// Performs K-means clustering
vector<vector<point> > kmeans(vector<point> initcluster, int k){
vector<point> centeroids, old_centeroids; // vector maintaining k centeroids
vector<vector<point> > clusterList; // for storing the list of clusters
vector<double> distance; // temp vector, used for finding the min_distance
int centeroids_change = 1; //Flag which helps the loop to terminate
double min;
int destination,num = 0;
// Intial centeroids are random
for(int i = 0; i < k; ++i){
centeroids.push_back(initcluster[i]);
}
// loop for finding the optimal clusters and centeroids
//while(centeroids_change){
while(num <1500){ // will loop 2000 times, will consider it near optimal solution
clusterList.clear(); // clearing the datastructure, after last loop.
// initializing the clusterList
vector<point> v;
for (int i = 0; i < k; ++i)
{
clusterList.push_back(v);
}
for (int i = 0; i < initcluster.size()-1; ++i){
distance.clear();// will empty the distance vector for future calculations
min = find_distance(initcluster[i], centeroids[0]); //assumption, will improve while running the code
destination = 0;
for (int j = 0; j < centeroids.size(); ++j){
distance.push_back(find_distance(initcluster[i], centeroids[j]));
if (distance[j] <= min){
min = distance[j];
destination = j;
}
}
clusterList.at(destination).push_back(initcluster[i]);
}
num ++;
//cout << "Itr" << num << endl;
old_centeroids = centeroids;
centeroids = mean_of_cluster(clusterList, k);
//centeroids_change = check_change(old_centeroids, centeroids);
}
cout<< "Number of loops iterated for finding Optimal Solution: "<< num<<"\n";
// Printing final K centeroids
cout << "Final Centeroids:"<<"\n";
for(int i=0; i <centeroids.size();++i){
cout<<centeroids[i].id<<" "<<centeroids[i].x<<" "<<centeroids[i].y<<" "<<centeroids[i].z<<"\n";
}
clusterList.push_back(centeroids);
return clusterList;
}
// returns the nodes closest to mean for each of the clusters in clusterlist
vector<point> mean_of_cluster(vector<vector<point> > clusterList){
point sum, mean,candidate;
float distance;
float min_distance;
vector<point> new_centeroids;
for (int i = 0; i < clusterList.size(); ++i){
sum.x = 0;
sum.y = 0;
for (int j = 0; j < clusterList.at(i).size(); ++j){
sum.x = sum.x + clusterList[i][j].x;
sum.y = sum.y + clusterList[i][j].y;
}
mean.x = sum.x / clusterList.at(i).size();
mean.y = sum.y / clusterList.at(i).size();
// have to find the closest node to this mean
min_distance = find_distance(mean, clusterList[i][0]);// assumption
candidate = clusterList[i][0];
for (int n = 0; n < clusterList.at(i).size(); ++n)
{
distance = find_distance(mean, clusterList[i][n]);
if(distance <= min_distance){
min_distance = distance;
candidate = clusterList[i][n];
}
}
new_centeroids.push_back(candidate);
}
return new_centeroids;
}
// Merge clusters which are close to each other
vector<vector<point> > unionClusters(vector<vector<point> > clusterList, vector<point> centeroids) {
vector<double> radius;
double centeroid_dist;
for (int i = 0; i < centeroids.size()-1; ++i){
for (int j = i+1; j < centeroids.size(); ++j){
centeroid_dist = find_distance(centeroids.at(i), centeroids.at(j));
radius = updateradius(clusterList, centeroids); // finding radius of each cluster and storing in radius vector
if(centeroids.size()<=5) continue; // will not merge if total num of clusters have reduced to 5
else if(centeroid_dist < (radius.at(i) + radius.at(j)/1.5)){ // merging of clusters
