193 research outputs found
Ranking the information content of distance measures
Full text linkReal-world data typically contain a large number of features that are often
heterogeneous in nature, relevance, and also units of measure. When assessing
the similarity between data points, one can build various distance measures
using subsets of these features. Using the fewest features but still retaining
sufficient information about the system is crucial in many statistical learning
approaches, particularly when data are sparse. We introduce a statistical test
that can assess the relative information retained when using two different
distance measures, and determine if they are equivalent, independent, or if one
is more informative than the other. This in turn allows finding the most
informative distance measure out of a pool of candidates. The approach is
applied to find the most relevant policy variables for controlling the Covid-19
epidemic and to find compact yet informative representations of atomic
structures, but its potential applications are wide ranging in many branches of
science
Ab initio thermodynamics of liquid and solid water
No full textThermodynamic properties of liquid water as well as hexagonal (Ih) and cubic (Ic) ice are predicted based on density functional theory at the hybrid-functional level, rigorously taking into account quantum nuclear motion, anharmonic fluctuations, and proton disorder. This is made possible by combining advanced free-energy methods and state-of-the-art machine-learning techniques. The ab initio description leads to structural properties in excellent agreement with experiments and reliable estimates of the melting points of light and heavy water. We observe that nuclear-quantum effects contribute a crucial 0.2 meV/H2O to the stability of ice Ih, making it more stable than ice Ic. Our computational approach is general and transferable, providing a comprehensive framework for quantitative predictions of ab initio thermodynamic properties using machine-learning potentials as an intermediate step.COSM
Methods and Material Developed beyond Conventional Nanoimprint
Full text linkNanoimprint lithography (NIL) has been regarded as one of the next-generation lithography techniques due to its ability to fabricate nanoscale structures with low cost and high throughput. Although both thermal and UV nanoimprint have demonstrated sub-10 nm resolution, the adoption of NIL by industry has been very limited. The main reasons are that the density of pattern defects and low throughput cannot satisfy the stringent requirement of commercial lithographic technique. In this study, methods and material have been developed to overcome the limitations beyond conventional nanoimprint by utilizing three main factors: mold, the interface between mold and resist, and resist.
In the study, we first developed a new synergistic thermal and UV nanoimprint lithography (STUV-NIL). A transparent mold is integrated with a transparent metal oxide heater, which enables resists to be cured by thermal energy and UV light spontaneously. This new STUV-NIL combines thermal and UV techniques into one module and helps throughput improvement and reducing mold-resist adhesion and hence defect generation.
In the second part of this study, the thermal behavior of a polycarbonate resist was investigated by characterization of polycarbonate gratings reflow after thermal annealing. The observation of exceptional thermal stability of entangled polycarbonate polymer opens up new routes of step-and-repeat thermal nanoimprint and high resolution patterning.
The adhesion characteristic between polymers and the mold is a critical factor in the demolding process. In the third part of our study, polycarbonate residual layer has been applied as an anti sticking treatment on nanoimprint molds, replacing the self-assembled monolayer currently used. It satisfies the requirements of not only low surface energy but also low reactivity for durability.
Polymerizations in UV NIL are generally accompanied by shrinkage in volume, which causes serious problems such as residual stress, demolding problems and defects. Epoxy-based UV resists have a volume shrinkage in the range of 3% to 10%. In the fourth part of our study, spiro-orthocarbonate, which undergoes volume expansion upon cationic ring-opening polymerization, has been mixed with an epoxy monomer to adjust the volume shrinkage of the cured resist and achieve zero volume change after curing
Computing chemical potentials of solutions from structure factors
Full text linkThe chemical potential of a component in a solution is defined as the free
energy change as the amount of the component changes. Computing this
fundamental thermodynamic property from atomistic simulations is notoriously
difficult, because of the convergence issues in free energy methods and finite
size effects. This paper presents the S0 method, which can be used to obtain
chemical potentials from static structure factors computed from equilibrium
molecular dynamics simulations under the isothermal-isobaric ensemble. The S0
method is demonstrated on the systems of binary Lennard-Jones particles,
urea--water mixtures, a NaCl aqueous solution, and a high-pressure
carbon-hydrogen mixture
Cartesian atomic cluster expansion for machine learning interatomic potentials
Full text linkMachine learning interatomic potentials are revolutionizing large-scale, accurate atomistic modeling in material science and chemistry. Many potentials use atomic cluster expansion or equivariant message-passing frameworks. Such frameworks typically use spherical harmonics as angular basis functions, followed by Clebsch-Gordan contraction to maintain rotational symmetry. We propose a mathematically equivalent and simple alternative that performs all operations in the Cartesian coordinates. This approach provides a complete set of polynormially independent features of atomic environments while maintaining interaction body orders. Additionally, we integrate low-dimensional embeddings of various chemical elements, trainable radial channel coupling, and inter-atomic message passing. The resulting potential, named Cartesian Atomic Cluster Expansion (CACE), exhibits good accuracy, stability, and generalizability. We validate its performance in diverse systems, including bulk water, small molecules, and 25-element high-entropy alloys
Cartesian atomic cluster expansion for machine learning interatomic potentials
Full text linkMachine learning interatomic potentials are revolutionizing large-scale, accurate atomistic modelling in material science and chemistry. Many potentials use atomic cluster expansion or equivariant message passing frameworks. Such frameworks typically use spherical harmonics as angular basis functions, and then use Clebsch-Gordan contraction to maintain rotational symmetry, which may introduce redundancies in representations and computational overhead. We propose an alternative: a Cartesian-coordinates-based atomic density expansion. This approach provides a complete set of polynormially indepedent features of atomic environments while maintaining interaction body orders. Additionally, we integrate low-dimensional embeddings of various chemical elements and inter-atomic message passing. The resulting potential, named Cartesian Atomic Cluster Expansion (CACE), exhibits good accuracy, stability, and generalizability. We validate its performance in diverse systems, including bulk water, small molecules, and 25-element high-entropy alloys
Predicting homogeneous nucleation rate from atomistic simulations
No full textPredictive modelling and quantitative understanding of nucleation is essential for predicting phase transformation processes in nature and precisely controlling material synthesis and processing. Atomistic modeling is a powerful tool for capturing the dynamical processes and investigating the underlying mechanism of nucleation, but it faces several key challenges.
