Searcharxiv⌕ Search

arXiv subjects

Nikolai V. Brilliantov

Publications and source records attributed to Nikolai V. Brilliantov.

At least 19 recordsLinked to original sources

Umbrella Reinforcement Learning -- computationally efficient tool for hard non-linear problems

We report a novel, computationally efficient approach for solving hard nonlinear problems of reinforcement learning (RL). Here we combine umbrella sampling, from computational physics/chemistry, with optimal control methods. The approach is realized on the basis of neural networks, with the use of policy gradient. It outperforms, by computational efficiency and implementation universality, all available state-of-the-art algorithms, in application to hard RL problems with sparse reward, state traps and lack of terminal states. The proposed approach uses an ensemble of simultaneously acting agents, with a modified reward which includes the ensemble entropy, yielding an optimal exploration-exploitation balance.

cs.LG↗

Supercluster states and phase transitions in aggregation-fragmentation processes

We study the evolution of aggregates triggered by collisions with monomers that either lead to the attachment of monomers or the break-up of aggregates into constituting monomers. Depending on parameters quantifying addition and break-up rates, the system falls into a jammed or a steady state. Supercluster states (SCSs) are very peculiar non-extensive jammed states that also arise in some models. Fluctuations underlie the formation of the SCSs. Conventional tools, such as the van Kampen expansion, apply to small fluctuations. We go beyond the van Kampen expansion and determine a set of critical exponents quantifying SCSs. We observe continuous and discontinuous phase transitions between the states. Our theoretical predictions are in good agreement with numerical results.

cond-mat.stat-mech↗

Aggregation kinetics at sedimentation: the impact of particles diffusion

We investigate the aggregation kinetics of sedimenting particles theoretically and numerically, using the advection-diffusion equation. Agglomeration, caused by both transport mechanisms (diffusion and advection), is important for small particles, like primary ash or soot particles in atmosphere, and large particles of equal or close size, where the advection mechanism is weak. For small Peclet numbers, which quantify the relative importance of diffusion and advection, we obtain the aggregation rates, as an expansion in Peclet numbers. For large Peclet numbers we use purely ballistic aggregation rates. Combining these results we obtain the rational approximant for the whole range of Peclet numbers. We also compute the aggregation rates by numerically solving the advection-diffusion equation. The results of the numerical simulations are in excellent agreement with the analytical theory for the studied Peclet numbers, varying by four orders of magnitude.

cond-mat.stat-mech↗

Aggregation in non-uniform systems with advection and localized source

We explore analytically and numerically agglomeration driven by advection and localized source. The system is inhomogeneous in one dimension, viz. along the direction of advection. We analyze a simplified model with mass-independent advection velocity, diffusion coefficient, and reaction rates. We also examine a model with mass-dependent coefficients describing aggregation with sedimentation. For the simplified model, we obtain an exact solution for the stationary spatially dependent agglomerate densities. In the model describing aggregation with sedimentation, we report a new conservation law and develop a scaling theory for the densities. For numerical efficiency we exploit the low-rank approximation technique; this dramatically increases the computational speed and allows simulations of large systems. The numerical results are in excellent agreement with the predictions of our theory.

cond-mat.stat-mech↗

Non-extensive super-cluster states in aggregation with fragmentation

Systems evolving through aggregation and fragmentation may possess an intriguing super-cluster state (SCS). Clusters constituting this state are mostly large, so the SCS resembles a gelling state. The formation of the SCS is controlled by fluctuations and in this aspect, it is similar to a critical state. The SCS is non-extensive as the number of clusters varies sub-linearly with the system size. In the parameter space, the SCS separates equilibrium and jamming (extensive) states. The conventional methods such as the van Kampen expansion fail to describe the SCS. To characterize the SCS, we propose a scaling approach with a set of critical exponents. Our theoretical findings are in good agreement with numerical results.

cond-mat.stat-mech↗

Collision of nanoparticles of covalently bound atoms. Impact of stress-dependent adhesion

The impact of nanoparticles (NPs) comprised of atoms with covalent bonding is investigated numerically and theoretically. We use recent models of covalent bonding of carbon atoms and elaborate a numerical model of amorphous carbon (a-C) NPs, which may be applied for modelling soot particles. We compute the elastic moduli of the a-C material which agree well with the available data. We reveal an interesting novel phenomenon - stress dependent adhesion, which refers to stress-enhanced formation of covalent bonds between contacting surfaces. We observe that the effective adhesion coefficient linearly depends on the maximal stress between the surfaces and explain this dependence. We compute the normal restitution coefficient for colliding NPs and explore the dependence of the critical velocity, demarcating bouncing and aggregative collisions, on the NP radius. Using the obtained elastic and stress-dependent adhesive coefficients we develop a theory for the critical velocity. The predictions of the theory agree very well with the simulation results.