// merging the two clusters
for(int n=0; n < clusterList.at(j).size(); ++n){
clusterList.at(i).push_back(clusterList[j][n]);
}
// clean up
clusterList.erase(clusterList.begin()+j);// erasing the extra cluster
centeroids.erase(centeroids.begin()+j);// erasing the extra centroid
// updating the centroids
centeroids = mean_of_cluster(clusterList);
break;
}
}
}
return clusterList;
}
// returns the nodes closest to mean for each of the clusters in clusterlist
vector<point> mean_of_cluster(vector<vector<point> > clusterList, int k){
point sum, mean,candidate;
vector<double> distance;
double min_distance;
vector<point> new_centeroids;
for (int i = 0; i < clusterList.size(); ++i)
{
sum.x = 0;
sum.y = 0;
sum.z = 0;
for (int j = 0; j < clusterList[i].size(); ++j)
{
sum.x = sum.x + clusterList[i][j].x;
sum.y = sum.y + clusterList[i][j].y;
sum.z = sum.z + clusterList[i][j].z;
}
mean.x = sum.x / k;
mean.y = sum.y / k;
mean.z = sum.z / k;
// have to find the closest node to this mean
distance.clear();
min_distance = find_distance(mean, clusterList[i][0]);// assumption
candidate = clusterList[i][0];
for (int n = 0; n < clusterList[i].size(); ++n)
{
distance.push_back(find_distance(mean, clusterList[i][n]));
if(distance[n] < min_distance){
min_distance = distance[n];
candidate.x = clusterList[i][n].x;
candidate.y = clusterList[i][n].y;
candidate.z = clusterList[i][n].z;
candidate.id = clusterList[i][n].id;
}
}
new_centeroids.push_back(candidate);
}
return new_centeroids;
}
int expand(int line_count, int vertex[][6], int m, int n, int current_x, int current_y){
int i, j;
for (i=0; i<=line_count; i++){
for (j=0; j<6; j+=2){
if (i==m && j==n){
}
else{
if (check_if_in_triangle(line_count, vertex, vertex[i][j], vertex[i][j+1]) == 0){
if (check_intersect(line_count, vertex, current_x, current_y, vertex[i][j], vertex[i][j+1])== 0){
if( vertex_exist(vertex[i][j], vertex[i][j+1], current_x, current_y) == 0){
valid_vertex[k][0] = current_x;
valid_vertex[k][1] = current_y;
valid_vertex[k][2] = vertex[i][j];
valid_vertex[k][3] = vertex[i][j+1];
valid_vertex[k][4] = find_distance(current_x, current_y, vertex[i][j], vertex[i][j+1]);
valid_vertex[k][5] = -1.0;
valid_vertex[k++][6] = -1.0;
if (vertex_exist(current_x, current_y, vertex[i][j], vertex[i][j+1]) == 0){
valid_vertex[k][0] = vertex[i][j];
valid_vertex[k][1] = vertex[i][j+1];
valid_vertex[k][2] = current_x;
valid_vertex[k][3] = current_y;
valid_vertex[k][4] = find_distance(current_x, current_y, vertex[i][j], vertex[i][j+1]);
valid_vertex[k][5] = -1.0;
valid_vertex[k++][6] = -1.0;
}
}
else{
}
}
else{
}
}
}
}
}
}
//finding radius of each cluster and storing in radius vector
vector<double> updateradius(vector<vector<point> > clusterList, vector<point> centeroids ){
double max, radius_dist;
point p1;
vector<point> chk;
vector<double> radius;
for ( int i = 0; i < clusterList.size(); ++i)
{ //cout<<"hello2.5"<<endl;
max = find_distance(centeroids.at(i), clusterList[i][0]); // assumption
p1 = clusterList[i][0];
for (int j = 0; j < clusterList.at(i).size(); ++j)
{ //cout<<"hello2.75"<<endl;
radius_dist = find_distance(centeroids.at(i),clusterList[i][j]);
if(radius_dist > max){
max = radius_dist;
p1 = clusterList[i][j];
}
}
chk.push_back(p1);
radius.push_back(max);
}
return radius;
}
const char *pick_move(MazeMap *mm, int distsq)
{
Point dst;
find_distance(mm, dist, mm->loc.r, mm->loc.c);
if (mm_count_squares(mm) < WIDTH*HEIGHT)
