public static Tuple <List <int>, double> UnifiedCGreedy(Graph graph, Bipartite bg, List <int> Type, double c, double B, double alpha)
{
List <double> P = new List <double>();
for (int id = 0; id < Type.Count; ++id)
{
int type = Type[id];
double p;
if (type == 1)
{
p = c * c;
}
else if (type == 2)
{
p = c;
}
else
{
p = (2 - c) * c;
}
P.Add(p);
}
int k = (int)(B / c);
Tuple <List <int>, double> tup = bg.Greedy(k, P);
Console.WriteLine("seeds have been selected");
return(tup);
}
public CoordinateDescent(Graph graph, Bipartite bg, List<int> initNodes, double init_c, List<int> type, int batch_num, double alpha, int mH)
{
this.graph = graph;
this.initNodes = initNodes;
this.init_c = init_c;
this.type = type;
this.batch_num = batch_num;
this.alpha = alpha;
this.mH = mH;
this.bg = bg;
C = new List<double>();
for (int i = 0; i < graph.numV; ++i)
C.Add(0.0);
foreach (int u in initNodes)
C[u] = init_c;
prob_edge = new List<double>();
for (int h = 0; h < bg.numS; ++h)
{
double res = 1.0;
foreach (int u in bg.S2V[h])
res *= (1 - SeedProb(u, C[u]));
//res = 1.0 - res;
prob_edge.Add(res);
}
}
public CoordinateDescent(Graph graph, Bipartite bg, List <int> initNodes, double init_c, int batch_num, double alpha, int mH)
{
this.graph = graph;
this.initNodes = initNodes;
this.init_c = init_c;
this.batch_num = batch_num;
this.alpha = alpha;
this.mH = mH;
this.bg = bg;
this.x = new List <int>();
foreach (int u in initNodes)
{
x.Add(u);
}
C = new List <double>();
for (int i = 0; i < graph.numV; ++i)
{
C.Add(init_c);
}
prob_edge = new List <double>();
for (int h = 0; h < bg.numS; ++h)
{
prob_edge.Add(1.0);
}
}
public CoordinateDescent(Graph graph, Bipartite bg, List <int> initNodes, List <int> seedNodes, double init_c, List <int> type, int batch_num, double alpha, int mH)
{
this.graph = graph;
this.initNodes = initNodes;
this.init_c = init_c;
this.type = type;
this.batch_num = batch_num;
this.alpha = alpha;
this.mH = mH;
this.bg = bg;
C = new List <double>();
for (int i = 0; i < graph.numV; ++i)
{
C.Add(0.0);
}
foreach (int u in seedNodes)
{
C[u] = init_c;
}
prob_edge = new List <double>();
for (int h = 0; h < bg.numS; ++h)
{
double res = 1.0;
foreach (int u in bg.S2V[h])
{
res *= (1 - SeedProb(u, C[u]));
}
//res = 1.0 - res;
prob_edge.Add(res);
}
}
public CoordinateDescent(Graph graph, List<int> initNodes, double init_c, List<int> type, int batch_num, double alpha, int mH)
{
this.graph = graph;
this.initNodes = initNodes;
this.init_c = init_c;
this.type = type;
this.batch_num = batch_num;
this.alpha = alpha;
this.mH = mH;
C = new List<double>();
for (int i = 0; i < graph.numV; ++i)
C.Add(0.0);
foreach (int u in initNodes)
C[u] = init_c;
ICModel icm = new ICModel(alpha);
List<List<int>> RR = new List<List<int>>();
for (int r = 0; r < mH; ++r)
{
HashSet<int> rawSet = icm.RR(graph);
List<int> rSet = rawSet.ToList();
RR.Add(rSet);
}
bg = new Bipartite(RR, graph.numV);
prob_edge = new List<double>();
for (int h = 0; h < bg.numS; ++h)
{
double res = 1.0;
foreach (int u in bg.S2V[h])
res *= (1 - SeedProb(u, C[u]));
//res = 1.0 - res;
prob_edge.Add(res);
}
}
public CoordinateDescent(Graph graph, List <int> initNodes, List <int> seedNodes, double init_c, List <int> type, int batch_num, double alpha, int mH)
{
this.graph = graph;
this.initNodes = initNodes;
this.init_c = init_c;
this.type = type;
