-
Predictability Analysis and Prediction of Discrete Weather and Financial Time-Series Data with a Hamiltonian-Based Filter-Projection Approach
Authors:
Henrik Kiefer,
Denis Furtel,
Cihan Ayaz,
Anton Klimek,
Jan O. Daldrop,
Roland R. Netz
Abstract:
The generalized Langevin equation (GLE), derived by projection from a general many-body Hamiltonian, exactly describes the dynamics of an arbitrary coarse-grained variable in a complex environment. However, analysis and prediction of real-world data with the GLE is hampered by slow transient or seasonal data components and time-discretization effects. Machine-learning (ML) techniques work but are…
▽ More
The generalized Langevin equation (GLE), derived by projection from a general many-body Hamiltonian, exactly describes the dynamics of an arbitrary coarse-grained variable in a complex environment. However, analysis and prediction of real-world data with the GLE is hampered by slow transient or seasonal data components and time-discretization effects. Machine-learning (ML) techniques work but are computer-resource demanding and difficult to interpret. We show that by convolution filtering, time-series data decompose into fast, transient and seasonal components that each obey Hamiltonian dynamics and, thus, can be separately analyzed by projection techniques. We introduce methods to extract all GLE parameters from highly discretized time-series data and to forecast future data including the environmental stochasticity. For daily-resolved weather data, our analysis reveals non-Markovian memory that decays over a few days. Our prediction accuracy is comparable to commercial (weather.com) and ML long short-term memory (LSTM) methods at a reduced computational cost by a factor of $10^2-10^3$ compared to LSTM. For financial data, memory is very short-ranged and the dynamics effectively is Markovian, in agreement with the efficient-market hypothesis; consequently, models simpler than the GLE are sufficient. Our GLE framework is an efficient and interpretable method for the analysis and prediction of complex time-series data.
△ Less
Submitted 24 September, 2024; v1 submitted 23 September, 2024;
originally announced September 2024.
-
Effect of Frequency-Dependent Viscosity on Molecular Friction in Liquids
Authors:
Henrik Kiefer,
Domenico Vitali,
Benjamin A. Dalton,
Laura Scalfi,
Roland R. Netz
Abstract:
The relation between the frequency-dependent friction of a molecule in a liquid and the hydrodynamic properties of the liquid is fundamental for molecular dynamics. We investigate this connection for a water molecule moving in liquid water using all-atomistic molecular dynamics simulations and linear hydrodynamic theory. We analytically calculate the frequency-dependent friction of a sphere with f…
▽ More
The relation between the frequency-dependent friction of a molecule in a liquid and the hydrodynamic properties of the liquid is fundamental for molecular dynamics. We investigate this connection for a water molecule moving in liquid water using all-atomistic molecular dynamics simulations and linear hydrodynamic theory. We analytically calculate the frequency-dependent friction of a sphere with finite surface slip moving in a viscoelastic compressible fluid by solving the linear transient Stokes equation, including frequency-dependent shear and volume viscosities, both determined from MD simulations of bulk liquid water. We also determine the frequency-dependent friction of a single water molecule moving in liquid water, as defined by the generalized Langevin equation from MD simulation trajectories. By fitting the effective sphere radius and the slip length, the frequency-dependent friction and velocity autocorrelation function from the transient Stokes equation and simulations quantitatively agree. This shows that the transient Stokes equation accurately describes the important features of the frequency-dependent friction of a single water molecule in liquid water and thus applies down to molecular length and time scales, provided accurate frequency-dependent viscosities are used. The frequency dependence of the shear viscosity of liquid water requires careful consideration of hydrodynamic finite-size effects to observe the asymptotic hydrodynamic power-law tail. In contrast, for a methane molecule moving in water, the frequency-dependent friction cannot be predicted based on a homogeneous model, which suggests, supported by the extraction of a frequency-dependent surface-slip profile, that a methane molecule is surrounded by a finite-thickness hydration layer with viscoelastic properties that are significantly different from bulk water.
△ Less
Submitted 22 August, 2024;
originally announced August 2024.
-
Nanoscopic Interfacial Hydrogel Viscoelasticity Revealed from Comparison of Macroscopic and Microscopic Rheology
Authors:
Robert F. Schmidt,
Henrik Kiefer,
Robert Dalgliesh,
Michael Gradzielski,
Roland R. Netz
Abstract:
Deviations between macrorheological and particle-based microrheological measurements are often considered a nuisance and neglected. We study aqueous poly(ethylene oxide) (PEO) hydrogels for varying PEO concentrations and chain lengths that contain microscopic tracer particles and show that these deviations in fact reveal the nanoscopic viscoelastic properties of the particle-hydrogel interface. Ba…
▽ More
Deviations between macrorheological and particle-based microrheological measurements are often considered a nuisance and neglected. We study aqueous poly(ethylene oxide) (PEO) hydrogels for varying PEO concentrations and chain lengths that contain microscopic tracer particles and show that these deviations in fact reveal the nanoscopic viscoelastic properties of the particle-hydrogel interface. Based on the transient Stokes equation, we first demonstrate that the deviations are not due to finite particle radius, compressibility or surface-slip effects. Small-angle neutron scattering rules out hydrogel heterogeneities. Instead, we show that a generalized Stokes-Einstein relation, accounting for a nanoscopic interfacial shell around tracers with viscoelastic properties that significantly deviate from bulk, consistently explains our macrorheological and microrheological measurements. The extracted shell diameter is comparable with the PEO end-to-end distance, indicating the importance of dangling chain ends. Our methodology reveals the nanoscopic interfacial rheology of hydrogels and is generally applicable to different kinds of viscoelastic fluids and particles.
