arXiv subjects

Stanislav Filippov

Publications and source records attributed to Stanislav Filippov.

2 recordsLinked to original sources

arXiv2023

Ab initio Determination of Phase Stabilities of Dynamically Disordered Solids: rotational C2 disorder in Li2C2

The temperature-induced orthorhombic to cubic phase transition in Li2C2 is a prototypical example of a solid to solid phase transformation between an ordered phase, which is well described within the phonon theory, and a dynamically disordered phase with rotating molecules, for which the standard phonon theory is not applicable. The transformation in Li2C2 happens from a phase with directionally ordered C2 dimers to a structure, where they are dynamically disorderd. We provide a description of this transition by employing ab initio molecular dynamics (AIMD) based stress-strain thermodynamic integration on a deformation path that connects the ordered and dynamically disordered phases. The free energy difference between the two phases is obtained. The entropy that stabilizes the dynamically disordered cubic phase is captured by the behavior of the stress on the deformation path. We also show that a combination of the stress-strain thermodynamic integration and machine learning force field methodologies appears as a promising path in studies of dynamically disordered materials.

cond-mat.mtrl-sci↗

arXiv2017

Deep Diagnostics: Applying Convolutional Neural Networks for Vessels Defects Detection

Coronary angiography is considered to be a safe tool for the evaluation of coronary artery disease and perform in approximately 12 million patients each year worldwide. [1] In most cases, angiograms are manually analyzed by a cardiologist. Actually, there are no clinical practice algorithms which could improve and automate this work. Neural networks show high efficiency in tasks of image analysis and they can be used for the analysis of angiograms and facilitate diagnostics. We have developed an algorithm based on Convolutional Neural Network and Neural Network U-Net [2] for vessels segmentation and defects detection such as stenosis. For our research we used anonymized angiography data obtained from one of the city's hospitals and augmented them to improve learning efficiency. U-Net usage provided high quality segmentation and the combination of our algorithm with an ensemble of classifiers shows a good accuracy in the task of ischemia evaluation on test data. Subsequently, this approach can be served as a basis for the creation of an analytical system that could speed up the diagnosis of cardiovascular diseases and greatly facilitate the work of a specialist.

cs.CV↗