1.4 Computer Simulation and Computational Experiment
7
4. Conducting a computational experiment. A study plan is being developed for
the experiment. The analysis of the results is carried out; conclusions are drawn
regarding the properties of the model obtained.
In the process of the experiment, it may become clear that it is necessary to:
• adjust study plan;
• choose another method for solving the problem;
• improve the algorithm for obtaining results;
• refine mathematical model;
• make changes to the formulation of the problem.
In this case, the process returns to the appropriate stage and begins again. That is,
the process of computer simulation is often iterative.
Currently, there are many software environments for computer modeling, to simplify the development of computer models and conduct computational experiment.
Many of these environments allow you to independently create new models from
scratch, as well as have extensive libraries of ready-made models (components),
on the basis of which you can build other more complex models. It is also usually
possible to create your own libraries.
An existing modeling environment is usually based on a modeling language.
Modern modeling environments provide features like:
• deriving mathematical equations in the most natural way possible for the user;
• selection of numerical methods for solving algebraic, differential, algebraic–differential, and other equations;
• graphical interface for creating multi-component models;
• interface for setting up and conducting a computational experiment;
• graphic primitives for creating model visualization;
• other opportunities (e.g., conducting an experiment in real time, the possibility of
user intervention in the experiment, and others).
Since equations are written in relatively free form, all modeling environments
have a tool like a solver. The solver analyzes and transforms the equations to apply
numerical methods to them. It is assumed that the user must monitor the fulfillment
of the conditions of existence and the uniqueness of the solution.
1.5 Classification of Computer Models
Computer implementations of mathematical models can be classified according to
several criteria: by interaction with the outside world, by internal structure, by type
of model time, and others.
By the type of interaction with the outside world, models can be divided into open
and isolated. The state variables that characterize the model are divided into internal
and external. In isolated models, all variables are internal; that is, they are not visible
7
4. Conducting a computational experiment. A study plan is being developed for
the experiment. The analysis of the results is carried out; conclusions are drawn
regarding the properties of the model obtained.
In the process of the experiment, it may become clear that it is necessary to:
• adjust study plan;
• choose another method for solving the problem;
• improve the algorithm for obtaining results;
• refine mathematical model;
• make changes to the formulation of the problem.
In this case, the process returns to the appropriate stage and begins again. That is,
the process of computer simulation is often iterative.
Currently, there are many software environments for computer modeling, to simplify the development of computer models and conduct computational experiment.
Many of these environments allow you to independently create new models from
scratch, as well as have extensive libraries of ready-made models (components),
on the basis of which you can build other more complex models. It is also usually
possible to create your own libraries.
An existing modeling environment is usually based on a modeling language.
Modern modeling environments provide features like:
• deriving mathematical equations in the most natural way possible for the user;
• selection of numerical methods for solving algebraic, differential, algebraic–differential, and other equations;
• graphical interface for creating multi-component models;
• interface for setting up and conducting a computational experiment;
• graphic primitives for creating model visualization;
• other opportunities (e.g., conducting an experiment in real time, the possibility of
user intervention in the experiment, and others).
Since equations are written in relatively free form, all modeling environments
have a tool like a solver. The solver analyzes and transforms the equations to apply
numerical methods to them. It is assumed that the user must monitor the fulfillment
of the conditions of existence and the uniqueness of the solution.
1.5 Classification of Computer Models
Computer implementations of mathematical models can be classified according to
several criteria: by interaction with the outside world, by internal structure, by type
of model time, and others.
By the type of interaction with the outside world, models can be divided into open
and isolated. The state variables that characterize the model are divided into internal
and external. In isolated models, all variables are internal; that is, they are not visible