/*****
# A empirical distribution
# def transition_prob(u, v):
# c = 1 / BETA
# Calculating route distance is expensive.
# We will discuss how to reduce the number of calls to this function later.
# delta = math.abs(route_distance(u, v) - great_circle_distance(u.measurement, v.measurement))
# return c * math.exp(-delta)
*****/
private double TransitionProbability(HiddenMarkovState u, HiddenMarkovState v)
{
if (u == s)
{
return(1.0f);
}
else if (v == t)
{
return(1.0f);
}
Node Source = RoadNetwork.ResolvePoint(u.P, 0.1f).First();
Node Target = RoadNetwork.ResolvePoint(v.P, 0.1f).First();
Router router = new Router(Source, Target, string.Empty, 0, TravelMode.Car, ObjFunction.Distance);
router.Start();
double routeDistance = router.SolutionCost;//router.Solution.Where(x => x.Id != -1).Sum(x => x.GetDistance(TravelMode.Car));
//
double beta = 3;
double c = 1 / beta;
double delta = Math.Abs(routeDistance - u.P.DistanceFrom(v.P));
return(c * Math.Exp(-delta));
}
private void Init()
{
s = new HiddenMarkovState(null, null, 0.0f, true);
t = new HiddenMarkovState(null, null, 0.0f, true);
foreach (Point P in Points)
{
CreateState(P, Points.IndexOf(P));
}
LinkStates();
}
private double EdgeCost(HiddenMarkovState u, HiddenMarkovState v)
{
return(-1 * Math.Log10(TransitionProbability(u, v)));
}
public void AddToAdjList(HiddenMarkovState State)
{
AdjacencyList.Add(State);
}
private void RunDijkstra()
{
Satsuma.PriorityQueue <HiddenMarkovState, double> Q = new Satsuma.PriorityQueue <HiddenMarkovState, double>();
Dictionary <HiddenMarkovState, double> distance = new Dictionary <HiddenMarkovState, double>();
Dictionary <HiddenMarkovState, HiddenMarkovState> parent = new Dictionary <HiddenMarkovState, HiddenMarkovState>();
if (!parent.ContainsKey(s))
{
parent.Add(s, null);
}
else
{
parent[s] = null;
}
Q[s] = 0.0;
while (Q.Count != 0)
{
// find the closest reached but unfixed node
double minDist;
HiddenMarkovState min = Q.Peek(out minDist);
Q.Pop();
/// fix the node.
if (!distance.ContainsKey(min))
{
distance.Add(min, minDist);
}
else
{
distance[min] = minDist; // fix the node
}
if (min == t)
{
break; // target node found and fixed.
}
foreach (HiddenMarkovState v in min.AdjacencyList)
{
if (distance.ContainsKey(v))
{
continue; // already processed
}
double newDist = minDist + min.StateCost() + EdgeCost(min, v);
double oldDist;
if (!Q.TryGetPriority(v, out oldDist))
{
oldDist = double.PositiveInfinity;
}
if (newDist < oldDist)
{
Q[v] = newDist;
if (parent.ContainsKey(v))
{
parent[v] = min;
}
else
{
parent.Add(v, min);
}
}
}
}
// Construct Solution.
HiddenMarkovState S = parent[t];
while (S != null)
{
//if (S.GetUnderlyingRNode() != null);
MatchingSolution.AddFirst(S.GetUnderlyingRNode());
S = parent[S];
}
}