public RRAStar(IGraph graph, AbstractTile start, AbstractTile target)
{
_graph = graph;
_start = start;
_open = new Priority_Queue.SimplePriorityQueue <AbstractTile, double>();
_gScore[target] = 0;
_open.Enqueue(target, Metrics.Octile(target, start));
Resume(start);
}
public static void Initialise()
{
pathfindQueue = new Priority_Queue.SimplePriorityQueue <PathfindTask>();
}
/// <summary>
/// constructor
/// </summary>
public Searcher()
{
openList = new Priority_Queue.SimplePriorityQueue <State <T> >();
evaluatedNodes = 0;
}
public static List <PathNode> GetLocalRoute(
PathNode target,
PathNode position,
MovementProperties movementProperties
)
{
Debug.Log("Начато построение пути");
// Вот тут вместо equals надо использовать == (как минимум), а лучше измерять расстояние между точками
// сравнивая с некоторым epsilon. Хотя может это для каких-то специальных случаев?
// if (position.Position.Equals(target.Position)) return new List<PathNode>();
if (Vector3.Distance(position.Position, target.Position) < movementProperties.epsilon)
{
return(new List <PathNode>());
}
//HashSet<(Vector3, Vector3)> closed = new HashSet<(Vector3, Vector3)>();
//HashSet<Vector3> closedDir = new HashSet<Vector3>();
Priority_Queue.SimplePriorityQueue <PathNode> opened = new Priority_Queue.SimplePriorityQueue <PathNode>();
// Тут тоже вопрос - а почему с 0 добавляем? Хотя она же сразу извлекается, не важно
opened.Enqueue(position, 0);
int steps = 0;
// Посещенные узлы (с некоторым шагом, аналог сетки)
HashSet <(int, int, int, int)> closed = new HashSet <(int, int, int, int)>();
closed.Add(position.ToGrid4DPoint(movementProperties.deltaDist, movementProperties.deltaTime));
PathNode last = opened.First;
while (opened.Count != 0 && steps < 5000)
{
steps++;
PathNode currentNode = opened.Dequeue();
last = currentNode;
//closed.Add(currentNode.ToGrid4DPoint(movementProperties.deltaDist, movementProperties.deltaTime)); // Для Sample-based список closed не нужен
// Тут что-то более сложное
if (currentNode.EqualsSigma(target, movementProperties.epsilon))
{
Debug.Log("Braked by closest point. Steps : " + steps.ToString());
break;
}
// Получаем список соседей
var neighbours = GetNeighbours(currentNode, movementProperties);
foreach (var nextNode in neighbours)
{
var discreteNode = nextNode.ToGrid4DPoint(movementProperties.deltaDist, movementProperties.deltaTime);
if (!closed.Contains(discreteNode))
{
nextNode.H = Heur(nextNode, target, movementProperties);
opened.Enqueue(nextNode, nextNode.H);
closed.Add(discreteNode);
}
}
}
if (last.EqualsSigma(target, movementProperties.epsilon) == false)
{
Debug.Log("Failed to build a way. Steps : " + steps.ToString());
return(new List <PathNode>());
}
List <PathNode> result = new List <PathNode>();
// Восстанавливаем путь от целевой к стартовой
// Может, заменить последнюю на целевую, с той же отметкой по времени? Но тогда с поворотом сложновато получается
var pathElem = last;
while (pathElem != null)
{
result.Add(pathElem);
pathElem = pathElem.Parent;
}
result.Reverse();
result.RemoveAt(0);
return(result);
}
private void NBS(object sender, EventArgs e)
{
totalDistance = 0;
Station[] ss = new Station[stations.Length];
stations.CopyTo(ss, 0);
Array.Sort(ss);
//找出不需要用送的站點作為補給站之用
List <Station> supplyStation = new List <Station>();
Priority_Queue.SimplePriorityQueue <Station> pq = new Priority_Queue.SimplePriorityQueue <Station>();
for (int i = 0; i < ss.Length; i++)
{
if (ss[i].Rate != 0)
{
pq.Enqueue(ss[i], 1);
}
if (ss[i].Rate == 0)
{
supplyStation.Add(ss[i]);
}
}
int[] Sequence = new int[totalStation];
for (int i = 0; i < Sequence.Length; i++)
{
Sequence[i] = -1;
}
//確定第一站為最需要先被服務的並且直接先決定卡車上的貨物數量
//直接先決定到的時間點剛好為最需要被服務的剛好缺的時點
int[] temp = new int[totalStation];
