) is investigated in the gas phase and in water using a combination of experiment and molecular dynamics (MD) simulations. The gas phase and solvent dynamics around F-PhOH and PhOH is characterized from atomistic simulations using empirical energy functions with point charges or multipoles for the electrostatics, Machine-Learning (ML) based parametrization and with full $\textit{ab initio}$ (QM) and mixed Quantum Mechanical/Molecular Mechanics (QM/MM) simulations with a particular focus on the CF- and OH-stretch region. The CF-stretch band is heavily mixed with other modes whereas the OH-stretch in solution displays a characteristic high-frequency peak around 3600 cm$^{-1}$ most likely associated with the -OH group of PhOH and F-PhOH together with a characteristic progression below 3000 cm$^{-1}$ due to coupling with water modes which is also reproduced by several of the simulations. Solvent and radial distribution functions indicate that the CF-site is largely hydrophobic except for simulations using point charges which renders them unsuited for correctly describing hydration and dynamics around fluorinated sites. △ Less

Submitted 21 June, 2022; originally announced June 2022.

Comments: Main Manuscript: 41 pages and 6 figures, SI: 10 pages and 10 figures

The Economics of Automated Market Makers

Authors: Robin Fritsch, Samuel Käser, Roger Wattenhofer

Abstract: This paper studies the question whether automated market maker protocols such as Uniswap can sustainably retain a portion of their trading fees for the protocol. We approach the problem by modelling how to optimally choose a pool's take rate, i.e\ the fraction of fee revenue that remains with the protocol, in order to maximize the protocol's revenue. The model suggest that if AMMs have a portion o… ▽ More This paper studies the question whether automated market maker protocols such as Uniswap can sustainably retain a portion of their trading fees for the protocol. We approach the problem by modelling how to optimally choose a pool's take rate, i.e\ the fraction of fee revenue that remains with the protocol, in order to maximize the protocol's revenue. The model suggest that if AMMs have a portion of loyal trade volume, they can sustainably set a non-zero take rate, even without losing liquidity to competitors with a zero take rate. Furthermore, we determine the optimal take rate depending on a number of model parameters including how much loyal trade volume pools have and how high the competitors' take rates are. △ Less

Submitted 9 June, 2022; originally announced June 2022.

Journal ref: Proceedings of the 4th ACM Conference on Advances in Financial Technologies (2022) 102-110

Double Proton Transfer in Hydrated Formic Acid Dimer: Interplay of Spatial Symmetry and Solvent-Generated Force on Reactivity

Authors: Kai Töpfer, Silvan Käser, Markus Meuwly

Abstract: The double proton transfer (DPT) reaction in hydrated formic acid dimer (FAD) is investigated at molecular-level detail. For this, a global and reactive machine learned (ML) potential energy surface (PES) is developed to run extensive (more than 100 ns) mixed ML/MM molecular dynamics (MD) simulations in explicit molecular mechanics (MM) solvent at MP2-quality for the solute. Simulations with fixed… ▽ More The double proton transfer (DPT) reaction in hydrated formic acid dimer (FAD) is investigated at molecular-level detail. For this, a global and reactive machine learned (ML) potential energy surface (PES) is developed to run extensive (more than 100 ns) mixed ML/MM molecular dynamics (MD) simulations in explicit molecular mechanics (MM) solvent at MP2-quality for the solute. Simulations with fixed - as in a conventional empirical force field - and conformationally fluctuating - as available from the ML-based PES - charge models for FAD shows significant impact on the competition between DPT and dissociation of FAD into two formic acid monomers. With increasing temperature the barrier height for DPT in solution changes by about 10% ($\sim 1$ kcal/mol) between 300 K and 600 K. The rate for DPT is largest, $\sim 1$ ns$^{-1}$, at 350 K and decreases for higher temperatures due to destabilisation and increased probability for dissociation of FAD. The water solvent is found to promote the first proton transfer by exerting a favourable solvent-induced Coulomb force along the O-H$\cdots$O hydrogen bond whereas the second proton transfer is significantly controlled by the O-O separation and other conformational degrees of freedom. Double proton transfer in hydrated FAD is found to involve a subtle interplay and balance between structural and electrostatic factors. △ Less

Submitted 13 May, 2022; v1 submitted 22 October, 2021; originally announced October 2021.

