MATHEMATICAL LIBRARIES AND SOLVERS 

Demanded both in applications and as a system software components

We’ve been specializing in the development of math libraries for various platforms and solvers for different science-intensive projects during almost twenty years. Herein are some examples:

Implementation of math libraries for Elbrus platform (Moscow, since 2014)

Development of new functions for high-performance math library, porting and optimization of existing widely used open source libraries for efficient operation on the hardware platform "Elbrus":

  • Library of  math functions EML - implementation of linear algebra functions for sparse matrices and image processing functions
  • OpenCV library - porting to the platform and optimizing the computer vision package
  • LibreSSL library - optimization to the platform the encryption library
  • JavaCV library - transfer to the platform the library of java-interfaces for OpenCV
  • Development of the library of digital signal processing functions for a dedicated  co-processor

Technologies: methods of program optimization (profiling, optimization of cycles), development of optimized math algorithms and signal processing algorithms, programming technologies (C, C ++, Assembler, etc.)

New solver for legacy software (USA, 2014-2015)

Development of the software solver for systems of second-order parabolic differential equations and the finite volume method (without a graphical interface). After testing solver was integrated with the customer software system.

Technologies: C ++ / gcc / OS Linux / OpenMP compiler for optimizing / make

Solvers for processing well logging data (USA, 2004-2006)

Superfast solution of the direct problem and fast solution of the inverse problems of well modeling, determination of the electromagnetic characteristics of the medium was implemented as a part of the software package for automated processing of electromagnetic  well logging data of oil-bearing strata.

Technologies: С ++, XML.

Modeling of electromigration processes (USA, 2002-2003)

Construction of the Cu-Si chip model. Carrying out calculations, comparison with real data.

Technologies: C ++, finite element method, iterations: GenCG, DD-preconditioning, Fourier transformation, Predictor-corrector explicit difference scheme. The simultaneous solution of the diffusion / thermal conductivity / filtration / electrostatics / thermoelasticity equations.