double time = 0;
double cost = 0;
Truck truck = null;
for (int k = ss.Length - 1; k > -1; k--)
{
if (ss[k].Rate != 0)
{
if (ss[k].Rate < 0) //rate < 0 => need to delivery
{
truck = new Truck(truckcapacity, truckcapacity);
temp[0] = truckcapacity / 2;
//cost = HoldingShortageSurplus(time, Sequence[0], truck.PickupDelivery(14));
}
else if (ss[k].Rate > 0)
{
truck = new Truck(truckcapacity, 0);
temp[0] = -truckcapacity / 2;
//cost = HoldingShortageSurplus(time, Sequence[0], truck.PickupDelivery(-14));
}
time = startTOvertex[ss[k].StationID] / speed;
Sequence[0] = ss[k].StationID;
temp[0] = truck.PickupDelivery(temp[0]);
cost += HoldingShortageSurplus(time, Sequence[0], temp[0], ref truck);
break;
}
}
Station s = null;
//開始建構解
int a = 1; bool flag = false;
double min = double.MaxValue; double m = 0; int selected = -1;
while (time <= timehorizon && a < totalStation)
{
while (pq.Count != 0)
{
s = pq.Max();
pq.Remove(s); flag = false;
//s = pq.Dequeue();
if (Sequence.Contains(s.StationID) == false)
{
Sequence[a] = s.StationID;
break;
}
if (pq.Count == 0)
{
if (supplyStation.Count != 0)
{
s = supplyStation.Max();
supplyStation.Remove(s);
}
break;
}
}
min = double.MaxValue;
//下一站是需要貨物的可是你卻沒有貨ㄌ
if (s.Declinebymin < 0)
{
//find truck lack of goods
if (truck.CurrentGoods == 0 && supplyStation.Count != 0)
{
min = double.MaxValue;
//find the nearest supply station
for (int i = 0; i < supplyStation.Count; i++)
{
m = distanceMatrix[Sequence[a - 1], supplyStation.ElementAt(i).StationID];
if (min > m)
{
min = m; selected = i;
}
}
temp[a] = truck.SupplyTruck(supplyStation.ElementAt(selected));
flag = true;
Sequence[a] = supplyStation.ElementAt(selected).StationID;
supplyStation.RemoveAt(selected);
if (a != stations.Length - 1)
{
Sequence[a + 1] = s.StationID;
}
}
if (flag == true)
{
time += distanceMatrix[Sequence[a - 1], Sequence[a]] / speed;//到下一站的時間點
totalDistance += distanceMatrix[Sequence[a - 1], Sequence[a]];
if (a != stations.Length - 1)
{
time += distanceMatrix[Sequence[a], Sequence[a + 1]] / speed;//到下一站的時間點
totalDistance += distanceMatrix[Sequence[a], Sequence[a + 1]];
temp[a + 1] = truck.PickupDelivery(truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a + 1], temp[a + 1], ref truck);
}
a += 2;
}
else
{
temp[a] = truck.PickupDelivery(truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a], temp[a], ref truck);
}
}
else
{
//find truck full of goods
if (truck.CurrentSpace == 0 && supplyStation.Count != 0)
{
//find the nearest supply station
min = double.MaxValue;
for (int i = 0; i < supplyStation.Count; i++)
{
m = distanceMatrix[Sequence[a - 1], supplyStation.ElementAt(i).StationID];
if (min > m)
{
min = m; selected = i;
}
}
temp[a] = truck.SupplyTruck(supplyStation.ElementAt(selected));
flag = true;
Sequence[a] = supplyStation.ElementAt(selected).StationID;
supplyStation.RemoveAt(selected);
if (a != stations.Length - 1)
{
Sequence[a + 1] = s.StationID;
}
}
if (flag == true)
{
time += distanceMatrix[Sequence[a - 1], Sequence[a]] / speed;//到下一站的時間點
totalDistance += distanceMatrix[Sequence[a - 1], Sequence[a]];
if (a != stations.Length - 1)
{
time += distanceMatrix[Sequence[a], Sequence[a + 1]] / speed;//到下一站的時間點
totalDistance += distanceMatrix[Sequence[a], Sequence[a + 1]];
temp[a + 1] = truck.PickupDelivery(-truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a + 1], temp[a + 1], ref truck);
}
a += 2;
}
else
{
temp[a] = truck.PickupDelivery(-truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a], temp[a], ref truck);
}
}
if (flag == false)
{
time += distanceMatrix[Sequence[a - 1], Sequence[a]] / speed;//到下一站的時間點
totalDistance += distanceMatrix[Sequence[a - 1], Sequence[a]];
a++;
}
}
if (a >= totalStation)
{
a = totalStation - 1;
}
//因為上面最後一次部會做所以...
