Laboratory of Structural Methods of Data Analysis in Predictive
Modeling Moscow Institute of Physics and Technology
ENG
 Логин: Пароль:

# О лаборатории

PreMoLab (Laboratory of Structural Methods of Data Analysis in Predictive Modeling) is a research lab at the Moscow Institute of Physics and Technology created in 2011 on a “mega-grant” of Russian government. The lab gathers top world expertise and conducts research in modern stochastics and optimization aimed at applications to modeling complex technological and economic systems. Research at PreMoLab is organized according to mathematical research fields, which represent academic subjects, and interdisciplinary application areas, in which the lab has special competence in modeling, analysis, stochastic and optimization treatment, and simulation. The lab has an extensive student training program, with most of the lab's researchers advising M.Sc. and Ph.D. projects of lab students (see list of suggested projects) and teaching at MIPT and other universities in Moscow. It also hosts an open weekly research seminar, a young researchers' seminar at the MIPT suburban campus, and organizes a series of conferences, workshops, and invited courses.

Ближайшие мероприятия:
 19.06.2016Зеленогорск (р-н Санкт-Петербурга) VIII Традиционная Школа «Управление, информация и оптимизация» Летняя школаVIII Традиционная Школа «Управление, информация и оптимизация»Традиционная молодежная школа ежегодно собирает вместе молодых ученых и ведущих мировых исследователей в области оптимизации, управления и статистики. Школа организуется Институтом проблем управления РАН и лабораторией ПреМоЛаб.
Остальные мероприятия
Последние публикации:
 Primal-Dual Subgradient Method for Huge-Scale Linear Conic Problems In this paper we develop a primal-dualsubgradient method for solving huge-scaleLinear Conic Optimization Problems. Our main assumption is that the primal cone isformed as a direct product of many small-dimensional convex cones, and that the matrix Aof corresponding linear operator isuniformly sparse. In this case, our method canapproximate the primal-dual optimal solution with accuracy ε in O(1/ε^2) iterations. Atthe same time, complexity of each iteration of this scheme does not exceed O(rq log_2 n) operations, where r and q are the maximal numbers of nonzero elements in the rows andcolumns of matrix A, and n is the number variables. Авторы: Нестеров Юрий Евгеньевич, Шпирко Сергей Валерьевич Дата: 30 декабря 2014 Теорема Бернштейна–фон Мизеса в непараметрическом случае Авторы: Спокойный Владимир Григорьевич, Панов Максим , Гончаров Федор Олегович Дата: 29 декабря 2014 О концентрации целевого параметра в статистических моделях с растущей размерностью Авторы: Панов Максим Дата: 29 декабря 2014
Остальные публикации