Transfer Learned Potential Energy Surfaces: Accurate Anharmonic Vibrational Dynamics and Dissociation Energies for the Formic Acid Monomer and Dimer

Authors: Silvan Käser, Markus Meuwly

Abstract: The vibrational dynamics of formic acid monomer (FAM) and dimer (FAD) is investigated from machine-learned potential energy surfaces at the MP2 (PES$_{\rm MP2}$) and transfer-learned (PES$_{\rm TL}$) to the CCSD(T) levels of theory. The normal modes and anharmonic frequencies of all modes below 2000 cm$^{-1}$ agree favourably with experiment whereas the OH-stretch mode is challenging for FAM and F… ▽ More The vibrational dynamics of formic acid monomer (FAM) and dimer (FAD) is investigated from machine-learned potential energy surfaces at the MP2 (PES$_{\rm MP2}$) and transfer-learned (PES$_{\rm TL}$) to the CCSD(T) levels of theory. The normal modes and anharmonic frequencies of all modes below 2000 cm$^{-1}$ agree favourably with experiment whereas the OH-stretch mode is challenging for FAM and FAD from normal mode analyses and finite-temperature MD simulations. VPT2 calculations on PES$_{\rm TL}$ for FAM reproduce the experimental OH frequency to within 22 cm$^{-1}$. For FAD the VPT2 calculations find the high-frequency OH stretch at 3011cm$^{-1}$, compared with an experimentally reported, broad ($\sim 100$ cm$^{-1}$) absorption band with center frequency estimated at $\sim 3050$ cm$^{-1}$. In agreement with earlier reports, MD simulations at higher temperature shift the position of the OH-stretch in FAM to the red, consistent with improved sampling of the anharmonic regions of the PES. However, for FAD the OH-stretch shifts to the blue and for temperatures higher than 1000 K the dimer partly or fully dissociates using PES$_{\rm TL}$. Including zero-point energy corrections from diffusion Monte Carlo simulations for FAM and FAD and corrections due to basis set superposition and completeness errors yield a dissociation energy of $D_0 = -14.23 \pm 0.08$ kcal/mol compared with an experimentally determined value of $-14.22 \pm 0.12$ kcal/mol. △ Less

Submitted 21 September, 2021; v1 submitted 17 September, 2021; originally announced September 2021.

Comments: Adapted order of pages

Feasibility study of a proton CT system based on 4D-tracking and residual energy determination via time-of-flight

Authors: Felix Ulrich-Pur, Thomas Bergauer, Alexander Burker, Albert Hirtl, Christian Irmler, Stefanie Kaser, Florian Pitters, Simon Rit

Abstract: For dose calculations in ion beam therapy, it is vital to accurately determine the relative stop** power (RSP) distribution within the treated volume. Currently, RSP values are extrapolated from Hounsfield units (HU), measured with x-ray computed tomography (CT), which entails RSP inaccuracies due to conversion errors. A suitable method to improve the treatment plan accuracy is proton computed t… ▽ More For dose calculations in ion beam therapy, it is vital to accurately determine the relative stop** power (RSP) distribution within the treated volume. Currently, RSP values are extrapolated from Hounsfield units (HU), measured with x-ray computed tomography (CT), which entails RSP inaccuracies due to conversion errors. A suitable method to improve the treatment plan accuracy is proton computed tomography (pCT). A typical pCT system consists of a tracking system and a separate residual energy (or range) detector to measure the RSP distribution directly. This paper introduces a novel pCT system based on a single detector technology, namely low gain avalanche detectors (LGADs). LGADs are fast 4D-tracking detectors, which can be used to simultaneously measure the particle position and time with precise timing and spatial resolution. In contrast to standard pCT systems, the residual energy is determined via a time-of-flight (TOF) measurement between different 4D-tracking stations. The design parameters for a realistic proton computed tomography system based on 4D-tracking detectors were studied and optimized using Monte Carlo simulations. The RSP accuracy and RSP resolution were measured inside the inserts of the CTP404 phantom to estimate the performance of the pCT system. After introducing a dedicated calibration procedure for the TOF calorimeter, RSP accuracies < 0.6 % could be achieved. Furthermore, the design parameters with the strongest impact on the RSP resolution were identified and a strategy to improve RSP resolution is proposed. △ Less

Submitted 10 September, 2021; originally announced September 2021.