this.batch_num = batch_num;
this.alpha = alpha;
this.mH = mH;
C = new List <double>();
for (int i = 0; i < graph.numV; ++i)
{
C.Add(0.0);
}
foreach (int u in seedNodes)
{
C[u] = init_c;
}
ICModel icm = new ICModel(alpha);
List <List <int> > RR = new List <List <int> >();
for (int r = 0; r < mH; ++r)
{
HashSet <int> rawSet = icm.RR(graph);
List <int> rSet = rawSet.ToList();
RR.Add(rSet);
}
bg = new Bipartite(MainClass.filepath, 0.6, graph.numV);
prob_edge = new List <double>();
for (int h = 0; h < bg.numS; ++h)
{
double res = 1.0;
foreach (int u in bg.S2V[h])
{
res *= (1 - SeedProb(u, C[u]));
}
//res = 1.0 - res;
prob_edge.Add(res);
}
}
public static Tuple <List <int>, double> UnifiedCGreedy(Graph graph, List <int> Type, double c, double B, double alpha)
{
List <double> P = new List <double>();
for (int id = 0; id < Type.Count; ++id)
{
int type = Type[id];
double p;
if (type == 1)
{
p = c * c;
}
else if (type == 2)
{
p = c;
}
else
{
p = (2 - c) * c;
}
P.Add(p);
}
ICModel icm = new ICModel(alpha);
int mH = GlobalVar.mH;
List <List <int> > RR = new List <List <int> >();
for (int r = 0; r < mH; ++r)
{
List <int> rSet = icm.RR(graph).ToList();
RR.Add(rSet);
}
Bipartite bg = new Bipartite(RR, graph.numV);
Console.WriteLine("Hyper-graph has been built");
int k = (int)(B / c);
Tuple <List <int>, double> tup = bg.Greedy(k, P);
Console.WriteLine("seeds have been selected");
return(tup);
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
List <int> seeds = new List <int>();
for (int i = 0; i < 100; i++)
{
seeds.Add(int.Parse(initial.ReadLine()));
}
double alpha = 0.6; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 1.0)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
int b = 10;
while (b <= 50)
// Build a random hyper graph with mH random hyper edges.
{
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
Tuple <List <int>, double> choose = bg.Greedy(b, seeds);
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
FileStream outfile = new FileStream(filepath + "_2o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:\t" + Hyper_time + "\t");
Hyper_start = DateTime.Now;
Tuple <double, double> results = icm.InfluenceSpread(graph, choose.Item1, 100);
Hyper_end = DateTime.Now;
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:\t" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10;
}
alpha += 0.2;
}
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
StreamReader threshold = new StreamReader(filepath + "_tu.txt");
List <double> thresh = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
thresh.Add(double.Parse(threshold.ReadLine()));
}
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <double> d = new List <double> {
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0
};
List <double> cu = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
double t = thresh[i];
double p = 0.0;
if (type[i] == 1)
{
p = System.Math.Pow(t, 0.5);
}
else if (type[i] == 2)
{
p = t;
}
else
{
p = 1 - System.Math.Pow(1 - t, 0.5);
}
foreach (double point in d)
{
if (point < p)
{
continue;
}
cu.Add(point);
break;
}
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 1.0; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 1.0)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
double b = 10.0;
while (b <= 10.0)
// Build a random hyper graph with mH random hyper edges.
