Function Nodes example in C#

// --------------------------------------------------------------------------------------------------------------------
// <copyright file="FunctionNodesExample.cs" company="Bayes Server">
//   Copyright (C) Bayes Server.  All rights reserved.
// </copyright>
// --------------------------------------------------------------------------------------------------------------------

namespace BayesServer.HelpSamples
{
    using System;
    using BayesServer.Inference;
    using BayesServer.Inference.RelevanceTree;

    public static class FunctionNodesExample
    {
        public static void Main()
        {
            // In this example we programatically create and query the sample
            // network 'Functions' included with the Bayes Server User Interface.

            var network = new Network("Functions");

            // add the variables/nodes

            var d1 = new State("D1");
            var d2 = new State("D2");
            var d = new Variable("D", d1, d2);
            d.ExpressionAlias = "d";
            network.Nodes.Add(new Node(d));

            var c = new Variable("C", VariableValueType.Continuous);
            c.ExpressionAlias= "c";
            network.Nodes.Add(new Node(c));

            var f1 = new Variable("F1", VariableValueType.Function);
            f1.ExpressionAlias= "f1";
            network.Nodes.Add(new Node(f1));

            var f2 = new Variable("F2", VariableValueType.Function);
            network.Nodes.Add(new Node(f2));

            var f3 = new Variable("F3", VariableValueType.Function);
            network.Nodes.Add(new Node(f3));

            // add some directed links

            network.Links.Add(new Link(d.Node, c.Node));
            network.Links.Add(new Link(f1.Node, f2.Node));  // the expression in f2 uses the result from f1, so we need a link here

            var tableD = d.Node.NewDistribution().Table;
           
            tableD[d1] = 0.6961106280502056;
            tableD[d2] = 0.3038893719497943;

            d.Node.Distribution = tableD;

            var gaussianC = (CLGaussian)c.Node.NewDistribution();
            gaussianC.SetMean(c, 100.0, d1);
            gaussianC.SetVariance(c, 20.0, d1);
            gaussianC.SetMean(c, -50.0, d2);
            gaussianC.SetVariance(c, 45.0, d2);

            c.Node.Distribution = gaussianC;

            f1.Function = new FunctionVariableExpression(
                @"
                var d1 = d.States.Get(""D1"", true);
                var probD1 = ctx.TableValue(d1);

                var meanC1 = ctx.Mean(c);

                return probD1 * meanC1;
                ",
                ExpressionReturnType.Double);

            f2.Function = new FunctionVariableExpression(
                @"
                var f1Val = ctx.FunctionValue(f1);

                return f1Val * 2.0;
                ",
                ExpressionReturnType.Double);
            
            f3.Function = new FunctionVariableExpression(
                @"
                return ""PI = "" + Math.PI;
                ",
                ExpressionReturnType.String);


            // The network is now fully specified

            // Now we will query the nodes

            var factory = new RelevanceTreeInferenceFactory();
            var inference = factory.CreateInferenceEngine(network);
            var queryOptions = factory.CreateQueryOptions();
            var queryOutput = factory.CreateQueryOutput();

            // set any evidence, e.g.
            // inference.Evidence.SetState(d1);  // set D = D1

            var queryD = new Table(d);
            inference.QueryDistributions.Add(queryD);

            var queryC = new CLGaussian(c);
            inference.QueryDistributions.Add(queryC);

            var queryF1 = new QueryFunctionOutput(f1);
            inference.QueryFunctions.Add(queryF1);

            var queryF2 = new QueryFunctionOutput(f2);
            inference.QueryFunctions.Add(queryF2);

            var queryF3 = new QueryFunctionOutput(f3);
            inference.QueryFunctions.Add(queryF3);

            inference.Query(queryOptions, queryOutput); // note that this can raise an exception (see help for details)

            Console.WriteLine("P(D|-) = { " + queryD[d1] + "," + queryD[d2] + " }.");
            Console.WriteLine("P(C|-) = { " + queryC.GetMean(c) + "," + queryC.GetVariance(c) + " }.");

            Console.WriteLine("F1 = { " + queryF1.Value + " }.");
            Console.WriteLine("F2 = { " + queryF2.Value + " }.");
            Console.WriteLine("F3 = { " + queryF3.Value + " }.");

            // expected output...

            // P(D | -) = { 0.6961106280502056,0.3038893719497943 }.
            // P(C | -) = { 54.41659420753085,4787.261219528207 }.
            // F1 = { 37.87996957015748 }.
            // F2 = { 75.75993914031496 }.
            // F3 = { PI = 3.141592653589793 }.

        }
    }
}