if (flag == false && stations[Sequence[a - 1]].Declinebymin < 0 && time <= timehorizon)
{
temp[a - 1] = truck.PickupDelivery(truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a - 1], temp[a - 1], ref truck);
}
else if (flag == false && stations[Sequence[a - 1]].Surplusbymin > 0 && time <= timehorizon)
{
temp[a - 1] = truck.PickupDelivery(-truckcapacity / 2);
cost += HoldingShortageSurplus(time, Sequence[a - 1], temp[a - 1], ref truck);
}
else
{
a--;
}
//全部跑完要回去depot
if (a == totalStation - 1)
{
totalDistance += startTOvertex[Sequence[a]];
}
else
{
//沒有跑完的車站繼續算成本
for (int b = 0; b < pq.Count; b++)
{
cost += HoldingShortageSurplus(timehorizon, pq.ElementAt(b).StationID, 0, ref truck);
}
}
richTextBox.Text = "routing path is depot ";
for (int i = 0; i < a + 1; i++)
{
richTextBox.Text += (Sequence[i] + " ");
}
double totalPandD = 0;
richTextBox.Text += "\n" + "pickup & delivery amount: ";
for (int i = 0; i < temp.Length; i++)
{
richTextBox.Text += temp[i].ToString() + " ";
totalPandD += Math.Abs(temp[i]);
}
richTextBox.Text += "\n" + "********performance********";
richTextBox.Text += "\n" + "unsatisfied amount :" + cost;
richTextBox.Text += ("\n" + "total distance is " + totalDistance);
richTextBox.Text += "\n" + "Total pickup & delivery : " + totalPandD;
}
/// <summary>
/// Finds the least cost path from a given road pixel (that has to be connected to an exit point) to the nearest exit point pixel using roads,
/// and tries to upgrade the road pixels of this path to the given road type ID. This function is used if there will be another harvest in a given
/// harvested cell later, and if the module has decided to construct a sturdy road that will be there when the harvest will be done a second time.
/// If that's the case, this function will make sure that there will be a path from the harvested cell to an exit point that will last until then.
/// </summary>
/// /// <param name="ModelCore">
/// The model's core framework.
/// </param>
/// /// <param name="startingSite">
/// The starting site of the search. Has to be a road pixel connected to an exit point.
/// </param>
/// /// <param name="roadTypeIdToUpgradeTo">
/// True if this is done for the initialisation of the module; false if not.
/// </param>
public static void DijkstraLeastCostPathUpgradeRoadForRepeatedEntry(ICore ModelCore, Site startingSite, int roadTypeIdToUpgradeTo)
{
// We initialize the frontier and everything else
Priority_Queue.SimplePriorityQueue <Site> frontier = new Priority_Queue.SimplePriorityQueue <Site>();
Dictionary <Site, Site> predecessors = new Dictionary <Site, Site>();
Dictionary <Site, double> costSoFar = new Dictionary <Site, Double>();
HashSet <Site> isClosed = new HashSet <Site>();
bool haveWeFoundAnExitPoint = false;
costSoFar[startingSite] = 0;
frontier.Enqueue(startingSite, 0);
Site siteToClose;
double newDistanceToStart;
// Useless assignement to please the gods of C#
Site arrivalSite = startingSite;
// We loop until the list is empty
while (frontier.Count > 0)
{
siteToClose = frontier.Dequeue();
// We look at each of its neighbours, road on them or not.
foreach (Site neighbourToOpen in MapManager.GetNeighbouringSitesWithRoads(siteToClose))
{
// We don't consider the neighbour if it is closed or if it's non-constructible.
if ((SiteVars.CostRasterWithRoads[neighbourToOpen] >= 0) && (!isClosed.Contains(neighbourToOpen)))
{
// We get the value of the distance to start by using the current node to close, which is just an addition of the distance to the start
// from the node to close + 1 since we are only considering existing roads
newDistanceToStart = costSoFar[siteToClose] + 1;
// If the node isn't opened yet, or if it is opened and going to start throught the current node to close is closer; then,
// this node to close will become its predecessor, and its distance to start will become this one.
if (!costSoFar.ContainsKey(neighbourToOpen) || newDistanceToStart < costSoFar[neighbourToOpen])
{
costSoFar[neighbourToOpen] = newDistanceToStart;
predecessors[neighbourToOpen] = siteToClose;
// Case of the node not being opened
if (!frontier.Contains(neighbourToOpen))
{
frontier.Enqueue(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
// Case of the node being already open (if it is already in the frontier, we have to update)
else
{
frontier.UpdatePriority(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
}
// We check if the neighbour is a node we want to find, meaning a node with a place where the wood can flow; or, a road that
// is connected to such a place. If so, we can stop the search.
if (SiteVars.RoadsInLandscape[neighbourToOpen].IsAPlaceForTheWoodToGo)
{
arrivalSite = neighbourToOpen; haveWeFoundAnExitPoint = true; goto End;
}
}
}
isClosed.Add(siteToClose);
}
End:
// ModelCore.UI.WriteLine("Dijkstra search is over.");
// If we're out of the loop, that means that the search is over. If it was successfull before we ran out of neighbours to check,
// We can now retrieve the list of the sites that
// are the least cost path, and make sure that all of these site now have a road constructed on them, and that it is
// indicated as connected to a place where we can make wood go.
if (haveWeFoundAnExitPoint)
{
List <Site> listOfSitesInLeastCostPath = MapManager.FindPathToStart(startingSite, arrivalSite, predecessors);
// We will now try to update the roads in the path.
double costOfUpgrades = 0;
// We don't take the last site of the path, as it will be an exit point, and cannot be upgraded.