Comments: Preprint submitted to Physics in Medicine and Biology. IOP Publishing Ltd is not responsible for any errors or omissions in this version of the manuscript or any version derived from it

A Proton Computed Tomography Demonstrator for Stop** Power Measurements

Authors: Felix Ulrich-Pur, Thomas Bergauer, Alexander Burker, Albert Hirtl, Christian Irmler, Stefanie Kaser, Florian Pitters

Abstract: Particle therapy is an established method to treat deep-seated tumours using accelerator-produced ion beams. For treatment planning, the precise knowledge of the relative stop** power (RSP) within the patient is vital. Conversion errors from x-ray computed tomography (CT) measurements to RSP introduce uncertainties in the applied dose distribution. Using a proton computed tomography (pCT) system… ▽ More Particle therapy is an established method to treat deep-seated tumours using accelerator-produced ion beams. For treatment planning, the precise knowledge of the relative stop** power (RSP) within the patient is vital. Conversion errors from x-ray computed tomography (CT) measurements to RSP introduce uncertainties in the applied dose distribution. Using a proton computed tomography (pCT) system to measure the SP directly could potentially increase the accuracy of treatment planning. A pCT demonstrator, consisting of double-sided silicon strip detectors (DSSD) as tracker and plastic scintillator slabs coupled to silicon photomultipliers (SiPM) as a range telescope, was developed. After a significant hardware upgrade of the range telescope, a 3D tomogram of an aluminium stair phantom was recorded at the MedAustron facility in Wiener Neustadt, Austria. In total, 80 projections with 6.5x10^5 primary events were acquired and used for the reconstruction of the RSP distribution in the phantom. After applying a straight-line approximation for the particle path inside the phantom, the most probable value (MPV) of the RSP distribution could be measured with an accuracy of 0.59%. The RSP resolution inside the phantom was only 9.3% due to a limited amount of projections and measured events per projection. △ Less

Submitted 24 June, 2021; originally announced June 2021.

Comments: Preprint submitted to the open-access Journal of Physics: Conference Series. (TIPP2021 conference proceedings). IOP Publishing Ltd is not responsible for any errors or omissions in this version of the manuscript or any version derived from it

Inter-orbital singlet pairing in Sr$_2$RuO$_4$: a Hund's superconductor

Authors: Stefan Käser, Hugo U. R. Strand, Nils Wentzell, Antoine Georges, Olivier Parcollet, Philipp Hansmann

Abstract: We study the superconducting gap function of Sr$_2$RuO$_4$. By solving the linearized Eliashberg equation with a correlated pairing vertex extracted from a dynamical mean-field calculation we identify the dominant pairing channels. An analysis of the candidate gap functions in orbital and quasiparticle band basis reveals that an inter-orbital singlet pairing of even parity is in agreement with exp… ▽ More We study the superconducting gap function of Sr$_2$RuO$_4$. By solving the linearized Eliashberg equation with a correlated pairing vertex extracted from a dynamical mean-field calculation we identify the dominant pairing channels. An analysis of the candidate gap functions in orbital and quasiparticle band basis reveals that an inter-orbital singlet pairing of even parity is in agreement with experimental observations. It reconciles in particular the occurrence of a two-component order parameter with the presence of line-nodes of quasiparticles along the c-axis in the superconducting phase. The strong angular dependence of the gap along the Fermi surface is in stark contrast to its quasi-locality when expressed in the orbital basis. We identify local inter-orbital spin correlations as the driving force for the pairing and thus reveal the continuation of Hund's physics into the superconducting phase. △ Less

Submitted 18 May, 2021; originally announced May 2021.

