rhine-bayes 0.9 → 1.0

raw patch · 4 files changed

+36/−20 lines, 4 filesdep ~rhinePVP ok

version bump matches the API change (PVP)

Dependency ranges changed: rhine

API changes (from Hackage documentation)

Files

• ChangeLog.md +6/−0
• app/Main.hs +10/−10
• rhine-bayes.cabal +3/−3
• src/FRP/Rhine/Bayes.hs +17/−7

@@ -1,5 +1,11 @@ # Revision history for rhine-gloss +## 1.0++* Removed schedules. See the [page about changes in version 1](/version1.md).+* Introduced type alias `StochasticProcess`.+* Added `whiteNoiseVarying`.+ ## 0.9  * Add simple Poisson, Gamma and Bernoulli processes

@@ -59,14 +59,14 @@  -- | Harmonic oscillator with white noise prior1d ::-  (MonadDistribution m, Diff td ~ Double) =>+  (Diff td ~ Double) =>   -- | Starting position   Double ->   -- | Starting velocity   Double ->-  BehaviourF m td Temperature Double+  StochasticProcessF td Temperature Double prior1d initialPosition initialVelocity = feedback 0 $ proc (temperature, position') -> do-  impulse <- arrM (normal 0) -< temperature+  impulse <- whiteNoiseVarying -< temperature   let acceleration = (-3) * position' + impulse   -- Integral over roughly the last 100 seconds, dying off exponentially, as to model a small friction term   velocity <- arr (+ initialVelocity) <<< decayIntegral 10 -< acceleration@@ -88,11 +88,11 @@ sensorNoiseTemperature = 1  -- | A generative model of the sensor noise-noise :: MonadDistribution m => Behaviour m td Pos+noise :: StochasticProcess td Pos noise = whiteNoise sensorNoiseTemperature &&& whiteNoise sensorNoiseTemperature  -- | A generative model of the sensor position, given the noise-generativeModel :: (MonadDistribution m, Diff td ~ Double) => BehaviourF m td Pos Sensor+generativeModel :: (Diff td ~ Double) => StochasticProcessF td Pos Sensor generativeModel = proc latent -> do   noiseNow <- noise -< ()   returnA -< latent ^+^ noiseNow@@ -286,7 +286,7 @@ mainSingleRate =   void $     sampleIO $-      launchGlossThread glossSettings $+      launchInGlossThread glossSettings $         reactimateCl glossClock mainClSF  -- ** Multi-rate: Simulation, inference, display at different rates@@ -337,18 +337,18 @@ mainRhineMultiRate =   userTemperature     @@ glossClockUTC GlossEventClockIO-      >-- keepLast initialTemperature -@- glossConcurrently -->+      >-- keepLast initialTemperature -->         modelRhine-        >-- keepLast (initialTemperature, (zeroVector, zeroVector)) -@- glossConcurrently -->+        >-- keepLast (initialTemperature, (zeroVector, zeroVector)) -->           inference-            >-- keepLast Result {temperature = initialTemperature, measured = zeroVector, latent = zeroVector, particles = []} -@- glossConcurrently -->+            >-- keepLast Result {temperature = initialTemperature, measured = zeroVector, latent = zeroVector, particles = []} -->               visualisationRhine {- FOURMOLU_ENABLE -}  mainMultiRate :: IO () mainMultiRate =   void $-    launchGlossThread glossSettings $+    launchInGlossThread glossSettings $       flow mainRhineMultiRate  -- * Utilities

@@ -1,5 +1,5 @@ name:                rhine-bayes-version:             0.9+version:             1.0 synopsis:            monad-bayes backend for Rhine description:   This package provides a backend to the `monad-bayes` library,@@ -23,7 +23,7 @@ source-repository this   type:     git   location: git@github.com:turion/rhine.git-  tag:      v0.9+  tag:      v1.0  library   exposed-modules:@@ -32,7 +32,7 @@     Data.MonadicStreamFunction.Bayes   build-depends:       base         >= 4.11 && < 4.18                      , transformers >= 0.5-                     , rhine        == 0.9+                     , rhine        == 1.0                      , dunai        ^>= 0.9                      , log-domain   >= 0.12                      , monad-bayes  >= 1.1.0

@@ -39,10 +39,20 @@  -- * Short standard library of stochastic processes +-- | A stochastic process is a behaviour that uses, as only effect, random sampling.+type StochasticProcess time a = forall m. MonadDistribution m => Behaviour m time a++-- | Like 'StochasticProcess', but with a live input.+type StochasticProcessF time a b = forall m. MonadDistribution m => BehaviourF m time a b+ -- | White noise, that is, an independent normal distribution at every time step.-whiteNoise :: MonadDistribution m => Double -> Behaviour m td Double+whiteNoise :: Double -> StochasticProcess td Double whiteNoise sigma = constMCl $ normal 0 sigma +-- | Like 'whiteNoise', that is, an independent normal distribution at every time step.+whiteNoiseVarying :: StochasticProcessF td Double Double+whiteNoiseVarying = arrMCl $ normal 0+ -- | Construct a Lévy process from the increment between time steps. levy ::   (MonadDistribution m, VectorSpace v (Diff td)) =>@@ -64,8 +74,8 @@ -- | The Wiener process, also known as Brownian motion, with varying variance parameter. wienerVarying   , brownianMotionVarying ::-    (MonadDistribution m, Diff td ~ Double) =>-    BehaviourF m td (Diff td) Double+    (Diff td ~ Double) =>+    StochasticProcessF td (Diff td) Double wienerVarying = proc timeScale -> do   diffTime <- sinceLastS -< ()   let stdDev = sqrt $ diffTime / timeScale@@ -78,16 +88,16 @@  -- | The 'wiener' process transformed to the Log domain, also called the geometric Wiener process. wienerLogDomain ::-  (MonadDistribution m, Diff td ~ Double) =>+  (Diff td ~ Double) =>   -- | Time scale of variance   Diff td ->-  Behaviour m td (Log Double)+  StochasticProcess td (Log Double) wienerLogDomain timescale = wiener timescale >>> arr Exp  -- | See 'wienerLogDomain' and 'wienerVarying'. wienerVaryingLogDomain ::-  (MonadDistribution m, Diff td ~ Double) =>-  BehaviourF m td (Diff td) (Log Double)+  (Diff td ~ Double) =>+  StochasticProcessF td (Diff td) (Log Double) wienerVaryingLogDomain = wienerVarying >>> arr Exp  {- | Inhomogeneous Poisson point process, as described in: