Strategy
- quest.strategy.random() x in the interval [0, 1).
- class quest.strategy.Strategy[source]
A Strategy represents how a sprite responds to the environment.
It is useful to extract
Strategyas a separate class (rather than making it part of the NPC class) because there might be many sprites in a game which need to use the same strategy. In the default implementation ofquest.sprite.NPC.on_update(), the NPC will set its course using itsStrategyif it has one.- choose_course(sprite, game)[source]
Returns a (x, y) vector representing the chosen course.
In the default implementation, returns (0, 0) which will cause the sprite to stop. To create new strategies, override this method and define helper methods as necessary.
- Parameters
sprite – The sprite who is about to act.
game – The game object (to access attributes useful in choosing the course).
- class quest.strategy.RandomWalk(change_prob=0.1)[source]
A strategy which causes the sprite to walk randomly, with some chance of changing direction at each step.
- Parameters
change_prob (float) – The probability (between 0 and 1) that the sprite will choose
step. (a new direction on a) –
- choose_course(sprite, game)[source]
Possibly chooses a new random direction, then returns the current direction.
- Parameters
sprite – The sprite who is about to act.
game – The game object (to access attributes useful in choosing the course).
- set_random_direction()[source]
Sets attributes x and y to a random (x, y) direction.
Think of this as choosing a random point on a circle. Tau is another word for 2 * pi, which represents the angle measure of the full circle, so if we multiply it by random() (which returns a random number evenly distributed between 0 and 1), we get a randomly-chosen angle measure between 0 and the full circle. Then we can use cosine and sine to get the x and y components of this point.
- class quest.strategy.DividedStrategy(strategy_a, strategy_b, ab_prob, ba_prob)[source]
Builds a complex strategy based on two existing strategies and transition probabilities.
One way to get more complex behavior is to define two simpler behaviors, and then to have a certain probability of switching between them. For example, some games have enemy NPCs which sometimes chase the player and other times wander around randomly. The transition probabilities (the chance of switching) are not symmetric. For example, you might want to define a game where sprites have a low chance of going from normal to desperate, and then zero chance of going from desperate back to normal. The idea of composing strategies from other strategies is an example of functional programming.
This pattern could be extended to n strategies, with a matrix of n * n transition probabilities. This more general case is called a Markov chain, and gets used quite a lot in computer science.
- Parameters