docs(examples): add example of using the Polynomial basis

examples/polynomial.rs

@@ -0,0 +1,49 @@
+extern crate rsrl;
+#[macro_use] extern crate slog;
+
+use rsrl::{run, logging, Parameter, SerialExperiment, Evaluation};
+use rsrl::agents::memory::Trace;
+use rsrl::agents::control::td::SARSALambda;
+use rsrl::domains::{Domain, MountainCar};
+use rsrl::fa::{Linear, Projector};
+use rsrl::fa::projection::Polynomial;
+use rsrl::geometry::Space;
+use rsrl::policies::EpsilonGreedy;
+
+
+fn main() {
+    let domain = MountainCar::default();
+    let mut agent = {
+        let n_actions = domain.action_space().span().into();
+
+        // Build the linear value function using a polynomial basis projection and the appropriate
+        // eligibility trace.
+        let bases = Polynomial::from_space(5, domain.state_space());
+        let trace = Trace::replacing(0.7, bases.activation());
+        let q_func = Linear::new(bases, n_actions);
+
+        // Build a stochastic behaviour policy with exponential epsilon.
+        let eps = Parameter::exponential(0.99, 0.05, 0.99);
+        let policy = EpsilonGreedy::new(eps);
+
+        SARSALambda::new(trace, q_func, policy, 0.1, 0.99)
+    };
+
+    let logger = logging::root(logging::stdout());
+    let domain_builder = Box::new(MountainCar::default);
+
+    // Training phase:
+    let _training_result = {
+        // Start a serial learning experiment up to 1000 steps per episode.
+        let e = SerialExperiment::new(&mut agent, domain_builder.clone(), 1000);
+
+        // Realise 1000 episodes of the experiment generator.
+        run(e, 1000, Some(logger.clone()))
+    };
+
+    // Testing phase:
+    let testing_result =
+        Evaluation::new(&mut agent, domain_builder).next().unwrap();
+
+    info!(logger, "solution"; testing_result);
+}