Comments: 20 pages, 12 figures

MP2 Is Not Good Enough: Transfer Learning ML Models for Accurate VPT2 Frequencies

Authors: Silvan Käser, Eric Boittier, Meenu Upadhyay, Markus Meuwly

Abstract: The calculation of the anharmonic modes of small to medium sized molecules for assigning experimentally measured frequencies to the corresponding type of molecular motions is computationally challenging at sufficiently high levels of quantum chemical theory. Here, a practical and affordable way to calculate coupled-cluster quality anharmonic frequencies using second order vibrational perturbation… ▽ More The calculation of the anharmonic modes of small to medium sized molecules for assigning experimentally measured frequencies to the corresponding type of molecular motions is computationally challenging at sufficiently high levels of quantum chemical theory. Here, a practical and affordable way to calculate coupled-cluster quality anharmonic frequencies using second order vibrational perturbation theory (VPT2) from machine-learned models is presented. The approach, referred to as "NN + VPT2", uses a high-dimensional neural network (PhysNet) to learn potential energy surfaces (PESs) at different levels of theory from which harmonic and VPT2 frequencies can be efficiently determined. The NN + VPT2 approach is applied to eight small to medium sized molecules (H$_2$CO, trans-HONO, HCOOH, CH$_3$OH, CH$_3$CHO, CH$_3$NO$_2$, CH$_3$COOH and CH$_3$CONH$_2$) and frequencies are reported from NN-learned models at the MP2/aug-cc-pVTZ, CCSD(T)/aug-cc-pVTZ and CCSD(T)-F12/aug-cc-pVTZ-F12 levels of theory. For the largest molecules and at the highest levels of theory, transfer learning (TL) is used to determine the necessary full-dimensional, near-equilibrium PESs. Overall, NN + VPT2 yields anharmonic frequencies to within 20 cm$^{-1}$ of experimentally determined frequencies for close to 90 % of the modes for the highest quality PES available and to within 10 cm$^{-1}$ for more than 60 % of the modes. For the MP2 PESs only around 60 % of the NN + VPT2 frequencies were within 20~cm$^{-1}$ of the experiment, with outliers up to 150 cm$^{-1}$ compared with experiment. It is also demonstrated that the approach allows to provide correct assignments for strongly interacting modes such as the OH bending and the OH torsional modes in formic acid monomer and the CO-stretch and OH-bend mode in acetic acid. △ Less

Submitted 12 March, 2021; v1 submitted 9 March, 2021; originally announced March 2021.

Comments: - Results unchanged - Minor rephrasing in Introduction and Methods - Added units on p. 20/21 - Added TL H2CO/CCSD(T)-F12 MAE in Table 1 - Added Acknowledgments and Data availability statement

Optimized Graphene Electrodes for contacting Graphene Nanoribbons

Authors: Oliver Braun, Jan Overbeck, Maria El Abbassi, Silvan Käser, Roman Furrer, Antonis Olziersky, Alexander Flasby, Gabriela Borin Barin, Rimah Darawish, Klaus Müllen, Pascal Ruffieux, Roman Fasel, Ivan Shorubalko, Mickael L. Perrin, Michel Calame

Abstract: Atomically precise graphene nanoribbons are a promising emerging class of designer quantum materials with electronic properties that are tunable by chemical design. However, many challenges remain in the device integration of these materials, especially regarding contacting strategies. We report on the device integration of uniaxially aligned and non-aligned 9-atom wide armchair graphene nanoribbo… ▽ More Atomically precise graphene nanoribbons are a promising emerging class of designer quantum materials with electronic properties that are tunable by chemical design. However, many challenges remain in the device integration of these materials, especially regarding contacting strategies. We report on the device integration of uniaxially aligned and non-aligned 9-atom wide armchair graphene nanoribbons (9-AGNRs) in a field-effect transistor geometry using electron beam lithography-defined graphene electrodes. This approach yields controlled electrode geometries and enables higher fabrication throughput compared to previous approaches using an electrical breakdown technique. Thermal annealing is found to be a crucial step for successful device operation resulting in electronic transport characteristics showing a strong gate dependence. Raman spectroscopy confirms the integrity of the graphene electrodes after patterning and of the GNRs after device integration. Our results demonstrate the importance of the GNR-graphene electrode interface and pave the way for GNR device integration with structurally well-defined electrodes. △ Less

Submitted 25 February, 2021; originally announced February 2021.

