SubjectsSubjects(version: 845)
Course, academic year 2018/2019
   Login via CAS
Modern computational physics II - NEVF161
Title in English: Moderní počítačová fyzika II
Guaranteed by: Department of Surface and Plasma Science (32-KFPP)
Faculty: Faculty of Mathematics and Physics
Actual: from 2013 to 2019
Semester: summer
E-Credits: 5
Hours per week, examination: summer s.:2/1 MC [hours/week]
Capacity: unlimited
Min. number of students: unlimited
State of the course: taught
Language: Czech
Teaching methods: full-time
Guarantor: RNDr. Štěpán Roučka, Ph.D.
doc. RNDr. Pavel Kocán, Ph.D.
Annotation -
Last update: doc. RNDr. Jiří Pavlů, Ph.D. (14.01.2019)
Modern methods of computational physics - application of neural networks in physics. Advanced techniques of computer modelling.
Course completion requirements - Czech Sign Language
Last update: RNDr. Štěpán Roučka, Ph.D. (08.10.2017)

Podmínkou pro získání zápočtu je vytvoření zápočtového programu

Literature - Czech
Last update: doc. RNDr. Jiří Pavlů, Ph.D. (14.01.2019)

Šíma J., Neruda R.: Teoretické otázky neuronových sítí. Matfyzpress, Praha 1996.

Gershenfeld N.: The Nature of Mathematical Modeling, Cambridge University Press, Cambridge 1999.

Landau D.P., Binder K.: A Guide to Monte Carlo Simulations in Statistical Physics, Cambridge Univ. Press, Cambridge 2005.

Syllabus -
Last update: doc. RNDr. Jiří Pavlů, Ph.D. (14.01.2019)
1. Neural networks
Neuron and neural network - biological neuron, formal neuron, biological and mathematical neural network.

Classical models of neural networks - network of perceptors, multilayer network, back propagation.

Learning of neural networks - training set, supervised learning, self organizing, learning of cyclic neural networks, problem of relearning.

Complexity of neural networks - biasing function, logic circuits, cyclic neural networks.

Neural networks and genetic algorithms. Use of neural networks for image processing. Further applications of neural networks in physics.

Fuzzy logic - basic concepts. Combination of fuzzy logic and neural networks.

2. Advanced techniques of computer modelling
Advanced algorithms of molecular dynamics method in more dimensions.

Particle-In-Cell method, efficient solvers of Poisson equation - conjugated gradients, multigrids, direct methods, LU decomposition, Fast Fourier Transform methods. Efficient force calculation - tree algorithms, Ewald summation, Fast Multipole Method. Deterministic modelling of trajectories of charged particles in external electric and magnetic fields.

Advanced algorithms in Monte Carlo method, sampling in statistical physics.


Particle hybrid modelling in several dimensions. Fluid modelling in several dimensions. Scattering processes in continuum models. Hybrid modelling - combination of fluid and particle approaches is spatial domain, in velocity domain, iterative approach.

Charles University | Information system of Charles University |