{-# LANGUAGE TypeFamilies #-} -- | -- Module : Simulation.Aivika.Trans.Internal.Types -- Copyright : Copyright (c) 2009-2016, David Sorokin -- License : BSD3 -- Maintainer : David Sorokin -- Stability : experimental -- Tested with: GHC 8.0.1 -- -- It defines the implementation details of some types. You should never -- use it in ordinary simulation models. The module is destined for those -- who will extend the library. -- module Simulation.Aivika.Trans.Internal.Types (Specs(..), Method(..), Run(..), Point(..), Parameter(..), Simulation(..), Dynamics(..), Event(..), EventProcessing(..), EventQueueing(..), invokeParameter, invokeSimulation, invokeDynamics, invokeEvent) where import Simulation.Aivika.Trans.Generator -- | It defines the simulation specs. data Specs m = Specs { spcStartTime :: Double, -- ^ the start time spcStopTime :: Double, -- ^ the stop time spcDT :: Double, -- ^ the integration time step spcMethod :: Method, -- ^ the integration method spcGeneratorType :: GeneratorType m -- ^ the type of random number generator } -- | It defines the integration method. data Method = Euler -- ^ Euler's method | RungeKutta2 -- ^ the 2nd order Runge-Kutta method | RungeKutta4 -- ^ the 4th order Runge-Kutta method deriving (Eq, Ord, Show) -- | It indentifies the simulation run. data Run m = Run { runSpecs :: Specs m, -- ^ the simulation specs runIndex :: Int, -- ^ the current simulation run index runCount :: Int, -- ^ the total number of runs within the experiment runEventQueue :: EventQueue m, -- ^ the event queue runGenerator :: Generator m -- ^ the random number generator } -- | It defines the simulation point appended with the additional information. data Point m = Point { pointSpecs :: Specs m, -- ^ the simulation specs pointRun :: Run m, -- ^ the simulation run pointTime :: Double, -- ^ the current time pointIteration :: Int, -- ^ the current iteration pointPhase :: Int -- ^ the current phase } -- | The 'Parameter' monad that allows specifying the model parameters. -- For example, they can be used when running the Monte-Carlo simulation. -- -- In general, this monad is very useful for representing a computation which is external -- relative to the model itself. newtype Parameter m a = Parameter (Run m -> m a) -- | A value in the 'Simulation' monad represents a computation -- within the simulation run. newtype Simulation m a = Simulation (Run m -> m a) -- | A value in the 'Dynamics' monad represents a polymorphic time varying function -- defined in the whole spectrum of time values as a single entity. It is ideal for -- numerical approximating integrals. newtype Dynamics m a = Dynamics (Point m -> m a) -- | A value in the 'Event' monad transformer represents a polymorphic time varying -- function which is strongly synchronized with the event queue. newtype Event m a = Event (Point m -> m a) -- | Invoke the 'Parameter' computation. invokeParameter :: Run m -> Parameter m a -> m a {-# INLINE invokeParameter #-} invokeParameter r (Parameter m) = m r -- | Invoke the 'Simulation' computation. invokeSimulation :: Run m -> Simulation m a -> m a {-# INLINE invokeSimulation #-} invokeSimulation r (Simulation m) = m r -- | Invoke the 'Dynamics' computation. invokeDynamics :: Point m -> Dynamics m a -> m a {-# INLINE invokeDynamics #-} invokeDynamics p (Dynamics m) = m p -- | Invoke the 'Event' computation. invokeEvent :: Point m -> Event m a -> m a {-# INLINE invokeEvent #-} invokeEvent p (Event m) = m p -- | Defines how the events are processed. data EventProcessing = CurrentEvents -- ^ either process all earlier and then current events, -- or raise an error if the current simulation time is less -- than the actual time of the event queue (safe within -- the 'Event' computation as this is protected by the type system) | EarlierEvents -- ^ either process all earlier events not affecting -- the events at the current simulation time, -- or raise an error if the current simulation time is less -- than the actual time of the event queue (safe within -- the 'Event' computation as this is protected by the type system) | CurrentEventsOrFromPast -- ^ either process all earlier and then current events, -- or do nothing if the current simulation time is less -- than the actual time of the event queue -- (do not use unless the documentation states the opposite) | EarlierEventsOrFromPast -- ^ either process all earlier events, -- or do nothing if the current simulation time is less -- than the actual time of the event queue -- (do not use unless the documentation states the opposite) deriving (Eq, Ord, Show) -- | A type class of monads that allow enqueueing the events. class EventQueueing m where -- | It represents the event queue. data EventQueue m :: * -- | Create a new event queue by the specified specs with simulation session. newEventQueue :: Specs m -> m (EventQueue m) -- | Enqueue the event which must be actuated at the specified time. enqueueEvent :: Double -> Event m () -> Event m () -- | Run the 'EventT' computation in the current simulation time -- within the 'DynamicsT' computation involving all pending -- 'CurrentEvents' in the processing too. runEvent :: Event m a -> Dynamics m a {-# INLINE runEvent #-} runEvent = runEventWith CurrentEvents -- | Run the 'EventT' computation in the current simulation time -- within the 'DynamicsT' computation specifying what pending events -- should be involved in the processing. runEventWith :: EventProcessing -> Event m a -> Dynamics m a -- | Return the number of pending events that should -- be yet actuated. eventQueueCount :: Event m Int