• No results found

Database in Modeling and Simulation

In document MCA 504 Modelling and Simulation (Page 127-130)

Structure 7.0 Objective

7.2. Database in Modeling and Simulation

The objective of the database and its data model in simulation is to provide a flexible representation of data for conducting computational simulations. The simulation data model is closely related to the data model for experimental testing .A data model is a specific description of the type of information and relationships between the information needed for a problem domain. For computational simulation, the simulation data model encompasses input data needed to define a simulation model and computational procedure, the output data from a simulation, and information about the simulation software and its execution that transformed the input data to the output data. To enable searching and querying in the data repository, the data model includes information that will be useful for a wide variety of queries. Generally the data model for simulation has been developed to allow development of simulation applications within independently of any specific simulation program.

The modeling of data is a critical activity for the database or repository to meet the needs of applications and users . The term data model was first introduced by the Codd in 1980 to include:

1. A collection of data object types that form the basic building blocks for a database system that defines the information of a problem domain.

3. Operators that can be applied to the objects for retrieving data or other purposes. Data models and the process of data modeling have become an essential first step in designing databases and repositories. Most data modeling is based on the concept of entities and relations, which were first developed as the so-called entity-relationship (ER) model. Entities are types of information of data and relationships represent the associations between entities. The modern approaches for data modeling use the concepts of object-oriented design in which entities are represented as classes, which are templates for the software behavior of the objects generated by a class. A class has a name, attributes, constraints, and relationships with objects of other classes. In an object- oriented sense, classes define valid operations on objects of the class. Relationships between classes may be associations with cardinality, an inheritance relationship, or other advanced relationships such as aggregation or composition.

The object-oriented approach is used for developing the high-level view of the simulation data model.

7.2.1 Definition of the Simulation Data Model

In defining a data model for simulation applications, the first decision was on the granularity of the data to be represented. The decision is primarily based on what are the most important user needs for querying, searching, and accessing computational simulation information. The decision was made to use a high-level view as interrelated files used in a simulation, along with descriptive information about the content of the files. This approach allows users to search for files for simulations that meet search criteria about the problem, software, project, and other high-level attributes. An alternative would be to provide a representation of the structural or geotechnical systems being analyzed. This latter approach would require a very detailed data on all aspects of a simulation including element types, materials, geometry, constraints, etc. It is not expected that many users would browse or query a database for simulation using search criteria that requires such specific information. The difficulty of the development of a detailed data model and implementation in a database system would outweigh the benefits to users.

Figure 1 is a basic view of the database required for simulation. The primary concepts are a simulation event, input files that may be parameterized, and output files produced by a software program.

There is a tradeoff between removing redundant data from the database. The functional relationships between the data and efficiency in the implementation of a database system. Redundant data or relationships should not be represented because such data are difficult to update and can lead to inconsistent information. Normalization theory is used to analyze the functional relationships and reduce redundancy. However, highly normalized data requires joins between data types, which may be inefficient unless the database system has very efficient query optimization. For the data model for simulation, redundancies are avoided, and most of the information is in at least third normal form.

An important aspect of a data model is constraints, which can take the form of valid ranges or values of attributes, cardinality, or on rules governing relationships. Database systems have varying capabilities for defining and enforcing constraints specified in a data model. In general, the data model for simulation has only the essential constraints needed for the integrity of the data. In the following subsections, the classes in the data model and their relationships are described. The assist in the definition, the description proceeds from simple classes to the more complicated classes.

Stored Data for Input Output Data For Analysis Simulation Model

Input Output

Analyst

Figure 1:A basic database concepts for Simulation

7.2.2 Data Representation for a Simulation Event

Using the notion of an event as defined in the data model for experiments, Simulation Event has attributes describing the event and time stamp information. Simulation Event is represents the execution of a specific computational simulation using a set of input data and producing a set of output data as computed on a computational resource. A simulation event is conducted by a Person for a Project. Simulation is executed on a computer hardware and software. The input for a Simulation Event is represented by an association with an Input File Parameter Set. The results of the simulation are stored in multiple files associated with Data Files.

7.2.3 Data Representation for Input Files for a Simulation

A key concept in conducting a computational simulation is that a problem may have a number of parameters and a specific simulation can be executed with a defined set of parameter values. This is represented in the data file by allowing input files to have parameter sets associated with them.

Input File is the top-level file that defines the input for a simulation. The input file is associated with a Person, Project. Since most models are defined with a number of files (such as using a source or include command), the data model represents the referenced files by an association with a Data File.

7.2.4 Data Representation for Output Files for a Simulation

A Simulation Event produces a number of output files, of which each is represented as a Data File. A Data File can be associated with two files, one representing the numerical values and the second containing descriptive information (such as an XML description) for the contents of the numerical file. Each type of file has a name, description, and a universal locator.

In document MCA 504 Modelling and Simulation (Page 127-130)