// Constructor
FloodFillFinder(bool shouldPause = false) : pause(shouldPause) {
coords.clearAll();
horizontal.clearAll();
vertical.clearAll();
heading = NORTH;
for(int i = MazeDefinitions::MAZE_LEN/2; i < MazeDefinitions::MAZE_LEN; i++){
for(int j = 0; j < MazeDefinitions::MAZE_LEN/2; j++)
Distance[i][j] = Manhattan(i, j, MazeDefinitions::MAZE_LEN/2, MazeDefinitions::MAZE_LEN/2 - 1);
}
for(int i = MazeDefinitions::MAZE_LEN/2; i < MazeDefinitions::MAZE_LEN; i++){
for(int j = MazeDefinitions::MAZE_LEN/2; j < MazeDefinitions::MAZE_LEN; j++)
Distance[i][j] = Manhattan(i, j, MazeDefinitions::MAZE_LEN/2, MazeDefinitions::MAZE_LEN/2);
}
for(int i = 0; i < MazeDefinitions::MAZE_LEN/2; i++){
for(int j = 0; j < MazeDefinitions::MAZE_LEN/2; j++)
Distance[i][j] = Manhattan(i, j, MazeDefinitions::MAZE_LEN/2 - 1, MazeDefinitions::MAZE_LEN/2 - 1);
}
for(int i = 0; i < MazeDefinitions::MAZE_LEN/2; i++){
for(int j = MazeDefinitions::MAZE_LEN/2; j < MazeDefinitions::MAZE_LEN; j++)
Distance[i][j] = Manhattan(i, j, MazeDefinitions::MAZE_LEN/2 - 1, MazeDefinitions::MAZE_LEN/2);
}
for(int i = 0; i < MazeDefinitions::MAZE_LEN; i++)
horizontal.set(0, i);
for(int i = 0; i < MazeDefinitions::MAZE_LEN; i++)
vertical.set(i, 0);
}
// (Basic heuristic) Returns the Manhattan distance between two points in 1D coordinates
int Manhattan(const int width, const int i, const int j)
{
int xStart;
int yStart;
int xTarget;
int yTarget;
OneDToTwoD(width, i, xStart, yStart);
OneDToTwoD(width, j, xTarget, yTarget);
return Manhattan(xStart, yStart, xTarget, yTarget);
}
// A* implementation
int FindPath(
const int nStartX,
const int nStartY,
const int nTargetX,
const int nTargetY,
const unsigned char* pMap,
const int nMapWidth,
const int nMapHeight,
int* pOutBuffer,
const int nOutBufferSize)
{
// Lock critical section (verrrry basic way)
simple_mutex.lock();
// If Start equals target path size is 0
if (nStartX == nTargetX && nStartY == nTargetY)
{
return 0;
}
// Define the open and the close list
std::vector<int> openList;
std::vector<int> closedList;
// Cost and heuristic buffers
const int mapSize = nMapWidth * nMapHeight;
int* costSoFar = new int[mapSize];
int* hEstimate = new int[mapSize];
int* camefrom = new int[mapSize];
// Initialize path history
for (int i = 0; i < mapSize; ++i)
{
camefrom[i] = INDEX_NULL;
}
int indexStart;
int indexEnd;
TwoDToOneD(nMapWidth, nStartX, nStartY, indexStart);
TwoDToOneD(nMapWidth, nTargetX, nTargetY, indexEnd);
costSoFar[indexStart] = 0;
hEstimate[indexStart] = Manhattan(nStartX, nStartY, nTargetX, nTargetY);
// Add the start point to the openList
openList.push_back(indexStart);
int current = openList[0];
bool success = false;
// Release the Kraken !
while (openList.size() > 0)
{
// Look for the smallest cost in the openList and make it the current point
int best = INTEGER_MAX;
const int openListSize = openList.size();
int i = 0;
for (; i < openListSize; ++i)
{
int cost = hEstimate[openList[i]];
if (cost < best)
{
current = openList[i];
best = cost;
}
}
// Remove the current point from the openList
openList.erase(openList.begin() + (i - 1));
// Add the current point to the closedList
closedList.push_back(current);
// Stop if we reached the end
if (current == indexEnd)
{
success = true;
break;
}
// Go over valid neighbors and get the most interesting
std::vector<int> neighbors;
NeighborsOneD(current, pMap, nMapWidth, nMapHeight, neighbors);
const int neighborsSize = neighbors.size();
i = 0;
for (; i < neighborsSize; ++i)
{
const int currentNeighbor = neighbors[i];
// If accessible and not already kicked out (could be done in Neighbors research function)
if ((int)pMap[currentNeighbor] == 1
&& std::find(closedList.begin(), closedList.end(), currentNeighbor) == closedList.end())
{
// Evaluate cost and add to the open list
const int cost = costSoFar[current] + Manhattan(nMapWidth, current, currentNeighbor);
if (std::find(openList.begin(), openList.end(), currentNeighbor) == openList.end())
{
openList.push_back(currentNeighbor);
}
// Else if already in open and is more expensive than before, forget it.