cond-mat.mes-hall↗

Molecular fields and statistical field theory of fluids. Application to interface phenomena

Using the integral transformation, the field-theoretical Hamiltonian of the statistical field theory of fluids is obtained, along with the microscopic expressions for the coefficients of the Hamiltonian. Applying this approach to the liquid-vapor interface, we derive an explicit analytical expression for the surface tension in terms of temperature, density and parameters of inter-molecular potential. We also demonstrate that a clear physical interpretation may be given to the formal statistical field arising in the integral transformation - it may be associated with the one-body local microscopic potential. The results of the theory, lacking any ad-hoc or fitting parameters are in a good agreement with available simulation data.

cond-mat.stat-mech↗

Size-polydisperse dust in molecular gas: Energy equipartition versus non-equipartition

We investigate numerically and analytically size-polydisperse granular mixtures immersed into a molecular gas. We show that the equipartition of granular temperatures of particles of different sizes is established; however, the granular temperatures significantly differ from the temperature of the molecular gas. This result is surprising since, generally, the energy equipartition is strongly violated in driven granular mixtures. Qualitatively, the obtained results do not depend on the collision model, being valid for a constant restitution coefficient $\varepsilon$, as well as for the $\varepsilon$ for viscoelastic particles. Our findings may be important for astrophysical applications, such as protoplanetary disks, interstellar dust clouds, and comets.

cond-mat.stat-mech↗

Kinetic regimes in aggregating systems with spontaneous and collisional fragmentation

We analyze systems of clusters and interacting upon colliding---a collision between two clusters may lead to merging or fragmentation---and we also investigate the influence of additional spontaneous fragmentation events. We consider both closed systems in which the total mass remains constant and open systems driven by a source of small-mass clusters. In closed systems, the size distribution of aggregates approaches a steady state. For these systems the relaxation time and the steady state distribution are determined mostly by spontaneous fragmentation while collisional fragmentation plays a minor role. For open systems, in contrast, the collisional fragmentation dominates. In this case, the system relaxes to a quasi-stationary state where cluster densities linearly grow with time, while the functional form of the cluster size distribution persists and coincides with the steady state size distribution of a system which has the same aggregation and fragmentation rates and only collisional fragmentation.

cond-mat.stat-mech↗

Mechanism of chain collapse of strongly charged polyelectrolytes

We perform extensive molecular dynamics simulations of a charged polymer in a good solvent in the regime where the chain is collapsed. We analyze the dependence of the gyration radius $R_g $ on the reduced Bjerrum length $\ell_B$ and find two different regimes. In the first one, called as a weak electrostatic regime, $R_g \sim \ell_B^{-1/2}$, which is consistent only with the predictions of the counterion-fluctuation theory. In the second one, called a strong electrostatic regime, we find $R_g \sim \ell_B^{-1/5}$. To explain the novel regime we modify the counterion-fluctuation theory.

cond-mat.soft↗

Collision of viscoelastic bodies: Rigorous derivation of dissipative force

We report a new theory of dissipative forces acting between colliding viscoelastic bodies. The impact velocity is assumed not to be large, to avoid plastic deformations and fragmentation at the impact. The bodies may be of an arbitrary convex shape and of different materials. We develop a mathematically rigorous perturbation scheme to solve the continuum mechanics equation that deals with both displacement and displacement rate fields and accounts for the dissipation in the bulk of the material. The perturbative solution of this equation allows to go beyond the previously used quasi-static approximation and obtain the dissipative force. This force does not suffer from the physical inconsistencies of the latter approximation and depends on particle deformation and deformation rate.

cond-mat.soft↗

Diffusive counter dispersion of mass in bubbly media

We consider a liquid bearing gas bubbles in a porous medium. When gas bubbles are immovably trapped in a porous matrix by surface-tension forces, the dominant mechanism of transfer of gas mass becomes the diffusion of gas molecules through the liquid. Essentially, the gas solution is in local thermodynamic equilibrium with vapor phase all over the system, i.e., the solute concentration equals the solubility. When temperature and/or pressure gradients are applied, diffusion fluxes appear and these fluxes are faithfully determined by the temperature and pressure fields, not by the local solute concentration, which is enslaved by the former. We derive the equations governing such systems, accounting for thermodiffusion and gravitational segregation effects which are shown not to be neglected for geological systems---marine sediments, terrestrial aquifers, etc. The results are applied for the treatment of non-high-pressure systems and real geological systems bearing methane or carbon dioxide, where we find a potential possibility of the formation of gaseous horizons deep below a porous medium surface. The reported effects are of particular importance for natural methane hydrate deposits and the problem of burial of industrial production of carbon dioxide in deep aquifers.

