arXiv subjects

Dilina Perera

Publications and source records attributed to Dilina Perera.

11 recordsLinked to original sources

arXiv2025

Confusion-driven machine learning of structural phases of a flexible, magnetic Stockmayer polymer

We use a semi-supervised, neural-network based machine learning technique, the confusion method, to investigate structural transitions in magnetic polymers, which we model as chains of magnetic colloidal nanoparticles characterized by dipole-dipole and Lennard-Jones interactions. As input for the neural network we use the particle positions and magnetic dipole moments of equilibrium polymer configurations, which we generate via replica-exchange Wang--Landau simulations. We demonstrate that by measuring the classification accuracy of neural networks, we can effectively identify transition points between multiple structural phases without any prior knowledge of their existence or location. We corroborate our findings by investigating relevant, conventional order parameters. Our study furthermore examines previously unexplored low-temperature regions of the phase diagram, where we find new structural transitions between highly ordered helicoidal polymer configurations.

cond-mat.soft↗

arXiv2022

Supervised and Unsupervised Machine Learning of Structural Phases of Polymers Adsorbed to Nanowires

We identify configurational phases and structural transitions in a polymer nanotube composite by means of machine learning. We employ various unsupervised dimensionality reduction methods, conventional neural networks, as well as the confusion method, an unsupervised neural-network-based approach. We find neural networks are able to reliably recognize all configurational phases that have been found previously in experiment and simulation. Furthermore, we locate the boundaries between configurational phases in a way that removes human intuition or bias. This could be done before only by relying on preconceived, ad-hoc order parameters.

cond-mat.soft↗

arXiv2022

Benchmarking quantum annealers using symmetries in embedded subgraphs

We investigate an efficient, generic method for evaluating the performance of quantum annealing devices that does not require the prior knowledge of the true ground state of the benchmark problem. This approach exploits symmetry properties inherent to the ground states of a composite Hamiltonian comprising the benchmark problem Hamiltonian and its symmetric counterpart. Using this method, we compare the performance of two generations of D-Wave machines. Although we do not observe a noticeable difference in the probability of finding solutions with the required symmetry, our results suggest that the current generation of D-Wave machines notably outperforms its predecessor in finding states closer to those with the required symmetry.

quant-ph↗

arXiv2021

Chook -- A comprehensive suite for generating binary optimization problems with planted solutions

We present Chook, an open-source Python-based tool to generate discrete optimization problems of tunable complexity with a priori known solutions. Chook provides a cross-platform unified environment for solution planting using a number of techniques, such as tile planting, Wishart planting, equation planting, and deceptive cluster loop planting. Chook also incorporates planted solutions for higher-order (beyond quadratic) binary optimization problems. The support for various planting schemes and the tunable hardness allows the user to generate problems with a wide range of complexity on different graph topologies ranging from hypercubic lattices to fully-connected graphs.

quant-ph↗

arXiv2020

Computational hardness of spin-glass problems with tile-planted solutions

We investigate the computational hardness of spin-glass instances on a square lattice, generated via a recently introduced tunable and scalable approach for planting solutions. The method relies on partitioning the problem graph into edge-disjoint subgraphs, and planting frustrated, elementary subproblems that share a common local ground state, which guarantees that the ground state of the entire problem is known a priori. Using population annealing Monte Carlo, we compare the typical hardness of problem classes over a large region of the multi-dimensional tuning parameter space. Our results show that the problems have a wide range of tunable hardness. Moreover, we observe multiple transitions in the hardness phase space, which we further corroborate using simulated annealing and simulated quantum annealing. By investigating thermodynamic properties of these planted systems, we demonstrate that the harder samples undergo magnetic ordering transitions which are also ultimately responsible for the observed hardness transitions on changing the sample composition.

cond-mat.dis-nn↗

arXiv2018

From Near to Eternity: Spin-glass planting, tiling puzzles, and constraint satisfaction problems

We present a methodology for generating Ising Hamiltonians of tunable complexity and with a priori known ground states based on a decomposition of the model graph into edge-disjoint subgraphs. The idea is illustrated with a spin-glass model defined on a cubic lattice, where subproblems, whose couplers are restricted to the two values {-1,+1}, are specified on unit cubes and are parametrized by their local degeneracy. The construction is shown to be equivalent to a type of three-dimensional constraint satisfaction problem known as the tiling puzzle. By varying the proportions of subproblem types, the Hamiltonian can span a dramatic range of typical computational complexity, from fairly easy to many orders of magnitude more difficult than prototypical bimodal and Gaussian spin glasses in three space dimensions. We corroborate this behavior via experiments with different algorithms and discuss generalizations and extensions to different types of graphs.