dst = explore(mm);
else /* mm_count_squares(mm) == WIDTH*HEIGHT */
dst = squash(mm, distsq);
if (dst.r == mm->loc.r && dst.c == mm->loc.c)
return "T";
return construct_turn(mm, dist, mm->loc.r, mm->loc.c, mm->dir, dst.r, dst.c,
NULL, NULL);
}
void main()
{
lcd_init();
init_serial();
Echo = Trig = 0;
TMOD |= 0x01;//Timer 0 in 16-bit mode
P1=0xf0; // set port as input port
P2=0x00; // set port as output port
DelayMs(100);
while(1)
{
if(P1==0xf8)
{
P2=0x0a; //00001010 front
}
else if(P1==0xf1)
{
P2=0x02; //00000010
}
else if(P1==0xf2)
{
P2=0x08; //00001000
}
else if(P1==0xf4)
{
P2=0x05; //00000101 back
}
else if(msg1=='*')
{
find_distance();
msg1=0;
}
else
{
P2=0x00;
}
}
}
//ugal now uses modified comparison, modefied getcredit
void ugal_flatfly_onchip( const Router *r, const Flit *f, int in_channel,
OutputSet *outputs, bool inject )
{
// ( Traffic Class , Routing Order ) -> Virtual Channel Range
int vcBegin = 0, vcEnd = gNumVCs-1;
if ( f->type == Flit::READ_REQUEST ) {
vcBegin = gReadReqBeginVC;
vcEnd = gReadReqEndVC;
} else if ( f->type == Flit::WRITE_REQUEST ) {
vcBegin = gWriteReqBeginVC;
vcEnd = gWriteReqEndVC;
} else if ( f->type == Flit::READ_REPLY ) {
vcBegin = gReadReplyBeginVC;
vcEnd = gReadReplyEndVC;
} else if ( f->type == Flit::WRITE_REPLY ) {
vcBegin = gWriteReplyBeginVC;
vcEnd = gWriteReplyEndVC;
}
assert(((f->vc >= vcBegin) && (f->vc <= vcEnd)) || (inject && (f->vc < 0)));
int out_port;
if(inject) {
out_port = 0;
} else {
int dest = flatfly_transformation(f->dest);
int rID = r->GetID();
int _concentration = gC;
int found;
int debug = 0;
int tmp_out_port, _ran_intm;
int _min_hop, _nonmin_hop, _min_queucnt, _nonmin_queucnt;
int threshold = 2;
if ( in_channel < gC ){
if(gTrace){
cout<<"New Flit "<<f->src<<endl;
}
f->ph = 0;
}
if(gTrace){
int load = 0;
cout<<"Router "<<rID<<endl;
cout<<"Input Channel "<<in_channel<<endl;
//need to modify router to report the buffere depth
load +=r->GetBuffer(in_channel);
cout<<"Rload "<<load<<endl;
}
if (debug){
cout << " FLIT ID: " << f->id << " Router: " << rID << " routing from src : " << f->src << " to dest : " << dest << " f->ph: " <<f->ph << " intm: " << f->intm << endl;
}
// f->ph == 0 ==> make initial global adaptive decision
// f->ph == 1 ==> route nonminimaly to random intermediate node
// f->ph == 2 ==> route minimally to destination
found = 0;
if (f->ph == 1){
dest = f->intm;
}
if (dest >= rID*_concentration && dest < (rID+1)*_concentration) {
if (f->ph == 1) {
f->ph = 2;
dest = flatfly_transformation(f->dest);
if (debug) cout << " done routing to intermediate ";
}
else {
found = 1;
out_port = dest % gC;
if (debug) cout << " final routing to destination ";
}
}
if (!found) {
if (f->ph == 0) {
_min_hop = find_distance(flatfly_transformation(f->src),dest);
_ran_intm = find_ran_intm(flatfly_transformation(f->src), dest);
tmp_out_port = flatfly_outport(dest, rID);
if (f->watch){
*gWatchOut << GetSimTime() << " | " << r->FullName() << " | "
<< " MIN tmp_out_port: " << tmp_out_port;
}
_min_queucnt = r->GetCredit(tmp_out_port, -1, -1);
_nonmin_hop = find_distance(flatfly_transformation(f->src),_ran_intm) + find_distance(_ran_intm, dest);
tmp_out_port = flatfly_outport(_ran_intm, rID);
if (f->watch){
*gWatchOut << GetSimTime() << " | " << r->FullName() << " | "
<< " NONMIN tmp_out_port: " << tmp_out_port << endl;
}
if (_ran_intm >= rID*_concentration && _ran_intm < (rID+1)*_concentration) {