for (int i = 0; i < listOfSitesInLeastCostPath.Count - 1; i++)
{
// We check if the type we want to upgrade it too is of higher rank than the road type on the pixel
if (PlugIn.Parameters.RoadCatalogueNonExit.IsRoadTypeOfHigherRank(roadTypeIdToUpgradeTo, SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].typeNumber))
{
// If so, we upgrade it !
int oldTypeNumber = SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].typeNumber;
costOfUpgrades += SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].ComputeUpdateCost(listOfSitesInLeastCostPath[i], oldTypeNumber, roadTypeIdToUpgradeTo);
SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].typeNumber = roadTypeIdToUpgradeTo;
}
}
// We register the informations relative to the arrival site and the path in the RoadNetwork static objects
RoadNetwork.costOfConstructionAndRepairsAtTimestep += costOfUpgrades;
}
else
{
throw new Exception("FOREST ROADS SIMULATION ERROR : A Dijkstra search wasn't able to connect the site " + startingSite.Location + " to any site. This isn't supposed to happen." + PlugIn.errorToGithub);
}
}
/// <summary>
/// Try to make two least cost path to two sites; the second path is only constructed if the site reached is far enough, and if the cost of the path isn't too high.
/// </summary>
/// /// <param name="ModelCore">
/// The model's core framework.
/// </param>
/// /// <param name="startingSite">
/// The starting site of the search.
/// </param>
/// /// <param name="loopingDistance">
/// The minimal distance that the second site must be from the first one.
/// </param>
/// /// <param name="loopingMaxCost">
/// The maximal cost that the second road must cost.
/// </param>
public static int DijkstraLeastCostPathWithLooping(ICore ModelCore, Site startingSite, List <RelativeLocation> searchNeighborhood, double loopingMaxCost)
{
// We initialize the frontier and everything else
Priority_Queue.SimplePriorityQueue <Site> frontier = new Priority_Queue.SimplePriorityQueue <Site>();
Dictionary <Site, Site> predecessors = new Dictionary <Site, Site>();
Dictionary <Site, double> costSoFar = new Dictionary <Site, Double>();
HashSet <Site> isClosed = new HashSet <Site>();
bool IsFirstSiteReached = false;
bool IsSecondSiteReached = false;
List <Site> forbiddenSites = new List <Site>();
costSoFar[startingSite] = 0;
frontier.Enqueue(startingSite, 0);
Site siteToClose;
double newDistanceToStart;
// Useless assignment to please the gods of C#
Site firstSiteReached = startingSite;
Site secondSiteReached = startingSite;
// We loop until the list is empty
while (frontier.Count > 0)
{
siteToClose = frontier.Dequeue();
// We look at each of its neighbours, road on them or not.
foreach (Site neighbourToOpen in MapManager.GetNeighbouringSites(siteToClose))
{
// We don't consider the neighbour if it is closed or if it's non-constructible, or if it's the forbiden list of sites for making a proper loop.
if ((SiteVars.CostRasterWithRoads[neighbourToOpen] >= 0) && (!isClosed.Contains(neighbourToOpen) && (!forbiddenSites.Contains(neighbourToOpen))))
{
// We get the value of the distance to start by using the current node to close, which is just an addition of the distance to the start
// from the node to close + the cost between it and the neighbor.
newDistanceToStart = costSoFar[siteToClose] + MapManager.CostOfTransition(siteToClose, neighbourToOpen);
// If the node isn't opened yet, or if it is opened and going to start throught the current node to close is closer; then,
// this node to close will become its predecessor, and its distance to start will become this one.
if (!costSoFar.ContainsKey(neighbourToOpen) || newDistanceToStart < costSoFar[neighbourToOpen])
{
costSoFar[neighbourToOpen] = newDistanceToStart;
predecessors[neighbourToOpen] = siteToClose;
// Case of the node not being opened
if (!frontier.Contains(neighbourToOpen))
{
frontier.Enqueue(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
// Case of the node being already open (if it is already in the frontier, we have to update)
else
{
frontier.UpdatePriority(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
}
// We check if the neighbour is a node we want to find, meaning a node with a place where the wood can flow; or, a road that
// is connected to such a place. If so, we can stop the search.
// If a first site is reached, we register it, we create the list of forbiden sites not to reach or usen as path, and we remove those sites from the frontier.
if (!IsFirstSiteReached && SiteVars.RoadsInLandscape[neighbourToOpen].isConnectedToSawMill)
{
IsFirstSiteReached = true; firstSiteReached = neighbourToOpen; forbiddenSites = MapManager.GetNearbySites(searchNeighborhood, firstSiteReached); foreach (Site road in forbiddenSites)
{
frontier.TryRemove(road);
}
}
// If a second site is reached, we register it, and we end the search.