Commissioning of low particle flux for proton beams at MedAustron

Authors: Felix Ulrich-Pur, Laurids Adler, Thomas Bergauer, Alexander Burker, Andrea De Franco, Greta Guidoboni, Albert Hirtl, Christian Irmler, Stefanie Kaser, Sebastian Nowak, Florian Pitters, Mauro Pivi, Dale Prokopovich, Claus Schmitzer, Alexander Wastl

Abstract: MedAustron is a synchrotron-based particle therapy centre located in Wiener Neustadt, Austria. It features three irradiation rooms for particle therapy, where proton beams with energies up to 252.7 MeV and carbon ions of up to 402.8 MeV/u are available for cancer treatment. In addition to the treatment rooms, MedAustron features a unique beamline exclusively for non-clinical research (NCR). This r… ▽ More MedAustron is a synchrotron-based particle therapy centre located in Wiener Neustadt, Austria. It features three irradiation rooms for particle therapy, where proton beams with energies up to 252.7 MeV and carbon ions of up to 402.8 MeV/u are available for cancer treatment. In addition to the treatment rooms, MedAustron features a unique beamline exclusively for non-clinical research (NCR). This research beamline is also commissioned for proton energies up to 800 MeV, while available carbon ion energies correspond to the ones available in the clinical treatment rooms. Based on the requirements for particle therapy, all irradiation rooms offer particle rates of up to 10^9 particles/s for protons and 10^7 particles/s for carbon ions. However, for research purposes, lower particle fluxes are required and were therefore commissioned for the NCR beamline. Three particle flux settings with particle rates ranging from ~2.4x10^3 particles/s to ~5.2x10^6 particles/s were established for seven proton energies below 252.7 MeV. In addition to the particle rate, the spot sizes and beam energies were measured for these settings. Furthermore, three low flux settings for 800 MeV protons with particle rates ranging from ~2x10^3 particles/s to ~1.3x10^6 particles/s were commissioned. Since the commissioned low flux settings are in a regime well below the limits of the available standard beam diagnostics, setting up the beam under these new operational conditions entirely relied on the use of external detectors. Furthermore, a beam position measurement based alignment without using the standard beam profile monitors was performed for 800 MeV protons. △ Less

Submitted 27 May, 2021; v1 submitted 11 February, 2021; originally announced February 2021.

Comments: Preprint submitted to NIM A (first revision)

Single particle tracking uncertainties in ion imaging

Authors: A. Burker, T. Bergauer, A. Hirtl, C. Irmler, S. Kaser, B. Knäusl, F. Pitters, F. Ulrich-Pur

Abstract: An extensive comparison of the path uncertainty in single particle tracking systems for ion imaging was carried out based on Monte Carlo simulations. The spatial resolution as function of system parameters such as geometry, detector properties and the energy of proton and helium beams was investigated to serve as a guideline for hardware developments. Primary particle paths were sampled within a… ▽ More An extensive comparison of the path uncertainty in single particle tracking systems for ion imaging was carried out based on Monte Carlo simulations. The spatial resolution as function of system parameters such as geometry, detector properties and the energy of proton and helium beams was investigated to serve as a guideline for hardware developments. Primary particle paths were sampled within a water volume and compared to the most likely path estimate obtained from detector measurements, yielding a depth-dependent uncertainty envelope. The maximum uncertainty along this curve was converted to a conservative estimate of the minimal radiographic pixel spacing for a single set of parameter values. Simulations with various parameter settings were analysed to obtain an overview of the reachable pixel spacing as function of system parameters. The results were used to determine intervals of detector material budget and position resolution that yield a pixel spacing small enough for clinical dose calculation. To ensure a pixel spacing below 2 mm, the material budget of a detector should remain below 0.25 % for a position resolution of 200 $\mathrm{μm}$ or below 0.75 % for a resolution of 10 $\mathrm{μm}$. Using protons, a sub-millimetre pixel size could not be achieved for a phantom size of 300 mm or at a large clearance. With helium ions, a sub-millimetre pixel spacing could be achieved even for a large phantom size and clearance, provided the position resolution was less than 100 $\mathrm{μm}$ and material budget was below 0.75 %. △ Less