△ Less
Submitted 16 February, 2024; v1 submitted 12 December, 2023;
originally announced December 2023.
-
Conformational isomerization dynamics in solvent violates both the Stokes-Einstein relation and Kramers' theory
Authors:
Benjamin A. Dalton,
Henrik Kiefer,
Roland R. Netz
Abstract:
Molecular isomerization kinetics in liquid solvents are determined by a complex interplay between the friction acting on a rotating dihedral due to interactions with the solvent, internal dissipation effects (also known as internal friction), the viscosity of the solvent, and the free energy profile over which a dihedral rotates. Currently, it is not understood how these quantities are related at…
▽ More
Molecular isomerization kinetics in liquid solvents are determined by a complex interplay between the friction acting on a rotating dihedral due to interactions with the solvent, internal dissipation effects (also known as internal friction), the viscosity of the solvent, and the free energy profile over which a dihedral rotates. Currently, it is not understood how these quantities are related at the molecular scale. Here, we combine molecular dynamics simulations of isomerizing n-alkane chains and dipeptide molecules in mixed water-glycerol solvents with memory-kernel extraction techniques to directly evaluate the frequency-dependent friction acting on a rotating dihedral. We extract the friction and isomerization times over a range of glycerol concentrations and accurately evaluate the relationships between solvent viscosity, isomerization kinetics, and dihedral friction. We show that the total friction acting on a rotating dihedral does not scale linearly with solvent viscosity, thus violating the Stokes-Einstein relation. Additionally, we demonstrate that the kinetics of isomerization are significantly faster compared to the Kramers prediction in the overdamped limit. We suggest that isomerization kinetics are determined by the multi-time-scale friction coupling between a rotating dihedral and its solvent environment, which results in non-Markovian kinetic speed-up effects.
△ Less
Submitted 14 September, 2023;
originally announced September 2023.
-
Frequency-dependent hydrodynamic finite size correction in molecular simulations reveals the long-time hydrodynamic tail
Authors:
Laura Scalfi,
Domenico Vitali,
Henrik Kiefer,
Roland R. Netz
Abstract:
Finite-size effects are challenging in molecular dynamics simulations because they have significant effects on computed static and dynamic properties, in particular diffusion constants, friction coefficients and time- or frequency-dependent response functions. We investigate the influence of periodic boundary conditions on the velocity autocorrelation function and the frequency-dependent friction…
▽ More
Finite-size effects are challenging in molecular dynamics simulations because they have significant effects on computed static and dynamic properties, in particular diffusion constants, friction coefficients and time- or frequency-dependent response functions. We investigate the influence of periodic boundary conditions on the velocity autocorrelation function and the frequency-dependent friction of a particle in a fluid and show that the long-time behavior (starting at the picosecond timescale) is significantly affected. We develop an analytical correction allowing to subtract the periodic boundary condition effects. By this we unmask the power-law long-time tails of the memory kernel and the velocity autocorrelation function in liquid water and a Lennard-Jones fluid from rather small simulation boxes.
△ Less
Submitted 17 February, 2023;
originally announced February 2023.
-
Photo-stability of super-hydrogenated PAHs determined by action spectroscopy experiments
Authors:
M. Wolf,
H. V. Kiefer,
J. Langeland,
L. H. Andersen,
H. Zettergren,
H. T. Schmidt,
H. Cederquist,
M. H. Stockett
Abstract:
We have investigated the photo-stability of pristine and super-hydrogenated pyrene cations C$_{16}$H$_{10+m}^+, m = 0,6, \mathrm{\ or\ } 16$) by means of gas-phase action spectroscopy. Optical absorption spectra and photo-induced dissociation mass spectra are presented. By measuring the yield of mass-selected photo-fragment ions as a function of laser pulse intensity, the number of photons (and he…
▽ More
We have investigated the photo-stability of pristine and super-hydrogenated pyrene cations C$_{16}$H$_{10+m}^+, m = 0,6, \mathrm{\ or\ } 16$) by means of gas-phase action spectroscopy. Optical absorption spectra and photo-induced dissociation mass spectra are presented. By measuring the yield of mass-selected photo-fragment ions as a function of laser pulse intensity, the number of photons (and hence the energy) needed for fragmentation of the carbon backbone was determined. Backbone fragmentation of pristine pyrene ions (C$_{16}$H$_{10}^+$) requires absorption of three photons of energy just below 3 eV, whereas super-hydrogenated hexahydropyrene (C$_{16}$H$_{16}^+$) must absorb two such photons and fully hydrogenated hexadecahydropyrene (C$_{16}$H$_{26}^+$) only a single photon. These results are consistent with previously reported dissociation energies for these ions. Our experiments clearly demonstrate that the increased heat capacity from the additional hydrogen atoms does not compensate for the weakening of the carbon backbone when pyrene is hydrogenated. In photodissociation regions, super-hydrogenated Polycyclic Aromatic Hydrocarbons (PAHs) have been proposed to serve as catalysts for H$_2$-formation. Our results indicate that carbon backbone fragmentation may be a serious competitor to H$_2$-formation at least for small hydrogenated PAHs like pyrene.
△ Less
Submitted 14 September, 2016;
originally announced September 2016.