if (IsFirstSiteReached)
{
if (SiteVars.RoadsInLandscape[neighbourToOpen].isConnectedToSawMill)
{
if (firstSiteReached != neighbourToOpen)
{
IsSecondSiteReached = true; secondSiteReached = neighbourToOpen; goto End;
}
}
}
}
}
isClosed.Add(siteToClose);
}
End:
// We start by computing the cost of the first road, and constructing it
if (IsFirstSiteReached)
{
List <Site> listOfSitesInFirstLeastCostPath = MapManager.FindPathToStart(startingSite, firstSiteReached, predecessors);
double costOfFirstPath = 0;
int yearsBeforeReturn = MapManager.GetTimeBeforeNextHarvest(ModelCore, startingSite);
int IDOfRoadToConstruct = PlugIn.Parameters.RoadCatalogueNonExit.GetIDofPotentialRoadForRepeatedEntry(yearsBeforeReturn);
for (int i = 0; i < listOfSitesInFirstLeastCostPath.Count; i++)
{
// If there is no road on this site, we construct it.
if (!SiteVars.RoadsInLandscape[listOfSitesInFirstLeastCostPath[i]].IsARoad)
{
SiteVars.RoadsInLandscape[listOfSitesInFirstLeastCostPath[i]].typeNumber = IDOfRoadToConstruct;
}
// Whatever it is, we indicate it as connected.
SiteVars.RoadsInLandscape[listOfSitesInFirstLeastCostPath[i]].isConnectedToSawMill = true;
// We update the cost raster that contains the roads.
SiteVars.CostRasterWithRoads[listOfSitesInFirstLeastCostPath[i]] = 0;
// We also add the cost of transition to the costs of construction and repair for this timestep : it's the cost of transition multiplied by the type of the road that we are constructing. If there are already roads of other types on these cells, it doesn't change anything, as the value in the cost raster is 0 for them.
if (i < listOfSitesInFirstLeastCostPath.Count - 1)
{
costOfFirstPath += MapManager.CostOfTransition(listOfSitesInFirstLeastCostPath[i], listOfSitesInFirstLeastCostPath[i + 1]) * PlugIn.Parameters.RoadCatalogueNonExit.GetCorrespondingMultiplicativeCostValue(IDOfRoadToConstruct);
}
}
// We register the informations relative to the arrival site and the path in the RoadNetwork static objects
RoadNetwork.lastArrivalSiteOfDijkstraSearch = firstSiteReached;
RoadNetwork.costOfLastPath = costOfFirstPath;
RoadNetwork.costOfConstructionAndRepairsAtTimestep += costOfFirstPath;
// Finally, we upgrade the rest of the way towards an exit point if needed (if we constructed something else than the lowest type of roads)
if (IDOfRoadToConstruct != PlugIn.Parameters.RoadCatalogueNonExit.GetIDofLowestRoadType())
{
DijkstraLeastCostPathUpgradeRoadForRepeatedEntry(ModelCore, firstSiteReached, IDOfRoadToConstruct);
}
// Now, if a second site was reached, we check how much it cost. Of it's not too costly AND a probabilities are OK (see probability of loop construction parameter), we build it.
if (IsSecondSiteReached)
{
List <Site> listOfSitesInSecondLeastCostPath = MapManager.FindPathToStart(startingSite, secondSiteReached, predecessors);
double costOfSecondPath = 0;
for (int i = 0; i < listOfSitesInSecondLeastCostPath.Count; i++)
{
if (i < listOfSitesInSecondLeastCostPath.Count - 1)
{
costOfSecondPath += MapManager.CostOfTransition(listOfSitesInSecondLeastCostPath[i], listOfSitesInSecondLeastCostPath[i + 1]) * PlugIn.Parameters.RoadCatalogueNonExit.GetCorrespondingMultiplicativeCostValue(IDOfRoadToConstruct);
}
}
// If this second least cost path is not too costly AND a probabilities are OK (see probability of loop construction parameter), we build it., then we'll build it too.
// The random number will be between 1 and 100, and it must be inferior to 100 - the probability parameter. This way, the higher the probability parameter,
// the higher the chance that the random number will be above the threshold.
// We use the random number generator from the LANDIS-II core, which implies that results will always be the same as long as the random number seed given with the scenario is the same.
if ((costOfSecondPath / loopingMaxCost) < costOfFirstPath && (PlugIn.ModelCore.GenerateUniform() * 100) > (100 - PlugIn.Parameters.LoopingProbability))
{
for (int i = 0; i < listOfSitesInSecondLeastCostPath.Count; i++)
{
// If there is no road on this site, we construct it.
if (!SiteVars.RoadsInLandscape[listOfSitesInSecondLeastCostPath[i]].IsARoad)
{
SiteVars.RoadsInLandscape[listOfSitesInSecondLeastCostPath[i]].typeNumber = IDOfRoadToConstruct;
}
// Whatever it is, we indicate it as connected.
SiteVars.RoadsInLandscape[listOfSitesInSecondLeastCostPath[i]].isConnectedToSawMill = true;
// We update the cost raster that contains the roads.