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
List <int> seed = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
CoordinateDescent cd = new CoordinateDescent(graph, bg, seed, 0.0, type, 20, alpha, mh);
double bused = 0.0;
while (bused <= b)
{
int flag = seed[0];
double maxE = 0.0;
List <int> bestseed = new List <int>();
List <double> bestallo = new List <double>();
foreach (int u in seed)
{
List <int> nrlist = graph.newreach(new List <int> {
u
}, cd.x);
double b2cub1 = 5 * cu[u];
if (b2cub1 >= 0.5 && nrlist.Count >= 1)
{
Tuple <List <int>, List <double>, double> result1 = cd.bestone(nrlist, b2cub1);
if (result1.Item3 / cu[u] >= maxE / cu[flag])
{
maxE = result1.Item3;
flag = u;
bestseed = new List <int>();
foreach (int point in result1.Item1)
{
bestseed.Add(point);
}
bestallo = new List <double>();
foreach (double resource in result1.Item2)
{
bestallo.Add(resource);
}
}
}
if (b2cub1 >= 1.0 && nrlist.Count >= 2)
{
Tuple <List <int>, List <double>, double> result2 = cd.besttwo(nrlist, b2cub1);
if (result2.Item3 / cu[u] >= maxE / cu[flag])
{
maxE = result2.Item3;
flag = u;
bestseed = new List <int>();
foreach (int point in result2.Item1)
{
bestseed.Add(point);
}
bestallo = new List <double>();
foreach (double resource in result2.Item2)
{
bestallo.Add(resource);
}
}
}
if (b2cub1 >= 1.5 && nrlist.Count >= 3)
{
Tuple <List <int>, List <double>, double> result3 = cd.bestone(nrlist, b2cub1);
if (result3.Item3 / cu[u] >= maxE / cu[flag])
{
maxE = result3.Item3;
flag = u;
bestseed = new List <int>();
foreach (int point in result3.Item1)
{
bestseed.Add(point);
}
bestallo = new List <double>();
foreach (double resource in result3.Item2)
{
bestallo.Add(resource);
}
}
}
}
if (bused + cu[flag] > b)
{
break;
}
bused += cu[flag];
seed.Remove(flag);
List <int> maxnr = graph.newreach(new List <int> {
flag
}, cd.x);
for (int u = 0; u < bestseed.Count; u++)
{
if (bused + bestallo[u] > b)
{
break;
}
cd.ChangeAllocation(bestseed[u], bestallo[u]);
cd.x.Add(bestseed[u]);
bused += bestallo[u];
}
Console.WriteLine("newreach:" + Convert.ToString(cd.C.Sum()) + "total:" + Convert.ToString(bused));
}
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Hyper_start = DateTime.Now;
List <double> prob = new List <double>();
foreach (int u in cd.x)
{
prob.Add(cd.SeedProb(u, cd.C[u]));
}
Tuple <double, double> results = icm.InfluenceSpread(graph, cd.x, prob, 200);
Hyper_end = DateTime.Now;
FileStream outfile = new FileStream(filepath + "_8o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10.0;
}
alpha += 0.2;
}
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
StreamReader threshold = new StreamReader(filepath + "_tu.txt");
List <double> thresh = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
thresh.Add(double.Parse(threshold.ReadLine()));
}
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <double> d = new List <double> {
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0
};
List <double> cu = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
double t = thresh[i];
double p = 0.0;
if (type[i] == 1)
{
p = System.Math.Pow(t, 0.5);
}
else if (type[i] == 2)
{
p = t;
}
else
{
p = 1 - System.Math.Pow(1 - t, 0.5);
}
foreach (double point in d)
{
if (point < p)
{
continue;
}
cu.Add(point);
break;
}
}
List <int> seed = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 0.8; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 0.8)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
double b = 10;
while (b <= 10)
// Build a random hyper graph with mH random hyper edges.
{
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
Tuple <List <int>, double> realize = bg.Greedy(cu, b, seed);
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
FileStream outfile = new FileStream(filepath + "_5o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
Console.WriteLine(realize.Item1.Count);
Hyper_start = DateTime.Now;
Tuple <double, double> results = icm.InfluenceSpread(graph, realize.Item1, 200);
Hyper_end = DateTime.Now;
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10.0;
}
alpha += 0.2;
}
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
StreamReader threshold = new StreamReader(filepath + "_tu.txt");
List <double> thresh = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
thresh.Add(double.Parse(threshold.ReadLine()));
}
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <double> d = new List <double> {
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0
};
List <double> cu = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
//cu.Add(1.0);
double t = thresh[i];
double p = 0.0;
if (type[i] == 1)
{
p = System.Math.Pow(t, 0.5);
}
else if (type[i] == 2)
{
p = t;
}
else
{
p = 1 - System.Math.Pow(1 - t, 0.5);
}
foreach (double point in d)
{
if (point < p)
{
continue;
}
cu.Add(point);
break;
}
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 0.6; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 1.0)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
double b = 10.0;
double ratio = 3.0;
while (b <= 50.0)
// Build a random hyper graph with mH random hyper edges.
