Reading Diary Week 5

5 terms

Article for this week (5): J. Schmidt, L. Pettersson, C. Verdozzi, S. Botti, and M. A. L. Marques, “Crystal graph attention networks for the prediction of stable materials.,” Sci. Adv., vol. 7, no. 49, p. eabi7948, Dec. 2021, doi: 10.1126/sciadv.abi7948. I decided to add another article becuse it's very important article for my current work.

Attention mechanism
phrase
The attention mechanism was introduced to improve the performance of the encoder-decoder model for machine translation. The idea behind the attention mechanism was to permit the decoder to utilize the most relevant parts of the input sequence in a flexible manner, by a weighted combination of all of the encoded input vectors, with the most relevant vectors being attributed the highest weights [https://machinelearningmastery.com/the-attention-mechanism-from-scratch/].
Example: The update function is based on the attention mechanism that has revolutionized natural language processing and has also found application in graph neural networks.
en: Мехнизм определения релевантной части входной последовательности (прямой перевод не отражает сути понятия, т.к. в русскоязычном сегменте еще не сформировано аналогичнсое понятие; представлен ниболее полный перевод)
null
Crystal graph
phrase
Representation of the crystal structure, in the form of a mathematical object = graph structure, which displays the bonds between atoms
Example: The crystal structure prototype will enter our model as a crystal graph.
en: графовая реперезнтация кристаллической структуры
In this case, the crystal structure is a mixed perovskite, and we consider the five nearest neighbors. Here, blue edges represent first neighbors, black edges represent second neighbors, and green edges represent third neighbors. During the message passing ste
Graph neural networks
phrase
Graph networks are a way of applying classical neural network models to graph data. 1) https://habr.com/ru/company/vk/blog/557280/ 2) https://distill.pub/2021/gnn-intro/
Example: Graph neural networks for crystal structures typically use the atomic positions and the atomic species as input.
en: Графовая нейронная сеть
Schematic for a GCN architecture, which updates node representations of a graph by pooling neighboring nodes at a distance of one degree.
K-points per reciprocal atom
phrase
A very complex highly specialized concept. Reflects the method of setting the accuracy of quantum mechanical calculations. The bottom line is that the accuracy of the calculation is set by a certain spatial grid (k-point grid), in the nodes of which the main values ​​of the electron density will be calculated and then integrated to obtain the full characteristics of the system under consideration. By specifying this grid as the density in the reciprocal space of the Fourier images of the lattice, one can introduce a dynamic value that can be given by just one number and will inversely depend on the ratio of the translation vectors describing the lattice. This method is interesting as an alternative to manually defining all three grid dimensions.
Example: We used the projector augmented wave datasets of version 5.2 with a cutoff of 520 eV. The Brillouin zone was sampled by Γ-centered k-point grids with a uniform density calculated to yield 1000 k-points per reciprocal atom.
en: количество К-точек, приведенное на обратный атом
null
Сonvex hull
phrase
The most complex concept among the most used. In geometry, the convex hull or convex envelope or convex closure of a shape is the smallest convex set that contains it. But the meaning of this phrase is most relevant in materials science. In order to explain this concept, one should take into account the energy of the ground state. Each substance or single-phase material has its own free energy; if, for example, we choose a certain binary composition and plot the energies of all crystalline modifications on a standard enthalpy diagram as a function of the composition under certain conditions (pressure and temperature), then a minimum line will clearly stand out on such a diagram. This line is a convex hull and symbolizes the main states of all compositions in this compositional space. Everything that lies above this line will be less preferable for formation or, in other words, lies above the convex hull. The convex hull is a cornerstone concept in modern computational materials science.
Example: The main and most efficient approach to high-throughput searches is to calculate the distance to the convex hull of thermodynamic stability for all compositions of a single prototype.
en: прямого перевода не существует, употребляется в контексте переводных статей "как есть". Однако по смыслу ближайший перевод может выглядеть, как: "поверхность минимальной энергии композиционной диаграммы свободной энергии".
null