_nonmin_queucnt = numeric_limits<int>::max();
} else {
_nonmin_queucnt = r->GetCredit(tmp_out_port, -1, -1);
}
if (debug){
cout << " _min_hop " << _min_hop << " _min_queucnt: " <<_min_queucnt << " _nonmin_hop: " << _nonmin_hop << " _nonmin_queucnt :" << _nonmin_queucnt << endl;
}
if (_min_hop * _min_queucnt <= _nonmin_hop * _nonmin_queucnt +threshold) {
if (debug) cout << " Route MINIMALLY " << endl;
f->ph = 2;
} else {
// route non-minimally
f->minimal = 0;
if (debug) { cout << " Route NONMINIMALLY int node: " <<_ran_intm << endl; }
f->ph = 1;
f->intm = _ran_intm;
dest = f->intm;
if (dest >= rID*_concentration && dest < (rID+1)*_concentration) {
f->ph = 2;
dest = flatfly_transformation(f->dest);
}
}
}
// find minimal correct dimension to route through
out_port = flatfly_outport(dest, rID);
// if we haven't reached our destination, restrict VCs appropriately to avoid routing deadlock
if(out_port >= gC) {
int const available_vcs = (vcEnd - vcBegin + 1) / 2;
assert(available_vcs > 0);
if(f->ph == 1) {
vcEnd -= available_vcs;
} else {
assert(f->ph == 2);
vcBegin += available_vcs;
}
}
found = 1;
}
if (!found) {
cout << " ERROR: output not found in routing. " << endl;
cout << *f; exit (-1);
}
if (out_port >= gN*(gK-1) + gC) {
cout << " ERROR: output port too big! " << endl;
cout << " OUTPUT select: " << out_port << endl;
cout << " router radix: " << gN*(gK-1) + gK << endl;
exit (-1);
}
if (debug) cout << " through output port : " << out_port << endl;
if(gTrace){cout<<"Outport "<<out_port<<endl;cout<<"Stop Mark"<<endl;}
}
outputs->Clear( );
outputs->AddRange( out_port , vcBegin, vcEnd );
}
bool
can_psidrain(struct creature *ch, struct creature *vict, int *out_dist, bool msg)
{
const char *to_char = NULL;
const char *to_vict = NULL;
int dist;
if (CHECK_SKILL(ch, SKILL_PSIDRAIN) < 30 || !IS_PSIONIC(ch)) {
to_char = "You have no idea how.";
goto failure;
}
if (ROOM_FLAGGED(ch->in_room, ROOM_NOPSIONICS) && GET_LEVEL(ch) < LVL_GOD) {
to_char = "Psychic powers are useless here!";
goto failure;
}
if (GET_LEVEL(vict) >= LVL_AMBASSADOR && GET_LEVEL(ch) < GET_LEVEL(vict)) {
to_char = "You cannot locate them.";
to_vict = "$n has just tried to psidrain you.";
goto failure;
}
if (vict == ch) {
to_char = "Ha ha... Funny!";
goto failure;
}
if (is_fighting(ch) && vict->in_room != ch->in_room) {
to_char = "You cannot focus outside the room during battle!";
goto failure;
}
if (ch->in_room != vict->in_room &&
ch->in_room->zone != vict->in_room->zone) {
to_char = "$N is not in your zone.";
goto failure;
}
if (ROOM_FLAGGED(vict->in_room, ROOM_NOPSIONICS)
&& GET_LEVEL(ch) < LVL_GOD) {
to_char = "Psychic powers are useless where $E is!";
goto failure;
}
if (!ok_to_attack(ch, vict, msg))
return false;
if (GET_MOVE(ch) < 20) {
to_char = "You are too physically exhausted.";
goto failure;
}
dist = find_distance(ch->in_room, vict->in_room);
if (out_dist)
*out_dist = dist;
if (dist > ((GET_LEVEL(ch) / 6))) {
to_char = "$N is out of your psychic range.";
goto failure;
}
if (NULL_PSI(vict)) {
to_char = "It is pointless to attempt this on $M.";
goto failure;
}
return true;
failure:
if (msg) {
if (to_char)
act(to_char, false, ch, NULL, vict, TO_CHAR);
if (to_vict)
act(to_vict, false, ch, NULL, vict, TO_VICT);
}
return false;
}
void start_graph(int line_count, int vertex[][6], int start_x, int start_y, int target_x, int target_y){
/*
Retrieve all valid vertex that will graph all possible routes in the graph.