cond-mat.stat-mech↗

Negative normal restitution coefficient for nanocluster collisions

The oblique impacts of nanoclusters is studied by means of Molecular Dynamics and theoretically. In simulations we explore two models -- Lennard-Jones clusters and particles with covalently bonded atoms. In contrast to the case of macroscopic bodies, the standard definition of the normal restitution coefficient yields for this coefficient negative values for oblique collisions of nanoclusters. We explain this effect and propose a proper definition of the restitution coefficient which is always positive. We develop a theory of an oblique impact based on continuum model of particles. A surprisingly good agreement between the macroscopic theory and simulations leads to the conclusion, that macroscopic concepts of elasticity, bulk viscosity and surface tension remain valid for nanoparticles of a few hundreds atoms.

physics.atm-clus↗

Granular Gas Cooling and Relaxation to the Steady State in Regard to the Overpopulated Tail of the Velocity Distribution

We present a universal description of the velocity distribution function of granular gases, $f(v)$, valid for both, small and intermediate velocities where $v$ is close to the thermal velocity and also for large $v$ where the distribution function reveals an exponentially decaying tail. By means of large-scale Monte Carlo simulations and by kinetic theory we show that the deviation from the Maxwell distribution in the high-energy tail leads to small but detectable variation of the cooling coefficient and to extraordinary large relaxation time.

cond-mat.stat-mech↗

Translations and Rotations are correlated in Granular Gases

In a granular gas of rough particles the axis of rotation is shown to be correlated with the translational velocity of the particles. The average relative orientation of angular and linear velocities depends on the parameters which characterise the dissipative nature of the collision. We derive a simple theory for these correlations and validate it with numerical simulations for a wide range of coefficients of normal and tangential restitution. The limit of smooth spheres is shown to be singular: even an arbitrarily small roughness of the particles gives rise to orientational correlations.

cond-mat.stat-mech↗

Breakdown of the Sonine expansion for the velocity distribution of Granular Gases

The velocity distribution of a granular gas is analyzed in terms of the Sonine polynomials expansion. We derive an analytical expression for the third Sonine coefficient a_3. In contrast to frequently used assumptions this coefficient is of the same order of magnitude as the second Sonine coefficient a_2. For small inelasticity the theoretical result is in good agreement with numerical simulations. The next-order Sonine coefficients a_4, a_5 and a_6 are determined numerically. While these coefficients are negligible for small dissipation, their magnitude grows rapidly with increasing inelasticity for 0< epsilon < 0.6. We conclude that this behavior of the Sonine coefficients manifests the break down of the Sonine polynomial expansion caused by the increasing impact of the overpopulated high-energy tail of the distribution function.

cond-mat.stat-mech↗

Collision of Adhesive Viscoelastic Particles

The collision of convex bodies is considered for small impact velocity, when plastic deformation and fragmentation may be disregarded. In this regime the contact is governed by forces according to viscoelastic deformation and by adhesion. The viscoelastic interaction is described by a modified Hertz law, while for the adhesive interactions, the model by Johnson, Kendall and Roberts (JKR) is adopted. We solve the general contact problem of convex viscoelastic bodies in quasi-static approximation, which implies that the impact velocity is much smaller than the speed of sound in the material and that the viscosity relaxation time is much smaller than the duration of a collision. We estimate the threshold impact velocity which discriminates restitutive and sticking collisions. If the impact velocity is not large as compared with the threshold velocity, adhesive interaction becomes important, thus limiting the validity of the pure viscoelastic collision model.

cond-mat.stat-mech↗

Adhesive interactions of viscoelastic spheres

We develop an analytical theory of adhesive interaction of viscoelastic spheres in quasistatic approximation. Deformations and deformation rates are assumed to be small, which allows for the application of the Hertz contact theory, modified to account for viscoelastic forces. The adhesion interactions are described by the Johnson, Kendall, and Roberts theory. Using the quasistatic approximation we derive the total force between the bodies which is not sufficiently described by the superposition of elastic, viscous and adhesive contributions, but instead an additional cross-term appears, which depends on the elastic, viscous and adhesive parameters of the material. Using the derived theory we estimate the contribution of adhesive forces to the normal coefficient of restitution and derive a criterion for the validity of the viscoelastic collision model.

cond-mat.mtrl-sci↗