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
List <int> seed = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
CoordinateDescent cd = new CoordinateDescent(graph, bg, seed, 0.0, type, 20, alpha, mh);
double bused = 0.0;
while (bused <= b)
{
int flag = seed[0];
double maxE = 0.0;
int circumstance = 0;
List <int> seedset = new List <int>();
foreach (int u in seed)
{
List <int> nrlist = graph.newreach(new List <int> {
u
}, cd.x);
double b2cub1 = ratio * cu[u];
Tuple <List <int>, double> choose = cd.gs(b2cub1, nrlist);
if (choose.Item2 / cu[u] >= maxE / cu[flag])
{
maxE = choose.Item2;
flag = u;
seedset = new List <int>();
foreach (int point in choose.Item1)
{
seedset.Add(point);
}
circumstance = 1;
}
b2cub1 = b2cub1 >= 1.0 ? 1.0 : b2cub1;
foreach (int v in nrlist)
{
double nowE = cd.singlepointgain(v, b2cub1);
if (nowE / cu[v] >= maxE / cu[flag])
{
maxE = nowE;
flag = u;
seedset = new List <int> {
v
};
circumstance = 2;
}
}
}
if (bused + cu[flag] > b)
{
break;
}
bused += cu[flag];
seed.Remove(flag);
List <int> maxnr = graph.newreach(new List <int> {
flag
}, cd.x);
if (circumstance <= 2)
{
foreach (int u in seedset)
{
double allocation = 0.0;
if (type[u] == 1)
{
allocation = 1.0;
}
else
{
allocation = 0.5;
}
//allocation = 1.0;
if (bused + allocation > b)
{
break;
}
cd.ChangeAllocation(u, allocation);
cd.x.Add(u);
Console.Write(Convert.ToString(u) + " ");
bused += allocation;
}
}
else
{
double allocation = cu[flag] * ratio;
allocation = allocation >= 1.0 ? 1.0 : allocation;
if (bused + allocation > b)
{
break;
}
cd.ChangeAllocation(seedset[0], allocation);
cd.x.Add(seedset[0]);
Console.Write(Convert.ToString(seedset[0]) + " ");
bused += allocation;
}
Console.WriteLine("newreach:" + Convert.ToString(cd.C.Sum()) + "total:" + Convert.ToString(bused));
}
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Console.WriteLine(cd.x.Count);
Hyper_start = DateTime.Now;
List <double> prob = new List <double>();
foreach (int u in cd.x)
{
prob.Add(cd.SeedProb(u, cd.C[u]));
}
Tuple <double, double> results = icm.InfluenceSpread(graph, cd.x, prob, 200);
Hyper_end = DateTime.Now;
FileStream outfile = new FileStream(filepath + "_7o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10.0;
}
alpha += 0.2;
}
}
//public static string filepath = "D:/iiot/csharp/im/Wiki-Vote";
public static void Main(string[] args)
{
Graph graph = new Graph(filepath + ".csv");
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 0.6; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 0.6)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
double b = 10.0;
double ratio = 4.0;
while (b <= 50.0)
// Build a random hyper graph with mH random hyper edges.