*/
int i, j;
//initialize valid_vertex[][] to zero
for (i=0; i<600; i++){
valid_vertex[i][0] = 0.0;
valid_vertex[i][1] = 0.0;
valid_vertex[i][2] = 0.0;
valid_vertex[i][3] = 0.0;
valid_vertex[i][4] = 0.0;
valid_vertex[i][5] = 0.0;
valid_vertex[i][6] = 0.0;
}
//store the start point in valid_vertex
valid_vertex[k][0] = start_x; // from x value
valid_vertex[k][1] = start_y; // from y value
valid_vertex[k][2] = start_x; // to x value
valid_vertex[k][3] = start_y; // to y value
valid_vertex[k][4] = 0.0; // distance from point a to b
valid_vertex[k][5] = 0.0; // total distance
valid_vertex[k++][6] = 0.0; // -2.0 target , 1.0 has visited not included, 0.0 using, -1.0 not yet visited
//handles the lines connecting from the starting point and branch out
for (i=0; i<=line_count; i++){
for (j=0; j<6; j+=2){
if (check_if_in_triangle(line_count, vertex, vertex[i][j], vertex[i][j+1]) == 0){
if (check_intersect(line_count, vertex, start_x, start_y, vertex[i][j], vertex[i][j+1]) == 0){
valid_vertex[k][0] = start_x;
valid_vertex[k][1] = start_y;
valid_vertex[k][2] = vertex[i][j];
valid_vertex[k][3] = vertex[i][j+1];
valid_vertex[k][4] = find_distance(vertex[i][j], vertex[i][j+1], start_x, start_y);
valid_vertex[k][5] = -1.0;
valid_vertex[k++][6] = -1.0;
valid_vertices(line_count, vertex, i, vertex[i][j], vertex[i][j+1]);
}
}
}
}
//handles the lines connecting to the target point.
for(i=0; i<=line_count; i++){
for(j=0; j<6; j+=2){
if(check_if_in_triangle(line_count, vertex, vertex[i][j], vertex[i][j+1])==0){
if(check_intersect(line_count, vertex, vertex[i][j], vertex[i][j+1], target_x, target_y) == 0){
if (vertex_exist(vertex[i][j], vertex[i][j+1], target_x, target_y) == 0){
valid_vertex[k][0] = vertex[i][j];
valid_vertex[k][1] = vertex[i][j+1];
valid_vertex[k][2] = target_x;
valid_vertex[k][3] = target_y;
valid_vertex[k][4] = find_distance(vertex[i][j], vertex[i][j+1], target_x, target_y);
valid_vertex[k][5] = -1.0;
valid_vertex[k++][6] = -2.0;
}
}
}
}
}
}
// Performs K-means clustering and return the sorted vector of clusters
vector<point> kmeans(vector<point> initcluster, int k){
vector<point> centeroids, old_centeroids; // vector maintaining k centeroids
vector<vector<point> > clusterList; // for storing the list of clusters
float distance1; // temp vector, used for finding the min_distance
int centeroids_change = 1; //Flag which helps the loop to terminate
float min;
int destination,num = 0;
// Intial centeroids are random
for(int i = 0; i < k; ++i){
centeroids.push_back(initcluster[i]);
}
// loop for finding the optimal clusters and centeroids
while(centeroids_change){
clusterList.clear(); // clearing the datastructure, after last loop.
// initializing the clusterList
vector<point> v;
for (int i = 0; i < k; ++i)
{
clusterList.push_back(v);
}
for (int i = 0; i < initcluster.size()-1; ++i){
min = find_distance(initcluster[i], centeroids[0]); //assumption, will improve while running the code
destination = 0;
for (int j = 0; j < centeroids.size(); ++j){
distance1 = find_distance(initcluster[i], centeroids[j]);
if (distance1 <= min){
min = distance1;
destination = j;
}
}
clusterList.at(destination).push_back(initcluster[i]);
}
num ++;
old_centeroids = centeroids;
centeroids = mean_of_cluster(clusterList);
centeroids_change = check_change(old_centeroids, centeroids);
}
cout<< "Number of loops iterated for finding Optimal Solution: "<< num<<"\n";
/*
// Printing final K centeroids
cout << "Initial Centeroids:"<<"\n";
for(int i=0; i <centeroids.size();++i){
cout<<centeroids[i].x<<", "<<centeroids[i].y<<"; "<<"\n";
}
*/
// merging clusters which overlap
vector<vector<point> > updatedClusterList = unionClusters(clusterList, centeroids);
// sorting the cluster List
sort (updatedClusterList.begin(), updatedClusterList.end(), sortfunction);
cout<<endl<<"Sorted Clusters";
print(updatedClusterList);
updatedClusterList.erase(updatedClusterList.begin()+5, updatedClusterList.end());
cout<<endl<<endl<<"Final 5 Clusters";
print(updatedClusterList);
centeroids = mean_of_cluster(updatedClusterList);
return centeroids;
}