else if (cost >= costSoFar[currentNeighbor])
{
continue;
}
costSoFar[currentNeighbor] = cost;
hEstimate[currentNeighbor] = cost + Manhattan(nMapWidth, currentNeighbor, indexEnd);
camefrom[current] = currentNeighbor;
}
}
}
// If we reached target, fill out output buffer
int pathIndex = 0;
if (success)
{
for (int idx = 0; camefrom[idx] != INDEX_NULL; ++idx)
{
pOutBuffer[pathIndex] = camefrom[idx];
++pathIndex;
}
pOutBuffer[pathIndex] = current;
}
//Clean up dynamic alloc
delete hEstimate;
delete camefrom;
delete costSoFar;
//Unlock critical section
simple_mutex.unlock();
return success ? pathIndex + 1 : INDEX_NULL;
}
void
run_test(Cost_Map &map, Point &start, Point &goal, vector<int> ×) {
struct timeval pre;
struct timeval post;
/// printf("UCS\n");
UCS ucs(&map, start, goal);
gettimeofday(&pre, NULL);
Path path = ucs.search();
gettimeofday(&post, NULL);
double reference_cost = path.length;
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("A*\n");
Manhattan Manhattan(&map, start, goal);
gettimeofday(&pre, NULL);
path = Manhattan.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("A*\n");
Euclidean euclidean(&map, start, goal);
gettimeofday(&pre, NULL);
path = euclidean.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("A*\n");
Octile octile(&map, start, goal);
gettimeofday(&pre, NULL);
path = octile.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("Coarse single\n");
CUCS_Heuristic cucs(&map, start, goal);
gettimeofday(&pre, NULL);
path = cucs.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("BB\n");
Boundaries_Blocking bb2(&map, start, goal, 2, xscale, yscale);
gettimeofday(&pre, NULL);
path = bb2.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
Boundaries_Blocking bb(&map, start, goal, levels, xscale, yscale);
gettimeofday(&pre, NULL);
path = bb.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("BN\n");
Boundaries_NonBlocking bnb2(&map, start, goal, 2, xscale, yscale);
gettimeofday(&pre, NULL);
path = bnb2.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
Boundaries_NonBlocking bnb(&map, start, goal, levels, xscale, yscale);
gettimeofday(&pre, NULL);
path = bnb.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("CB\n");
Corners_Blocking cb2(&map, start, goal, 2, xscale, yscale);
gettimeofday(&pre, NULL);
path = cb2.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
Corners_Blocking cb(&map, start, goal, levels, xscale, yscale);
gettimeofday(&pre, NULL);
path = cb.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
/// printf("CN\n");
Corners_NonBlocking cnb2(&map, start, goal, 2, xscale, yscale);
gettimeofday(&pre, NULL);
path = cnb2.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
Corners_NonBlocking cnb(&map, start, goal, levels, xscale, yscale);
gettimeofday(&pre, NULL);
path = cnb.search();
gettimeofday(&post, NULL);
check_and_update(start, goal, pre, post, reference_cost, path.length, times);
}
bool UnitTests::WeightedDataMatrixMothur()
{
std::vector< std::vector<double> > dissMatrix;
// weighted Bray-Curtis
DiversityCalculator BC("../unit-tests/DataMatrixMothur.env", "", "Bray-Curtis", 1000, true, false, false, false, false);
BC.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.8))
return false;
if(!Compare(dissMatrix[2][0], 0.6))
return false;
if(!Compare(dissMatrix[2][1], 0.8))
return false;
// weighted Canberra
DiversityCalculator Canberra("../unit-tests/DataMatrixMothur.env", "", "Canberra", 1000, true, false, false, false, false);
Canberra.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 6.35152))
return false;
if(!Compare(dissMatrix[2][0], 8.11111))
return false;
if(!Compare(dissMatrix[2][1], 5.92063))
return false;
// weighted Gower
DiversityCalculator Gower("../unit-tests/DataMatrixMothur.env", "", "Gower", 1000, true, false, false, false, false);
Gower.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 6.58333))
return false;
if(!Compare(dissMatrix[2][0], 7.88889))
return false;
if(!Compare(dissMatrix[2][1], 5.52778))
return false;
// weighted Hellinger
DiversityCalculator Hellinger("../unit-tests/DataMatrixMothur.env", "", "Hellinger", 1000, true, false, false, false, false);
Hellinger.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 1.13904))
return false;
if(!Compare(dissMatrix[2][0], 1.05146))
return false;
if(!Compare(dissMatrix[2][1], 1.17079))
return false;
// weighted Manhattan
DiversityCalculator Manhattan("../unit-tests/DataMatrixMothur.env", "", "Manhattan", 1000, true, false, false, false, false);
Manhattan.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 1.6))
return false;
if(!Compare(dissMatrix[2][0], 1.2))
return false;
if(!Compare(dissMatrix[2][1], 1.6))
return false;
// weighted Morisita-Horn
DiversityCalculator MH("../unit-tests/DataMatrixMothur.env", "", "Morisita-Horn", 1000, true, false, false, false, false);
MH.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.873239))
return false;
if(!Compare(dissMatrix[2][0], 0.333333))
return false;
if(!Compare(dissMatrix[2][1], 0.859155))
return false;
// weighted Soergel
DiversityCalculator Soergel("../unit-tests/DataMatrixMothur.env", "", "Soergel", 1000, true, false, false, false, false);
Soergel.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.88888901))
return false;
if(!Compare(dissMatrix[2][0], 0.75))
return false;
if(!Compare(dissMatrix[2][1], 0.88888901))
return false;
// weighted species profile
/*
DiversityCalculator SP("../unit-tests/DataMatrixMothur.env", "", "Species profile", 1000, true, false, false, false, false);
SP.Dissimilarity("../unit-tests/temp.diss");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.78740102))
return false;
if(!Compare(dissMatrix[2][0], 0.44721401))
return false;
if(!Compare(dissMatrix[2][1], 0.78102499))
return false;
*/
// weighted Chi-squared
// Note: EBD uses a slightly different form of the Chi-squared measure as suggested in Numerical Ecology by Legendre adn Legendre
// Nonetheless, it is easy to verify this using mothur. Simply divide by sqrt(N), N is the total number of sequences.