In the present thesis, we tackled the problem of nucleation by attacking it on several fronts. First of all we employed as well as devised enhanced sampling strategies to make the computation affordable. In the case of homogeneous ice nucleation, we computed the free energy profile associated with a single nucleation pathway, and then accounted for the free energy gain from the possibility of exhibiting stacking disorders in the nucleus.
We then formulated a thermodynamic framework to bridge the gap between the microscopic and macroscopic pictures of nucleation, and thus provides a simple and elegant framework to verify and extend classical nucleation theory. Using this framework, we accurately and rigorously extracted different physical quantities that affect nucleation,
including the chemical potential, the interfacial free energy, and the Tolman length. By comparing the results that we obtained from simulations of homogeneous nucleation to the ones computed at the planar limit, we verified our thermodynamic framework, as well as benchmarked the accuracy of the classical nucleation theory.
Finally, we constructed a machine learning potential based on hybrid DFT data, in order to better model the interatomic interactions in water systems. We predicted the thermodynamic properties of liquid water as well as hexagonal and cubic ice, rigorously taking into account quantum nuclear motion, anharmonic fluctuations and proton disorder.
The ab initio description not only leads to structural properties, density isobar, and melting point in excellent agreement with experiments, but also provides insights on how nuclear quantum effects modulate the stabilities of different phases of water. In addition, this ab initio modelling of water opens up many possibilities of future work, including a first principle description of ice nucleation.
To sum up, the present thesis provides the key instruments for investigating nucleation using atomistic simulations, and represents a substantial development in the quantitative understanding of the nucleation phenomenon.COSM
Response Matching for generating materials and molecules
Full text linkMachine learning has recently emerged as a powerful tool for generating new
molecular and material structures. The success of state-of-the-art models stems
from their ability to incorporate physical symmetries, such as translation,
rotation, and periodicity. Here, we present a novel generative method called
Response Matching (RM), which leverages the fact that each stable material or
molecule exists at the minimum of its potential energy surface. Consequently,
any perturbation induces a response in energy and stress, driving the structure
back to equilibrium. Matching to such response is closely related to score
matching in diffusion models. By employing the combination of a machine
learning interatomic potential and random structure search as the denoising
model, RM exploits the locality of atomic interactions, and inherently respects
permutation, translation, rotation, and periodic invariances. RM is the first
model to handle both molecules and bulk materials under the same framework. We
demonstrate the efficiency and generalization of RM across three systems: a
small organic molecular dataset, stable crystals from the Materials Project,
and one-shot learning on a single diamond configuration
Computing chemical potentials of adsorbed or confined fluids
No full textThe chemical potential of adsorbed or confined fluids provides insight into their unique thermodynamic properties and determines adsorption isotherms. However, it is often difficult to compute this quantity from atomistic simulations using existing statistical mechanical methods. We introduce a computational framework that utilizes static structure factors, thermodynamic integration, and free energy perturbation for calculating the absolute chemical potential of fluids. For demonstration, we apply the method to compute the adsorption isotherms of carbon dioxide in a metal-organic framework and water in carbon nanotubes
Streaming Graph Embeddings via Incremental Neighborhood Sketching
No full textGraph embeddings have become a key paradigm to learn node representations and facilitate downstream graph analysis tasks. Many real-world scenarios such as online social networks and communication networks involve streaming graphs, where edges connecting nodes are continuously received in a streaming manner, making the underlying graph structures evolve over time. Such a streaming graph raises great challenges for graph embedding techniques not only in capturing the structural dynamics of the graph, but also in efficiently accommodating high-speed edge streams. Against this background, we propose SGSketch, a highly-efficient streaming graph embedding technique via incremental neighborhood sketching. SGSketch cannot only generate high-quality node embeddings from a streaming graph by gradually forgetting outdated streaming edges, but also efficiently update the generated node embeddings via an incremental embedding updating mechanism. Our extensive evaluation compares SGSketch against a sizable collection of state-of-the-art techniques using both synthetic and real-world streaming graphs. The results show that SGSketch achieves superior performance on different graph analysis tasks, showing 31.9% and 21.9% improvement on average over the best-performing static and dynamic graph embedding baselines, respectively. Moreover, SGSketch is significantly more efficient in both embedding learning and incremental embedding updating processes, showing 54x-1813x and 118x-1955x speedup over the baseline techniques, respectively.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Web Information System
- …