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
List <int> seed = new List <int>();
List <int> final = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
CoordinateDescent cd = new CoordinateDescent(graph, bg, seed, 0.0, 10, alpha, mh);
double bused = 0.0;
while (bused <= b)
{
int flag = seed[0];
double maxE = 0.0;
List <int> seedset = new List <int>();
foreach (int u in seed)
{
List <int> nrlist = graph.newreach(new List <int> {
u
}, cd.x);
double b2cub1 = ratio;
Tuple <List <int>, double> choose = cd.gs(b2cub1, nrlist);
if (choose.Item2 >= maxE)
{
maxE = choose.Item2;
flag = u;
seedset = new List <int>();
foreach (int point in choose.Item1)
{
seedset.Add(point);
}
}
}
if (bused + 1.0 > b)
{
break;
}
bused += 1.0;
seed.Remove(flag);
foreach (int u in seedset)
{
if (bused + 1.0 > b)
{
break;
}
cd.ChangeAllocation(u, 1.0);
cd.x.Add(u);
final.Add(u);
Console.Write(Convert.ToString(u) + " ");
bused += 1.0;
}
Console.WriteLine("newreach:" + Convert.ToString(cd.C.Sum()) + "total:" + Convert.ToString(bused));
}
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Console.WriteLine(final.Count);
Hyper_start = DateTime.Now;
Tuple <double, double> results = icm.InfluenceSpread(graph, final, 20000, 0.7);
Hyper_end = DateTime.Now;
FileStream outfile = new FileStream(filepath + "_9n.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10.0;
}
alpha += 0.2;
}
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph, List<int> Type, string RsltDir)
{
double b = GlobalVar.b; // Step of c of searching the best discount in th Unified Discount Algorithm
double alpha = GlobalVar.Alpha;
string Dir = RsltDir + "/Alpha=" + alpha;
string Path = Dir + "/AllResults.txt";
StreamWriter writer = new StreamWriter(Path);
// Build a random hyper graph with mH random hyper edges.
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
int mH = GlobalVar.mH;
Console.WriteLine(mH);
List<List<int>> RR = new List<List<int>>();
for (int r = 0; r < mH; ++r)
{
List<int> rSet = icm.RR(graph).ToList();
RR.Add(rSet);
if (r > 0 && r % 100000 == 0)
Console.WriteLine(r + " samples");
}
Bipartite bg = new Bipartite(RR, graph.numV);
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Console.WriteLine("Hyper-graph has been built");
writer.WriteLine("Hyper-graph time:\t" + Hyper_time);
writer.Flush();
for (int ind = GlobalVar.St; ind <= GlobalVar.End; ++ind)
{
double B = ind * 10.0;
bg.Greedy((int)B);
// Unified Discount Algorithm
DateTime startTime = DateTime.Now;
List<Tuple<List<int>, double>> Res = new List<Tuple<List<int>, double>>();
for (int i = 1; i*b - 1.0 <= 1e-4; ++i)
{
double c = i * b;
Tuple<List<int>, double> tup = UnifiedCGreedy(graph, bg, Type, c, B, alpha);
Res.Add(tup);
Console.WriteLine("Alpha=" + alpha + "\tc=" + c + "\t" + tup.Item2);
}
// Discrete Influence Maximization
double IM_greedy = 0;
DateTime IM_greedy_start = DateTime.Now;
Tuple<List<int>, double> tup_IM = UnifiedCGreedy(graph, bg, Type, 20 * b, B, alpha);
Res.Add(tup_IM);
Console.WriteLine("Alpha=" + alpha + "\tc=1\t" + tup_IM.Item2);
DateTime IM_greedy_end = DateTime.Now;
IM_greedy = (IM_greedy_end - IM_greedy_start).TotalMilliseconds;
//Influence Maximization results
double IM_sp = Res[Res.Count - 1].Item2;
List<int> IM_seeds = Res[Res.Count - 1].Item1;
double IM_ts = Hyper_time + IM_greedy;
// Unified Discount results (except standard deviation)
int max_i = 0;
for (int i = 0; i < Res.Count - 1; ++i)
{
if (Res[i].Item2 > Res[max_i].Item2)
max_i = i;
}
DateTime UC_endTime = DateTime.Now;