Submitted 22 February, 2021; v1 submitted 19 August, 2020; originally announced August 2020.

Comments: 19 pages, 6 figures, submitted to Phys. Med. Biol

ML Models of Vibrating H$_2$CO: Comparing Reproducing Kernels, FCHL and PhysNet

Authors: Silvan Käser, Debasish Koner, Anders S. Christensen, O. Anatole von Lilienfeld, Markus Meuwly

Abstract: Machine Learning (ML) has become a promising tool for improving the quality of atomistic simulations. Using formaldehyde as a benchmark system for intramolecular interactions, a comparative assessment of ML models based on state-of-the-art variants of deep neural networks (NN), reproducing kernel Hilbert space (RKHS+F), and kernel ridge regression (KRR) is presented. Learning curves for energies a… ▽ More Machine Learning (ML) has become a promising tool for improving the quality of atomistic simulations. Using formaldehyde as a benchmark system for intramolecular interactions, a comparative assessment of ML models based on state-of-the-art variants of deep neural networks (NN), reproducing kernel Hilbert space (RKHS+F), and kernel ridge regression (KRR) is presented. Learning curves for energies and atomic forces indicate rapid convergence towards excellent predictions for B3LYP, MP2, and CCSD(T)-F12 reference results for modestly sized (in the hundreds) training sets. Typically, learning curve off-sets decay as one goes from NN (PhysNet) to RKHS+F to KRR (FCHL). Conversely, the predictive power for extrapolation of energies towards new geometries increases in the same order with RKHS+F and FCHL performing almost equally. For harmonic vibrational frequencies, the picture is less clear, with PhysNet and FCHL yielding respectively flat learning at $\sim$ 1 and $\sim$ 0.2 cm$^{-1}$ no matter which reference method, while RKHS+F models level off for B3LYP, and exhibit continued improvements for MP2 and CCSD(T)-F12. Finite-temperature molecular dynamics (MD) simulations with the same initial conditions yield indistinguishable infrared spectra with good performance compared with experiment except for the high-frequency modes involving hydrogen stretch motion which is a known limitation of MD for vibrational spectroscopy. For sufficiently large training set sizes all three models can detect insufficient convergence (``noise'') of the reference electronic structure calculations in that the learning curves level off. Transfer learning (TL) from B3LYP to CCSD(T)-F12 with PhysNet indicates that additional improvements in data efficiency can be achieved. △ Less

Submitted 30 June, 2020; originally announced June 2020.

Journal ref: J. Phys. Chem. A 2020, 124, 42, 8853-8865

Machine Learning for Observables: Reactant to Product State Distributions for Atom-Diatom Collisions

Authors: Julian Arnold, Debasish Koner, Silvan Käser, Narendra Singh, Raymond J. Bemish, Markus Meuwly

Abstract: Machine learning-based models to predict product state distributions from a distribution of reactant conditions for atom-diatom collisions are presented and quantitatively tested. The models are based on function-, kernel- and grid-based representations of the reactant and product state distributions. While all three methods predict final state distributions from explicit quasi-classical trajector… ▽ More Machine learning-based models to predict product state distributions from a distribution of reactant conditions for atom-diatom collisions are presented and quantitatively tested. The models are based on function-, kernel- and grid-based representations of the reactant and product state distributions. While all three methods predict final state distributions from explicit quasi-classical trajectory simulations with R$^2$ > 0.998, the grid-based approach performs best. Although a function-based approach is found to be more than two times better in computational performance, the kernel- and grid-based approaches are preferred in terms of prediction accuracy, practicability and generality. The function-based approach also suffers from lacking a general set of model functions. Applications of the grid-based approach to nonequilibrium, multi-temperature initial state distributions are presented, a situation common to energy distributions in hypersonic flows. The role of such models in Direct Simulation Monte Carlo and computational fluid dynamics simulations is also discussed. △ Less

Submitted 29 May, 2020; originally announced May 2020.

