Makogon Volodymyr Mykolayovych

Faculty: Computer Science and Technology
Speciality: System programming

Scientific adviser: Svyatniy V.A.

Abstract of the theme of master's work

Organisation of dialog subsystem of distributed parallel simulation environment

Currently, the complexity of the tasks that appear before all the areas of science and technology is constantly growing. Increasingly, scientists and experts have to  use models of complex dynamical systems (CDS) to obtain the best solution, or to simulate situations when the system crashes. There are such complex dynamic systems as: technological systems, transfer lines, automated technical facilities, power plants, thermodynamic installation of the pipeline network, etc.

Application of parallelization in solving of CDS opens fundamentally new opportunities. However, it makes an additional complications:
1) The main problem of the parallel simulation of SDS - the construction of distributed parallel modeling environments (DPME), which will satisfy all the requirements, have interface for efficient interaction with user. User or expert in any case should not be distracted from the tasks of his scientific area, and thinl over the software and hardware architecture of DPME.
2) Powerful computing resources are required. From this perspective, the scientific and technological progress depends on the level of development of computer technology that can provide a model supporting various problems of science and technology.

As it turned out as a result of the analysis of existing distributed parallel modeling environments, the organization of friendly dialogue between the user, domain expert and resources of DPME is a very serious problem. Interfaces, which provide such projects as SEGL (project of the Institute IHR and HLRS of Stuttgart University), DEISA and many others, are not user friendly. User-friendliness means that the domain expert can fully concentrate himself on his task in terms that are understandable to him (terms of its domain). In this case, must be organized highly intelligent user support in all phases of his work. About importance of interface quality of any software wrote Jeff Raskin (creator of the project Apple "Mac" and one of the top experts in user interfaces) in his book "Interface: New Directions in the design of computer systems:" If a system’s one-on-one interaction with its human user is not pleasant and facile, the resulting deficiency will poison the performance of the entire system, however fine that system might be in its other "[10]. So, consideration of the DPME as a dialog of user, of developer models (domain expert) with a multifaceted resource environment, the theoretical grounding and the development of effective support to the dialogue at all stages of construction and application of models of dynamical systems is a new and very important task.

1. 1. Masyuk A.L. Subsystem of dialogue in parallel modeling environment, for mineventilation network / A.L. Masyuk. / / Scientific works of Donetsk National Technical University. Issue 70. IKOT Series: - Donetsk: Donetsk National Technical University, 2003.
At the local level, only few Masters have devoted their works conversational media access to the parallel simulation. The most complete subsystem dialogue considered Masyuk A.L. In his work the selection of the subsystem of dialogue as a separate unit of the system modeling environment is grounded, the main requirements are  demanded.

2. Master DonNTU Brahnov P.G.
This problem was also involved Brahnov P.G. Theme of master's work: "Parallel simulation of the system of management of network dynamic object with lumped parameters", 2008. http://www.masters.donntu.ru/2008/fvti/brakhnov/diss/index.htm.
In Brahnov’s work there are following scientific results:

• Subsystem of dialogue (DS) as a system unit of POME is defined, the requirements to DS [11];
• The functional and algorithmic structure of the DS POME-oriented implementation of a MIMD-systems is developed [11];
• The functional and algorithmic structure of the PD POMS for interactive support is developed [11];
• The theoretical analysis and development of a technique of PD POMS [11];
• A graphic design PD POMS, implementing visual communication model developers with the system and the required mapping of simulation results [11];
• Completed implementation and experimental study of the basic components of PD POMS [11].

DEISA (Distributed European Infrastructure for Supercomputing Applications) – a project that includes supercomputers of the European Union. DEISA project supports the European research, providing them with access to distributed supercomputer environment. In a distributed network of supercomputers combined eleven supercomputers from seven European countries [12]:

• Max Planck Gesellschaft, Germany;
• Bavarian Academy of Sciences and Humanities, Germany;
• Barcelona Supercomputing Center, Spain;
• CINECA, Italy;
• CSC, Scientific Computing Ltd, Finland;
• European Centre for Medium-Range Weather Forecasts, United Kingdom;
• Julich Research Centre, Germany;
• Institut du Developpement et des Ressources en Informatique Scientifique (CNRS), France;
• Stichting Academisch Rekencentrum Amsterdam, Netherlands;
• Edinburgh Parallel Computing Centre, United Kingdom;
• High Performance Computing Center Stuttgart (HLRS), University of Stuttgart, Germany;

DEISA’s infrastructure integrated high-speed (10 Gbps) network connection, which provides service providers GEANT2 (official site http://www.geant2.net/) - at the European level, and the NRENs - at the national level [12]. In order to take advantage of supercomputing resources DEISA, the user should contact one of the partner institutions of this project. This partners must have special software installed. In addition, after receiving permission the user must install on his computer certain software product that makes some inconveniences. More information about the project DEISA can be found on the official site http://www.deisa.eu/ .

DIANA (Dynamic simulation and numerical analysis tool) – simulating environment, designed for numerical analysis of problems of chemical engineering and systems biology. DIANA - «open source» Project of Magdeburg Institute of complex dynamic systems, of Max Planck. Key components of the system DIANA: an object-oriented models, numerical kernel, and use a scripting language Python as a powerful command line interface. Models are created using the tool ProMoT (Process Modeling Tool, the official site http://www.mpi-magdeburg.mpg.de/projects/promot) and converted to C + + code, but can also be controlled interactively as ordinary objects Python. ProMoT - software that is designed to create complex technical and biological systems, and further control of them. For applications relating to systems biology ProMoT uses a special modular simulation models and libraries. More information about the project DIANA can be found at the official site http://www.mpi-magdeburg.mpg.de/projects/diana.

DIVA (Dynamische Simulation verfahrenstechnischer Anlagen) – the predecessor of DIANA - simulating environment, a project of the Institute ISR (currently ISYS) University of Stuttgart,. Numerical Algorithms Project DIVA were written in Fortran, and the interface with users is through the command line. It was later written GUI (graphic user interface) - extension of DIVA, which provides a graphical interactive control of simulation, numerical algorithms and models. To implement the GUI and connect with DIVA-simulator were used Java and Corba [13].

SEGL (Science Experimental Grid Laboratory) – "open source" project of the Institute IHR and HLRS University of Stuttgart to describe, test, launch, monitoring and analysis of large scientific and industrial experiments requiring high-performance computing. To describe experiments SEGL uses GriCol (Grid Concurrent Languge), which represents the entire experiment as a set of blocks which can be connected in any sequence, forward data between them. More information about the project SEGL is in the official website http://segl.hlrs.de/.

Develop the simulator software, which implements the functions of the subsystem of dialogue of DPME, which will improve user-friendliness DPME to users and developers of models and expand the area of using of parallel simulations in different domains. A new approaches to parallelization will developed, as well as methods for constructing models of complex dynamical systems to functions within the subsystem will developed, experimentally confirmed the decomposition of DPME into subsystems.
Objectives:

• To analize the state of development in the modeling software (MSW);
• To develop subsystem’s concepts (requirements, functions and structure of the HW / SW);
• To develop algorithms, structures, approaches, implementation and experimental research of them in DPME.

Building of prototype of the subsystem, using of MSW in the DPME. Theme is part of the scientific co-operation department of the CI cathedra with HLRS of Stuttgart University. Using a subsystem in the learning process and in ready-made DPME.