Console.WriteLine("Best c=" + (max_i + 1) * b + "\t" + Res[max_i].Item2);
double UC_sp = Res[max_i].Item2;
List<int> UC_seeds = Res[max_i].Item1;
double UC_ts = (UC_endTime - startTime).TotalMilliseconds + Hyper_time;
// Proceed to Coordinate Descent. Use the best result of Unified Discount as initial value
double init_c = (max_i + 1) * b;
List<int> initNodes = UC_seeds;
CoordinateDescent cd = new CoordinateDescent(graph, bg, initNodes, init_c, Type, GlobalVar.batch_num, alpha, GlobalVar.mH);
List<double> C = cd.IterativeMinimize();
DateTime CD_endTime = DateTime.Now;
double CD_ts = (CD_endTime - startTime).TotalMilliseconds + Hyper_time;
// Evaluation using Monte Carlo Simulations
List<Pair> IM_P = new List<Pair>();
foreach (int u in IM_seeds)
{
Pair pair = new Pair(u, 1.0);
IM_P.Add(pair);
}
List<Pair> UC_P = new List<Pair>();
foreach (int u in UC_seeds)
{
double p = 0;
if (Type[u] == 1)
p = init_c * init_c;
else if (Type[u] == 2)
p = init_c;
else
p = (2 - init_c) * init_c;
Pair pair = new Pair(u, p);
UC_P.Add(pair);
}
List<Pair> CD_P = new List<Pair>();
for (int i = 0; i < graph.numV; ++i)
{
double p = 0;
if (Type[i] == 1)
p = C[i] * C[i];
else if (Type[i] == 2)
p = C[i];
else
p = (2 - C[i]) * C[i];
Pair pair = new Pair(i, p);
if (p > GlobalVar.epsilon)
CD_P.Add(pair);
}
// //ICModel icm = new ICModel(alpha);
Tuple<double, double> IM_tup = icm.InfluenceSpread(graph, IM_P, GlobalVar.MC);
Tuple<double, double> UC_tup = icm.InfluenceSpread(graph, UC_P, GlobalVar.MC);
Tuple<double, double> CD_tup = icm.InfluenceSpread(graph, CD_P, GlobalVar.MC);
double Numerator = 2 * graph.numV * (1 - 1.0 / Math.E) * (Math.Log(Cnk(graph.numV, (int)B)) + Math.Log(graph.numV) + Math.Log(2.0));
double appro = 1 - 1.0 / Math.E - Math.Sqrt(Numerator / (IM_tup.Item1*(double)GlobalVar.mH));
writer.WriteLine("B=" + B);
writer.WriteLine("IM:\t" + IM_tup.Item1 + "\t" + IM_tup.Item2 + "\t" + IM_ts + "\t" + appro);
writer.WriteLine("UC:\t" + UC_tup.Item1 + "\t" + UC_tup.Item2 + "\t" + UC_ts);
writer.WriteLine("CD:\t" + CD_tup.Item1 + "\t" + CD_tup.Item2 + "\t" + CD_ts);
writer.Flush();
string outPath = Dir + "/B=" + B + ".txt";
StreamWriter outWriter = new StreamWriter(outPath);
outWriter.WriteLine("IM\t" + IM_ts + "\t" + IM_seeds.Count);
foreach (int u in IM_seeds)
outWriter.WriteLine(u + "\t1\t1");
outWriter.WriteLine("UC\t" + UC_ts + "\t" + UC_seeds.Count);
foreach (Pair pair in UC_P)
outWriter.WriteLine(pair.id + "\t" + init_c + "\t" + pair.prob);
outWriter.WriteLine("CD\t" + CD_ts + "\t" + CD_P.Count);
foreach (Pair pair in CD_P)
outWriter.WriteLine(pair.id + "\t" + C[pair.id] + "\t" + pair.prob);
outWriter.Close();
string ccurvePath = Dir + "/curve_c(" + B + ").txt";
StreamWriter cWriter = new StreamWriter(ccurvePath);
for (int i = 0; i < Res.Count; ++i)
{
double c = (i + 1) * b;
cWriter.WriteLine(c + "\t" + Res[i].Item2);
}
cWriter.Close();
}
writer.Close();
ProcessData4Visual.Fig3(Path, Dir + "/Fig3.txt");
ProcessData4Visual.Fig4(Path, Dir + "/Fig4.txt");
ProcessData4Visual.Fig5(Dir + "/curve_c(50).txt", Dir + "/Fig5.txt");
ProcessData4Visual.Fig6(Path, Dir + "/Fig6.txt");
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph, List <int> Type, string RsltDir)
{
double b = GlobalVar.b; // Step of c of searching the best discount in th Unified Discount Algorithm
double alpha = GlobalVar.Alpha;
string Dir = RsltDir + "/Alpha=" + alpha;
string Path = Dir + "/AllResults.txt";
StreamWriter writer = new StreamWriter(Path);
// Build a random hyper graph with mH random hyper edges.