Journal ref: J. Phys. Chem. A 2020, 124, 35, 7177-7190

Isomerization and Decomposition Reactions of Acetaldehyde Relevant to Atmospheric Processes from Dynamics Simulations on Neural Network-Based Potential Energy Surfaces

Authors: Silvan Käser, Oliver T. Unke, Markus Meuwly

Abstract: Acetaldehyde (AA) isomerization (to vinylalcohol, VA) and decomposition (into either CO+CH$_4$ and H$_2$+H$_2$CCO) is studied using a fully dimensional, reactive potential energy surface represented as a neural network (NN). The NN, trained on 432'399 reference structures from MP2/aug-cc-pVTZ calculations has a MAE of 0.0453 kcal/mol and an RMSE of 1.186 kcal/mol for a test set of 27'399 structure… ▽ More Acetaldehyde (AA) isomerization (to vinylalcohol, VA) and decomposition (into either CO+CH$_4$ and H$_2$+H$_2$CCO) is studied using a fully dimensional, reactive potential energy surface represented as a neural network (NN). The NN, trained on 432'399 reference structures from MP2/aug-cc-pVTZ calculations has a MAE of 0.0453 kcal/mol and an RMSE of 1.186 kcal/mol for a test set of 27'399 structures. For the isomerization process AA $\rightarrow$ VA the minimum dynamical path implies that the C-H vibration, and the C-C-H (with H being the transferring H-atom) and the C-C-O angles are involved to surmount the 68.2 kcal/mol barrier. Using an excess energy of 93.6 kcal/mol - the energy available in the solar spectrum and sufficient to excite to the first electronically excited state - to initialize the molecular dynamics, no isomerization to VA is observed on the 500 ns time scale. Only with excess energies of $\sim$ 127.6 kcal/mol (including the zero point energy of the AA molecule), isomerization occurs on the nanosecond time scale. Given that collisional de-excitation at atmospheric conditions in the stratosphere occurs on the 100 ns time scale, it is concluded that formation of VA following photoexcitation of AA from actinic photons is unlikely. This also limits the relevance of this reaction pathway to be a source for formic acid. △ Less

Submitted 18 March, 2020; originally announced March 2020.

Imaging with protons at MedAustron

Authors: F. Ulrich-Pur, T. Bergauer, A. Burker, S. Hatamikia, A. Hirtl, C. Irmler, S. Kaser, P. Paulitsch, F. Pitters, V. Teufelhart

Abstract: Ion beam therapy has become a frequently applied form of cancer therapy over the last years. The advantage of ion beam therapy over conventional radiotherapy using photons is the strongly localized dose deposition, leading to a reduction of dose applied to surrounding healthy tissue. Currently, treatment planning for proton therapy is based on X-ray computed tomography, which entails certain sourc… ▽ More Ion beam therapy has become a frequently applied form of cancer therapy over the last years. The advantage of ion beam therapy over conventional radiotherapy using photons is the strongly localized dose deposition, leading to a reduction of dose applied to surrounding healthy tissue. Currently, treatment planning for proton therapy is based on X-ray computed tomography, which entails certain sources of inaccuracy in alculation of the stop** power (SP). A more precise method to acquire the SP is to directly use high energy protons (or other ions such as carbon) and perform proton computed tomography (pCT). With this method, the ions are tracked prior to entering and after leaving the patient and finally their residual energy is measured at the very end. Therefore, an ion imaging demonstrator, comprising a tracking telescope made from double-sided silicon strip detectors and a range telescope as a residual energy detector, was set up. First measurements with this setup were performed at beam tests at MedAustron, a center for ion therapy and research in \mbox{Wiener Neustadt}, \mbox{Austria}. The facility provides three rooms for cancer treatment with proton beams as well as one which is dedicated to non-clinical research. This contribution describes the principle of ion imaging with proton beams in general as well as the design of the experimental setup. Moreover, first results from simulations and recent beam tests as well as ideas for future developments will be presented. △ Less

Submitted 15 July, 2020; v1 submitted 28 February, 2020; originally announced March 2020.