cond-mat.dis-nn↗

arXiv2018

Combined Molecular and Spin Dynamics Simulation of BCC Iron with Vacancy Defects

Utilizing an atomistic computational model which handles both translational and spin degrees of freedom, combined molecular and spin dynamics simulations have been performed to investigate the effect of vacancy defects on spin wave excitations in ferromagnetic iron. Fourier transforms of space and time-displaced correlation functions yield the dynamic structure factor, providing characteristic frequencies and lifetimes of the spin wave modes. Comparison of the system with a 5% vacancy concentration with pure lattice data shows a decrease in frequency as well as a decrease in lifetime for all transverse spin wave excitations observed. Additionally, a rugged spin wave line shape for low-q spin waves indicates the presence of multiple localized excitations near defect sites resulting in reduced excitation lifetimes due to increased magnon-magnon scattering. We observe further evidence of increased magnon-magnon scattering as the peaks in the longitudinal spin wave spectrum become less distinct.

physics.comp-ph↗

arXiv2017

Collective dynamics in atomistic models with coupled translational and spin degrees of freedom

Using an atomistic model that simultaneously treats the dynamics of translational and spin degrees of freedom, we perform combined molecular and spin dynamics simulations to investigate the mutual influence of the phonons and magnons on their respective frequency spectra and lifetimes in ferromagnetic bcc iron. By calculating the Fourier transforms of the space- and time-displaced correlation functions, the characteristic frequencies and the linewidths of the vibrational and magnetic excitation modes were determined. Comparison of the results with that of the stand-alone molecular dynamics and spin dynamics simulations reveals that the dynamic interplay between the phonons and magnons leads to a shift in the respective frequency spectra and a decrease in the lifetimes. Moreover, in the presence of lattice vibrations, additional longitudinal magnetic excitations were observed with the same frequencies as the longitudinal phonons.

cond-mat.mtrl-sci↗

arXiv2016

Reinventing atomistic magnetic simulations with spin-orbit coupling

We propose a powerful extension to combined molecular and spin dynamics that fully captures the coupling between the atomic and spin subsystems via spin-orbit interactions. Its foundation is the inclusion of the local magnetic anisotropies that arise as a consequence of the lattice symmetry breaking due to phonons or defects. We demonstrate that our extension enables the exchange of angular momentum between the atomic and spin subsystems, which is critical to the challenges arising in the study of fluctuations and non-equilibrium processes in complex, natural, and engineered magnetic materials.

cond-mat.mtrl-sci↗

arXiv2016

Magnetic phase transition in coupled spin-lattice systems: A replica-exchange Wang-Landau study

Coupled, dynamical spin-lattice models provide a unique test ground for simulations investigating the finite-temperature magnetic properties of materials under the direct influence of the lattice vibrations. These models are constructed by combining a coordinate-dependent interatomic potential with a Heisenberg-like spin Hamiltonian, facilitating the treatment of both the atomic coordinates and spins as explicit phase variables. Using a model parameterized for bcc iron, we study the magnetic phase transition in these complex systems via the recently introduced, massively parallel replica-exchange Wang-Landau Monte Carlo method. Comparison with the results obtained from rigid lattice (spin only) simulations show that the transition temperature as well as the amplitude of the peak in the specific heat curve is marginally affected by the lattice vibrations. Moreover, the results were found to be sensitive to the particular choice of the interatomic potential.

cond-mat.stat-mech↗

arXiv2014

Replica-exchange Wang-Landau sampling: Pushing the limits of Monte Carlo simulations in materials sciences

We describe the study of thermodynamics of materials using replica-exchange Wang-Landau (REWL) sampling, a generic framework for massively parallel implementations of the Wang-Landau Monte Carlo method. To evaluate the performance and scalability of the method, we investigate the magnetic phase transition in body-centered cubic (bcc) iron using the classical Heisenberg model parametrized with first principles calculations. We demonstrate that our framework leads to a significant speedup without compromising the accuracy and precision and facilitates the study of much larger systems than is possible with its serial counterpart.

cond-mat.mtrl-sci↗