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
int mH = GlobalVar.mH;
Console.WriteLine(mH);
List <List <int> > RR = new List <List <int> >();
for (int r = 0; r < mH; ++r)
{
List <int> rSet = icm.RR(graph).ToList();
RR.Add(rSet);
if (r > 0 && r % 100000 == 0)
{
Console.WriteLine(r + " samples");
}
}
Bipartite bg = new Bipartite(RR, graph.numV);
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Console.WriteLine("Hyper-graph has been built");
writer.WriteLine("Hyper-graph time:\t" + Hyper_time);
writer.Flush();
for (int ind = GlobalVar.St; ind <= GlobalVar.End; ++ind)
{
double B = ind * 10.0;
bg.Greedy((int)B);
// Unified Discount Algorithm
DateTime startTime = DateTime.Now;
List <Tuple <List <int>, double> > Res = new List <Tuple <List <int>, double> >();
for (int i = 1; i <= 20; ++i)
{
double c = i * b;
Tuple <List <int>, double> tup = UnifiedCGreedy(graph, bg, Type, c, B, alpha);
Res.Add(tup);
Console.WriteLine("Alpha=" + alpha + "\tc=" + c + "\t" + tup.Item2);
}
// Discrete Influence Maximization
double IM_greedy = 0;
DateTime IM_greedy_start = DateTime.Now;
Tuple <List <int>, double> tup_IM = UnifiedCGreedy(graph, bg, Type, 20 * b, B, alpha);
Res.Add(tup_IM);
Console.WriteLine("Alpha=" + alpha + "\tc=1\t" + tup_IM.Item2);
DateTime IM_greedy_end = DateTime.Now;
IM_greedy = (IM_greedy_end - IM_greedy_start).TotalMilliseconds;
//Influence Maximization results
double IM_sp = Res[Res.Count - 1].Item2;
List <int> IM_seeds = Res[Res.Count - 1].Item1;
double IM_ts = Hyper_time + IM_greedy;
// Unified Discount results (except standard deviation)
int max_i = 0;
for (int i = 0; i < Res.Count - 1; ++i)
{
if (Res[i].Item2 > Res[max_i].Item2)
{
max_i = i;
}
}
DateTime UC_endTime = DateTime.Now;
Console.WriteLine("Best c=" + (max_i + 1) * b + "\t" + Res[max_i].Item2);
double UC_sp = Res[max_i].Item2;
List <int> UC_seeds = Res[max_i].Item1;
double UC_ts = (UC_endTime - startTime).TotalMilliseconds + Hyper_time;
// Proceed to Coordinate Descent. Use the best result of Unified Discount as initial value
double init_c = (max_i + 1) * b;
List <int> initNodes = UC_seeds;
CoordinateDescent cd = new CoordinateDescent(graph, bg, initNodes, init_c, Type, GlobalVar.batch_num, alpha, GlobalVar.mH);
List <double> C = cd.IterativeMinimize();
DateTime CD_endTime = DateTime.Now;
double CD_ts = (CD_endTime - startTime).TotalMilliseconds + Hyper_time;
// Evaluation using Monte Carlo Simulations
List <Pair> IM_P = new List <Pair>();
foreach (int u in IM_seeds)
{
Pair pair = new Pair(u, 1.0);
IM_P.Add(pair);
}
List <Pair> UC_P = new List <Pair>();
foreach (int u in UC_seeds)
{
double p = 0;
if (Type[u] == 1)
{
p = init_c * init_c;
}
else if (Type[u] == 2)
{
p = init_c;
}
else
{
p = (2 - init_c) * init_c;
}
Pair pair = new Pair(u, p);
UC_P.Add(pair);
}
List <Pair> CD_P = new List <Pair>();
for (int i = 0; i < graph.numV; ++i)
{
double p = 0;
if (Type[i] == 1)
{
p = C[i] * C[i];
}
else if (Type[i] == 2)
{
p = C[i];
}
else
{
p = (2 - C[i]) * C[i];
}
Pair pair = new Pair(i, p);
if (p > GlobalVar.epsilon)
{
CD_P.Add(pair);
}
}
// //ICModel icm = new ICModel(alpha);
Tuple <double, double> IM_tup = icm.InfluenceSpread(graph, IM_P, GlobalVar.MC);