SiteVars.CostRasterWithRoads[listOfSitesInSecondLeastCostPath[i]] = 0;
}
// We register the informations relative to the arrival site and the path in the RoadNetwork static objects
RoadNetwork.lastArrivalSiteOfDijkstraSearch = secondSiteReached;
RoadNetwork.costOfLastPath = costOfSecondPath;
RoadNetwork.costOfConstructionAndRepairsAtTimestep += costOfSecondPath;
// Finally, we upgrade the rest of the way towards an exit point if needed (if we constructed something else than the lowest type of roads)
if (IDOfRoadToConstruct != PlugIn.Parameters.RoadCatalogueNonExit.GetIDofLowestRoadType())
{
DijkstraLeastCostPathUpgradeRoadForRepeatedEntry(ModelCore, secondSiteReached, IDOfRoadToConstruct);
}
// If both roads have been constructed, we return that it's the case
return(2);
}
}
// If only one road has been constructed, we return that that's the case.
return(1);
}
else
{
throw new Exception("FOREST ROADS SIMULATION ERROR : A Dijkstra search wasn't able to connect the site " + startingSite.Location + " to any site. This isn't supposed to happen. Check if there are exit points in your landscape, and if they are reachable by the pathfinding algorithm (e.g. not surrounded by areas we roads can't be built)." + PlugIn.errorToGithub);
}
}
/// <summary>
/// Finds the least cost path from a given site to a site with a road already connected to somewhere where the wood can flow.
/// Different from DijkstraSearchForPlaceToPutWood in the sense that we do not only look for sites with paths, but all sites (even with no paths),
/// and that we build new roads where there was no path.
/// </summary>
/// /// <param name="ModelCore">
/// The model's core framework.
/// </param>
/// /// <param name="startingSite">
/// The starting site of the search.
/// </param>
/// /// <param name="initialisation">
/// True if this is done for the initialisation of the module; false if not.
/// </param>
public static void DijkstraLeastCostPathToClosestConnectedRoad(ICore ModelCore, Site startingSite, bool initialisation = false)
{
// We initialize the frontier of the algorithm and everything else
Priority_Queue.SimplePriorityQueue <Site> frontier = new Priority_Queue.SimplePriorityQueue <Site>();
Dictionary <Site, Site> predecessors = new Dictionary <Site, Site>();
Dictionary <Site, double> costSoFar = new Dictionary <Site, Double>();
HashSet <Site> isClosed = new HashSet <Site>();
bool haveWeFoundARoadToConnectTo = false;
costSoFar[startingSite] = 0;
frontier.Enqueue(startingSite, 0);
Site siteToClose;
double newDistanceToStart;
// Useless assignement to please the gods of C#
Site arrivalSite = startingSite;
// We loop until the list is empty (all cells of the landscape have been considered), or until we find an arrival cell
while (frontier.Count > 0)
{
siteToClose = frontier.Dequeue();
// We look at each of its neighbours, road on them or not.
foreach (Site neighbourToOpen in MapManager.GetNeighbouringSites(siteToClose))
{
// We don't consider the neighbour if it is closed or if it's non-constructible.
if ((SiteVars.CostRasterWithRoads[neighbourToOpen] >= 0) && (!isClosed.Contains(neighbourToOpen)))
{
// We get the value of the distance to start by using the current node to close, which is just an addition of the distance to the start
// from the node to close + the cost between it and the neighbor.
newDistanceToStart = costSoFar[siteToClose] + MapManager.CostOfTransition(siteToClose, neighbourToOpen);
// If the node isn't opened yet, or if it is opened and going to start throught the current node to close is closer; then,
// this node to close will become its predecessor, and its distance to start will become this one.
if (!costSoFar.ContainsKey(neighbourToOpen) || newDistanceToStart < costSoFar[neighbourToOpen])
{
costSoFar[neighbourToOpen] = newDistanceToStart;
predecessors[neighbourToOpen] = siteToClose;
// Case of the node not being opened
if (!frontier.Contains(neighbourToOpen))
{
frontier.Enqueue(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
// Case of the node being already open (if it is already in the frontier, we have to update)
else
{
frontier.UpdatePriority(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
}
// We check if the neighbour is a node we want to find, meaning a node with a place where the wood can flow; or, a road that
// is connected to such a place. If so, we can stop the search.
if (SiteVars.RoadsInLandscape[neighbourToOpen].isConnectedToSawMill)
{
arrivalSite = neighbourToOpen; haveWeFoundARoadToConnectTo = true; goto End;
}
}
}
isClosed.Add(siteToClose);
}
End:
// ModelCore.UI.WriteLine("Dijkstra search is over.");
// If we're out of the loop, that means that the search is over. If it was successfull before we ran out of neighbours to check,
// We can now retrieve the list of the sites that
// are the least cost path, and make sure that all of these site now have a road constructed on them, and that it is
// indicated as connected to a place where we can make wood go.
if (haveWeFoundARoadToConnectTo)
{
List <Site> listOfSitesInLeastCostPath = MapManager.FindPathToStart(startingSite, arrivalSite, predecessors);
double costOfPath = 0;
// If we are in the initiation phase, there is no need to check the strategy of the type of road we are going to construct.
if (initialisation)
{
for (int i = 0; i < listOfSitesInLeastCostPath.Count; i++)
{
// If there is no road on this site, we construct it. It'll be the smallest type of road.
if (!SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].IsARoad)
{
SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].typeNumber = PlugIn.Parameters.RoadCatalogueNonExit.GetIDofLowestRoadType();
}
// Whatever it is, we indicate it as connected.
SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].isConnectedToSawMill = true;
// We update the cost raster that contains the roads.
SiteVars.CostRasterWithRoads[listOfSitesInLeastCostPath[i]] = 0;
// We also add the cost of transition to the costs of construction and repair for this timestep : it's the cost of transition multiplied by the type of the road that we are constructing. If there are already roads of other types on these cells, it doesn't change anything, as the value in the cost raster is 0 for them.
if (i < listOfSitesInLeastCostPath.Count - 1)
{
costOfPath += MapManager.CostOfTransition(listOfSitesInLeastCostPath[i], listOfSitesInLeastCostPath[i + 1]) * PlugIn.Parameters.RoadCatalogueNonExit.GetCorrespondingMultiplicativeCostValue(PlugIn.Parameters.RoadCatalogueNonExit.GetIDofLowestRoadType());
}
}
}
// If we are building a road during the simulation, we are going to check wherever it's better to build a simple road or not.
else
{
// Before we decide what type of road to construct, we need information on the stand in which the starting site is. In particular,
// we want to know if there is going to be a repeated entry in this stand soon.
// To know that, we need to know which is the last presciption applied to this stand; if it's a multipleRepeat, we need to know the period.
// If not, it's a single repeat; if so, we need to know the time for which the stand is set aside.
int yearsBeforeReturn = MapManager.GetTimeBeforeNextHarvest(ModelCore, startingSite);
int IDOfRoadToConstruct = PlugIn.Parameters.RoadCatalogueNonExit.GetIDofPotentialRoadForRepeatedEntry(yearsBeforeReturn);
for (int i = 0; i < listOfSitesInLeastCostPath.Count; i++)
{
// If there is no road on this site, we construct it.
if (!SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].IsARoad)
{
SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].typeNumber = IDOfRoadToConstruct;
}
// Whatever it is, we indicate it as connected.
SiteVars.RoadsInLandscape[listOfSitesInLeastCostPath[i]].isConnectedToSawMill = true;
// We update the cost raster that contains the roads.
SiteVars.CostRasterWithRoads[listOfSitesInLeastCostPath[i]] = 0;
// We also add the cost of transition to the costs of construction and repair for this timestep : it's the cost of transition multiplied by the type of the road that we are constructing. If there are already roads of other types on these cells, it doesn't change anything, as the value in the cost raster is 0 for them.
if (i < listOfSitesInLeastCostPath.Count - 1)
{
costOfPath += MapManager.CostOfTransition(listOfSitesInLeastCostPath[i], listOfSitesInLeastCostPath[i + 1]) * PlugIn.Parameters.RoadCatalogueNonExit.GetCorrespondingMultiplicativeCostValue(IDOfRoadToConstruct);
}
}
// Finally, we upgrade the rest of the way towards an exit point if needed (if we constructed something else than the lowest type of roads)
if (IDOfRoadToConstruct != PlugIn.Parameters.RoadCatalogueNonExit.GetIDofLowestRoadType())
{
DijkstraLeastCostPathUpgradeRoadForRepeatedEntry(ModelCore, arrivalSite, IDOfRoadToConstruct);
}
}
// We register the informations relative to the arrival site and the path in the RoadNetwork static objects
RoadNetwork.lastArrivalSiteOfDijkstraSearch = arrivalSite;
RoadNetwork.costOfLastPath = costOfPath;
RoadNetwork.costOfConstructionAndRepairsAtTimestep += costOfPath;
}
else
{
throw new Exception("FOREST ROADS SIMULATION ERROR : A Dijkstra search wasn't able to connect the site " + startingSite.Location + " to any site. This isn't supposed to happen. Check if there are exit points in your landscape, and if they are reachable by the pathfinding algorithm (e.g. not surrounded by areas we roads can't be built)." + PlugIn.errorToGithub);
}
}
/// <summary>
/// Finds the least cost path from roads to roads to a exit point for the wood in the landscape, and add the given wood flux to every road visited.
/// Warning : The starting site must not be inside an existing fluxpath.
/// </summary>
/// /// <param name="ModelCore">
/// The model's core framework.
/// </param>
/// /// <param name="startingSite">
/// The starting site of the search.
/// </param>
/// /// <param name="Woodflux">
/// The flux of wood that is going to flow to the exit point.
