genetic algorithms

Evolutionary algorithms (modeling of the general laws of evolution) use only evolutionary principles. They have been successfully used for tasks like functional optimization and can easily be described in mathematical language.

Evolutionary algorithms include:

• genetic algorithms
• genetic programming
• evolutionary strategies
• evolutionary programming
• classifier systems

They model the basic principles in the theory of biological evolution — selection processes, mutations, and the restoration of a population of individuals.

The behavior of individuals is determined by the external environment. Many individuals are called populations. Such a population evolves in accordance with the rules of selection in accordance with the objective function, given by the external environment. Each individual (individual) population is assigned the value of its fitness in the external environment. Only the most adapted species breed. Recombination and mutation allow individuals to change and adapt to the environment. Such algorithms are related to adaptive search engines.

Evolutionary algorithms have been successfully used for problems like functional optimization and can easily be described in mathematical language.

Simulations of the evolution of cars

http://boxcar2d.com/

The program generates devices from wheels and body parts, the purpose of which is to drive as far as possible. A special algorithm evaluates the passage of the route by each car. The following machines are generated taking into account the previous successful, viable copies. They seem to evolve - if in the first generation samples without wheels can not drive even a meter, then in the third generation they will already have wheels and shape will be improved, and in the tenth generation a car will be able to overcome decent distances.

http://megaswf.com/serve/102223/

A genetic algorithm is used to create a car from the Box2D library. The colors show crossing-over and mutation for each individual of the population. You can choose the coefficient of mutation. Viable models are often generated.

This program uses a genetic algorithm to develop a two-dimensional car that will be “optimal” for a particular locality. The car has two wheels and two loads. The initial positions and radii of these four objects can be selected using an algorithm. The objects are connected by springs, the length of which is also chosen according to the algorithm. The load should never touch the ground.

The optimality of a partial solution (fitness fitness function) is determined by how long it exists before:

• Mass touches the ground.
• Run out.

In the beginning, the algorithm does not even know that the wheels touch the surface. Sometimes you can see how different types appear and disappear - for example, “unicycle”, especially in the early stages of the progress of the algorithm.

Genetic algorithms can converge much faster when choosing a fitness fitness function, population size. The developer made the simulation more interesting in visualization optimization than fast.

Word conversion using a genetic algorithm

http://planetcalc.ru/475/
http://planetcalc.ru/638/

The emulator turns the given word into another given word step by step, replacing one letter in the previous word, so that at each step the correct word is obtained. The emulator works with words that consist of 4 or 5 letters.

The user sets the maximum number of generations and the size of the population. After the calculations, the whole chain of transformations with the interpretation of each word is displayed.

Solving the Traveling Salesman Problem Using Genetic Algorithms

http://www.lalena.com/AI/Tsp/

The traveling salesman problem (TSP) is one of the most famous combinatorial optimization problems; it is finding the most profitable route that passes through the specified cities at least once and then returns to the original city. In terms of the task, the criterion of the route profitability (the shortest, cheapest, aggregate criterion, etc.) and the corresponding matrices of distances, costs, etc. are indicated. As a rule, it is indicated that the route should pass through each city only once.

The task of teamwork using genetic algorithms

http://www.lalena.com/AI/Ant/

ANT is a genetic program that emulates the behavior of ants in the search for food. To solve this problem, ants must act collectively and agree. The program of ant behavior is incorporated in the simulator, which contains such actions as: Move Forward, Take Food and the like. The genetic program is evolving, which allows ants to get to food faster.