Comments: 7 pages, 11 figures, proceedings of the "HSTD12: 12th International "Hiroshima" Symposium on the Development and Application of Semiconductor Tracking Detectors (HSTD12)", 14-18 Dec 2019, Hiroshima (Japan), to be submitted to Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment

Reactive Dynamics and Spectroscopy of Hydrogen Transfer from Neural Network-Based Reactive Potential Energy Surfaces

Authors: Silvan Käser, Oliver T. Unke, Markus Meuwly

Abstract: The in silico exploration of chemical, physical and biological systems requires accurate and efficient energy functions to follow their nuclear dynamics at a molecular and atomistic level. Recently, machine learning tools gained a lot of attention in the field of molecular sciences and simulations and are increasingly used to investigate the dynamics of such systems. Among the various approaches,… ▽ More The in silico exploration of chemical, physical and biological systems requires accurate and efficient energy functions to follow their nuclear dynamics at a molecular and atomistic level. Recently, machine learning tools gained a lot of attention in the field of molecular sciences and simulations and are increasingly used to investigate the dynamics of such systems. Among the various approaches, artificial neural networks (NNs) are one promising tool to learn a representation of potential energy surfaces. This is done by formulating the problem as a map** from a set of atomic positions $\mathbf{x}$ and nuclear charges $Z_i$ to a potential energy $V(\mathbf{x})$. Here, a fully-dimensional, reactive neural network representation for malonaldehyde (MA), acetoacetaldehyde (AAA) and acetylacetone (AcAc) is learned. It is used to run finite-temperature molecular dynamics simulations, and to determine the infrared spectra and the hydrogen transfer rates for the three molecules. The finite-temperature infrared spectrum for MA based on the NN learned on MP2 reference data provides a realistic representation of the low-frequency modes and the H-transfer band whereas the CH vibrations are somewhat too high in frequency. For AAA it is demonstrated that the IR spectroscopy is sensitive to the position of the transferring hydrogen at either the OCH- or OCCH$_3$ end of the molecule. For the hydrogen transfer rates it is demonstrated that the O-O vibration is a gating mode and largely determines the rate at which the hydrogen is transferred between the donor and acceptor. Finally, possibilities to further improve such NN-based potential energy surfaces are explored. They include the transferability of an NN-learned energy function across chemical species (here methylation) and transfer learning from a lower level of reference data (MP2) to a higher level of theory (pair natural orbital-LCCSD(T)). △ Less

Submitted 21 November, 2019; originally announced November 2019.

High-Dimensional Potential Energy Surfaces for Molecular Simulations

Authors: Oliver T. Unke, Debasish Koner, Sarbani Patra, Silvan Käser, Markus Meuwly

Abstract: An overview of computational methods to describe high-dimensional potential energy surfaces suitable for atomistic simulations is given. Particular emphasis is put on accuracy, computability, transferability and extensibility of the methods discussed. They include empirical force fields, representations based on reproducing kernels, using permutationally invariant polynomials, and neural network-l… ▽ More An overview of computational methods to describe high-dimensional potential energy surfaces suitable for atomistic simulations is given. Particular emphasis is put on accuracy, computability, transferability and extensibility of the methods discussed. They include empirical force fields, representations based on reproducing kernels, using permutationally invariant polynomials, and neural network-learned representations and combinations thereof. Future directions and potential improvements are discussed primarily from a practical, application-oriented perspective. △ Less

Submitted 7 October, 2019; originally announced October 2019.

Journal ref: Mach. Learn.: Sci. Technol. 1 (2020) 013001