/// </param>
public static void DijkstraWoodFlux(ICore ModelCore, Site startingSite, double woodFlux)
{
// We initialize the frontier and everything else
Priority_Queue.SimplePriorityQueue <Site> frontier = new Priority_Queue.SimplePriorityQueue <Site>();
Dictionary <Site, Site> predecessors = new Dictionary <Site, Site>();
Dictionary <Site, double> costSoFar = new Dictionary <Site, Double>();
HashSet <Site> isClosed = new HashSet <Site>();
bool haveWeFoundARoadToConnectTo = false;
costSoFar[startingSite] = 0;
frontier.Enqueue(startingSite, 0);
Site siteToClose;
double newDistanceToStart;
// Useless assignement to please the gods of C#
Site arrivalSite = startingSite;
// First, we got to check the possibility that the starting site IS the arrival site. This is possible in very rare situations,
// as the given starting site can be an exit point...That is not surounded by roads. If that happens, the dijkstra will not be able to
// open any neighbors, and will fail. Again, very rare, but still important.
if (RoadNetwork.fluxPathDictionary.ContainsKey(startingSite) || SiteVars.RoadsInLandscape[startingSite].IsAPlaceForTheWoodToGo)
{
haveWeFoundARoadToConnectTo = true; goto End;
}
// We loop until the list is empty
while (frontier.Count > 0)
{
// We take the site with the lowest distance to start.
siteToClose = frontier.Dequeue();
// We look at each of its neighbours, but only those with roads on them.
foreach (Site neighbourToOpen in MapManager.GetNeighbouringSitesWithRoads(siteToClose))
{
// We don't consider the neighbour if it is closed
if (!isClosed.Contains(neighbourToOpen))
{
// We get the value of the distance to start by using the current node to close, which is just an addition of the distance to the start
// from the node to close + the cost between it and the neighbor.
newDistanceToStart = costSoFar[siteToClose] + MapManager.CostOfTransition(siteToClose, neighbourToOpen);
// If the node isn't opened yet, or if it is opened and going to start throught the current node to close is closer; then,
// this node to close will become its predecessor, and its distance to start will become this one.
if (!costSoFar.ContainsKey(neighbourToOpen) || newDistanceToStart < costSoFar[neighbourToOpen])
{
costSoFar[neighbourToOpen] = newDistanceToStart;
predecessors[neighbourToOpen] = siteToClose;
// Case of the node not being opened
if (!frontier.Contains(neighbourToOpen))
{
frontier.Enqueue(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
// Case of the node being already open
else
{
frontier.UpdatePriority(neighbourToOpen, (float)costSoFar[neighbourToOpen]);
}
}
// We check if the neighbour is a exit node for the wood. If that's the case, search is other.
if (RoadNetwork.fluxPathDictionary.ContainsKey(neighbourToOpen) || SiteVars.RoadsInLandscape[neighbourToOpen].IsAPlaceForTheWoodToGo)
{
arrivalSite = neighbourToOpen; haveWeFoundARoadToConnectTo = true; goto End;
}
}
}
// Now that we have checked all of its neighbours, we can close the current node.
isClosed.Add(siteToClose);
}
End:
// ModelCore.UI.WriteLine("Dijkstra search for wood flux is over.");
// If we're out of the loop, that means that the search is over. If it was successfull before we ran out of neighbours to check,
// We can now retrieve the list of the sites that
// are the least cost path, and make sure that all of these site now have a road constructed on them, and that it is
// indicated as connected to a place where we can make wood go.
if (haveWeFoundARoadToConnectTo)
{
// If we found a fluxPath to connect to, we create a new one starting from the starting site, and stopping just before the arrival (which is our connection site to the fluxpath, and thus already belongs to a fluxpath)
if (RoadNetwork.fluxPathDictionary.ContainsKey(arrivalSite))
{
List <Site> listOfSitesInLeastCostPath = MapManager.FindPathToStart(startingSite, arrivalSite, predecessors);
// We have to reverse the list, because we want to go from the harvested zones to the connection point, and not the opposite.
listOfSitesInLeastCostPath.Reverse();
// We don't put the arrival site in the fluxpath, because it is part of another fluxpath.
listOfSitesInLeastCostPath.Remove(arrivalSite);
FluxPath newFluxPath = new FluxPath(listOfSitesInLeastCostPath, arrivalSite);
// Now that this new path is created, we flux the wood.
newFluxPath.FluxPathFromSite(startingSite, woodFlux);
}
// Else, if we didn't found a fluxPath to connect to but an exit point for the wood, we create a new "isAnEnd" path.
else
{
List <Site> listOfSitesInLeastCostPath = MapManager.FindPathToStart(startingSite, arrivalSite, predecessors);
// We have to reverse the list, because we want to go from the harvested zones to the exit point, and not the opposite.
listOfSitesInLeastCostPath.Reverse();
FluxPath newFluxPathIsAnEnd = new FluxPath(listOfSitesInLeastCostPath);
// Then, we flux the wood down this path.
newFluxPathIsAnEnd.FluxPathFromSite(startingSite, woodFlux);
}
}
else
{
throw new Exception("FOREST ROADS SIMULATION ERROR : A Dijkstra search wasn't able to flux the wood from site " + startingSite.Location + " to any exit point. This isn't supposed to happen. Please check the output raster containg the road network for the current timestep (" + ModelCore.CurrentTime + " years) and see if a road is interrupted somewhere." + PlugIn.errorToGithub);
}
}
