RESUMO
Machine-learning force fields (MLFF) should be accurate, computationally and data efficient, and applicable to molecules, materials, and interfaces thereof. Currently, MLFFs often introduce tradeoffs that restrict their practical applicability to small subsets of chemical space or require exhaustive datasets for training. Here, we introduce the Bravais-Inspired Gradient-Domain Machine Learning (BIGDML) approach and demonstrate its ability to construct reliable force fields using a training set with just 10-200 geometries for materials including pristine and defect-containing 2D and 3D semiconductors and metals, as well as chemisorbed and physisorbed atomic and molecular adsorbates on surfaces. The BIGDML model employs the full relevant symmetry group for a given material, does not assume artificial atom types or localization of atomic interactions and exhibits high data efficiency and state-of-the-art energy accuracies (errors substantially below 1 meV per atom) for an extended set of materials. Extensive path-integral molecular dynamics carried out with BIGDML models demonstrate the counterintuitive localization of benzene-graphene dynamics induced by nuclear quantum effects and their strong contributions to the hydrogen diffusion coefficient in a Pd crystal for a wide range of temperatures.
RESUMO
The Covid-19 pandemic has caused the alteration of many aspects of the solid waste management chain, such as variations in the waste composition, generation and disposal. Various studies have examined these changes with analysis of integrated waste management strategies; qualitative studies on perceived variations and statistical evaluations based on waste collected or disposed in landfills. Despite this information there is a need for updated data on waste generation and composition, especially in developing countries. The objective of this article is to develop a data sampling and analytical approach for the collection of data on household waste generation and composition during the pandemic; and, in addition, estimate the daily generation of masks in the study area. The proposed methodology is based on the principles of citizen science and utilizes virtual tools to contact participants, and for the training and collection of information. The study participants collected the information, installed segregation bins in their homes and trained their relatives in waste segregation. The article presents the results of the application of the methodology in an urban district of Lima (Peru) in August 2020. The results suggest an apparent decrease in household waste per capita and a slight increase in plastics composition in the study area. It is estimated that each participant generates 0.124 masks per day and 0.085 pairs of gloves per day. The method developed and results presented can be used as a tool for public awareness and training on household waste characterization and segregation. Furthermore it can provide the necessary evidence to inform policy directives in response household waste issues and Covid-19 restrictions.