DiversityCalculator ChiSquared("../unit-tests/DataMatrixMothur.env", "", "Chi-squared", 1000, true, false, false, false, false);
ChiSquared.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 1.0973200))
return false;
if(!Compare(dissMatrix[2][0], 0.96513098))
return false;
if(!Compare(dissMatrix[2][1], 1.1147900))
return false;
// weighted Euclidean
DiversityCalculator Euclidean("../unit-tests/DataMatrixMothur.env", "", "Euclidean", 1000, true, false, false, false, false);
Euclidean.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.78740102))
return false;
if(!Compare(dissMatrix[2][0], 0.44721401))
return false;
if(!Compare(dissMatrix[2][1], 0.78102499))
return false;
// weighted Kulczynski
DiversityCalculator Kulczynski("../unit-tests/DataMatrixMothur.env", "", "Kulczynski", 1000, true, false, false, false, false);
Kulczynski.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.8))
return false;
if(!Compare(dissMatrix[2][0], 0.6))
return false;
if(!Compare(dissMatrix[2][1], 0.8))
return false;
// weighted Pearson
DiversityCalculator uPearson("../unit-tests/DataMatrixMothur.env", "", "Pearson", 1000, true, false, false, false, false);
uPearson.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 1.22089))
return false;
if(!Compare(dissMatrix[2][0], 0.5))
return false;
if(!Compare(dissMatrix[2][1], 1.2008))
return false;
// weighted Yue-Clayton
DiversityCalculator YC("../unit-tests/DataMatrixMothur.env", "", "YueClayton", 1000, true, false, false, false, false);
YC.Dissimilarity("../unit-tests/temp", "UPGMA");
ReadDissMatrix("../unit-tests/temp.diss", dissMatrix);
if(!Compare(dissMatrix[1][0], 0.93233103))
return false;
if(!Compare(dissMatrix[2][0], 0.5))
return false;
if(!Compare(dissMatrix[2][1], 0.92424202))
return false;
return true;
}
void IdentifySuccessors_QP(PathJob* pj, PathNode* node)
{
Vec2i npos = PathNodePos(node);
int thisdistance = Magnitude2(Vec2i(npos.x - pj->ngoalx, npos.y - pj->ngoalz));
if( !pj->closestnode || thisdistance < pj->closest )
{
pj->closestnode = node;
pj->closest = thisdistance;
}
int runningD = 0;
if(node->previous)
runningD = node->previous->totalD;
bool standable[DIRS];
for(int i=0; i<DIRS; i++)
standable[i] = Standable(pj, npos.x + offsets[i].x, npos.y + offsets[i].y);
bool passable[DIRS];
passable[DIR_NW] = standable[DIR_NW] && standable[DIR_N] && standable[DIR_W];
passable[DIR_N] = standable[DIR_N];
passable[DIR_NE] = standable[DIR_NE] && standable[DIR_N] && standable[DIR_E];
passable[DIR_E] = standable[DIR_E];
passable[DIR_SE] = standable[DIR_SE] && standable[DIR_S] && standable[DIR_E];
passable[DIR_S] = standable[DIR_S];
passable[DIR_SW] = standable[DIR_SW] && standable[DIR_S] && standable[DIR_W];
passable[DIR_W] = standable[DIR_W];
for(int i=0; i<DIRS; i++)
{
if(!passable[i])
continue;
int newD = runningD + stepdist[i];
Vec2i nextnpos(npos.x + offsets[i].x, npos.y + offsets[i].y);
PathNode* nextn = PathNodeAt(nextnpos.x, nextnpos.y);
if(!nextn->closed && (!nextn->opened || newD < nextn->totalD))
{
g_toclear.push_back(nextn); // Records this node to reset its properties later.
nextn->totalD = newD;
int H = Manhattan( nextnpos - Vec2i(pj->ngoalx, pj->ngoalz) );
nextn->F = nextn->totalD + H;
nextn->previous = node;
if( !nextn->opened )
{
g_openlist.insert(nextn);
nextn->opened = true;
}
else
{
g_openlist.heapify(nextn);
}
}
}
}
