WORKSHOP IN SOLITON THEORY, NONLINEAR DYNAMICS AND MACHINE LEARNING STARTS IN AUGUST
12 August 2023
The workshop in Soliton Theory, Nonlinear Dynamics and Machine Learning brings together over sixty participants at Varna Free University in the period 12-17 August 2023. One of the objectives of the workshop is to familiarise the participants with two of the main topics of the scientific forum , namely with the properties and behaviour of solitons and methods for their production, as well as with applications of ideas from theoretical and experimental physics in creating more mathematically based methods for machine learning. Through reports and discussions, students, young scientists and leading researchers from Bulgaria, France, Germany, Ireland, Romania, Switzerland and England will present their achievements in these and other fields.
The theory of solitons and their applications in sections of physics such as nonlinear optics, plasma physics, hydrodynamics, etc. emerged about 50 years ago as a new field in mathematical physics. Efficient mathematical methods were developed, allowing solving exactly entire classes of nonlinear partial differential equations that describe important processes in the aforementioned physical phenomena. In the process of its development, the theory of solitons stimulated the emergence of new sections in mathematics as well and in particular in the theory of Lie algebras, the theory of analytic functions, etc. It is about the theory of graded infinite-dimensional algebras, now known as Kac-Moody algebras, new approaches to solving Riemann-Hilbert problems. Solitons themselves can be described as sustained nonlinear waves. Their properties practically do not depend on the physical nature of the medium in which they propagate: plasma, fluids, nonlinear optical medium. During an interaction, two solitons can be strongly deformed, but after the interaction they regain their shape and speed.
On the other hand, machine learning (ML), with its methods, serves as the formal basis of artificial intelligence and is the most rapidly developing field of science in the last decade. The conference will present new ideas on the development of both discriminative and generative ML models using classical and graph artificial neural networks. Of particular interest in the forum are the "transformer" type architectures that have gained great popularity. In a special session, solutions to equations from the general theory of relativity by means of neural networks with nested laws of mechanics will be presented.
The workshop was organised in a face-to-face and remote mode of participation by the Institute for Modern Physical Research and Varna Free University, supported by the Scientific Research Fund of the Ministry of Education and Science of the Republic of Bulgaria, through a procedure for supporting international scientific forums held in the Republic of Bulgaria under contract KP-06-MNF/7 dated 23.05.2023.
Sponsors of the event are Management Financial Group AD and Quanterall OOD.
The theory of solitons and their applications in sections of physics such as nonlinear optics, plasma physics, hydrodynamics, etc. emerged about 50 years ago as a new field in mathematical physics. Efficient mathematical methods were developed, allowing solving exactly entire classes of nonlinear partial differential equations that describe important processes in the aforementioned physical phenomena. In the process of its development, the theory of solitons stimulated the emergence of new sections in mathematics as well and in particular in the theory of Lie algebras, the theory of analytic functions, etc. It is about the theory of graded infinite-dimensional algebras, now known as Kac-Moody algebras, new approaches to solving Riemann-Hilbert problems. Solitons themselves can be described as sustained nonlinear waves. Their properties practically do not depend on the physical nature of the medium in which they propagate: plasma, fluids, nonlinear optical medium. During an interaction, two solitons can be strongly deformed, but after the interaction they regain their shape and speed.
On the other hand, machine learning (ML), with its methods, serves as the formal basis of artificial intelligence and is the most rapidly developing field of science in the last decade. The conference will present new ideas on the development of both discriminative and generative ML models using classical and graph artificial neural networks. Of particular interest in the forum are the "transformer" type architectures that have gained great popularity. In a special session, solutions to equations from the general theory of relativity by means of neural networks with nested laws of mechanics will be presented.
The workshop was organised in a face-to-face and remote mode of participation by the Institute for Modern Physical Research and Varna Free University, supported by the Scientific Research Fund of the Ministry of Education and Science of the Republic of Bulgaria, through a procedure for supporting international scientific forums held in the Republic of Bulgaria under contract KP-06-MNF/7 dated 23.05.2023.
Sponsors of the event are Management Financial Group AD and Quanterall OOD.