Tuple <double, double> UC_tup = icm.InfluenceSpread(graph, UC_P, GlobalVar.MC);
Tuple <double, double> CD_tup = icm.InfluenceSpread(graph, CD_P, GlobalVar.MC);
double Numerator = 2 * graph.numV * (1 - 1.0 / Math.E) * (Math.Log(Cnk(graph.numV, (int)B)) + Math.Log(graph.numV) + Math.Log(2.0));
double appro = 1 - 1.0 / Math.E - Math.Sqrt(Numerator / (IM_tup.Item1 * (double)GlobalVar.mH));
writer.WriteLine("B=" + B);
writer.WriteLine("IM:\t" + IM_tup.Item1 + "\t" + IM_tup.Item2 + "\t" + IM_ts + "\t" + appro);
writer.WriteLine("UC:\t" + UC_tup.Item1 + "\t" + UC_tup.Item2 + "\t" + UC_ts);
writer.WriteLine("CD:\t" + CD_tup.Item1 + "\t" + CD_tup.Item2 + "\t" + CD_ts);
writer.Flush();
string outPath = Dir + "/B=" + B + ".txt";
StreamWriter outWriter = new StreamWriter(outPath);
outWriter.WriteLine("IM\t" + IM_ts + "\t" + IM_seeds.Count);
foreach (int u in IM_seeds)
{
outWriter.WriteLine(u + "\t1\t1");
}
outWriter.WriteLine("UC\t" + UC_ts + "\t" + UC_seeds.Count);
foreach (Pair pair in UC_P)
{
outWriter.WriteLine(pair.id + "\t" + init_c + "\t" + pair.prob);
}
outWriter.WriteLine("CD\t" + CD_ts + "\t" + CD_P.Count);
foreach (Pair pair in CD_P)
{
outWriter.WriteLine(pair.id + "\t" + C[pair.id] + "\t" + pair.prob);
}
outWriter.Close();
string ccurvePath = Dir + "/curve_c(" + B + ").txt";
StreamWriter cWriter = new StreamWriter(ccurvePath);
for (int i = 0; i < Res.Count; ++i)
{
double c = (i + 1) * b;
cWriter.WriteLine(c + "\t" + Res[i].Item2);
}
cWriter.Close();
}
writer.Close();
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <int> seed = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 0.6; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 0.6)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
int b1 = 5;
while (b1 <= 5)
// Build a random hyper graph with mH random hyper edges.
{
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
int count = Convert.ToInt16(1.5 * Convert.ToDouble(b1));
List <int> seed1 = graph.findlarge(seed, count);
CoordinateDescent cd1 = new CoordinateDescent(graph, bg, seed, seed1, 0.66667, type, 50, alpha, mh);
Random random = new Random();
int b2 = 25;
count = Convert.ToInt16(1.5 * Convert.ToDouble(b2));
int i1 = 0;
while (i1 < 5)
{
i1++;
int i = seed1[random.Next(0, seed1.Count)];
if (Math.Abs(1.0 - cd1.C[i]) < GlobalVar.epsilon)
{
continue;
}
int j = seed1[random.Next(0, seed1.Count)];
if ((i == j) || Math.Abs(1.0 - cd1.C[j]) < GlobalVar.epsilon)
{
continue;
}
double bp = cd1.C[i] + cd1.C[j];
double left = (bp - 1 > 0.0) ? (bp - 1) : 0.0;
double right = (bp < 1.0) ? bp : 1.0;
if (left >= right)
{
continue;
}
double sumns = 0.0, sumnsi = 0.0, sumnsj = 0.0, sumnsij = 0.0;
for (int i2 = 0; i2 < 5; i2++)
{
List <int> realize = new List <int>();
foreach (int u in seed1)
{
if ((u == i) || (u == j))
{
continue;
}
if (random.NextDouble() < cd1.SeedProb(u, cd1.C[u]))
{
realize.Add(u);
}
}
List <int> ns = graph.newreach(realize, seed);
List <int> seed2 = graph.findlarge(ns, count);
CoordinateDescent cd2 = new CoordinateDescent(graph, bg, ns, seed2, 0.66667, type, 10, alpha, mh);
List <double> cns = cd2.IterativeMinimize();
double qns = cd2.expectation();
sumns += qns;
realize.Add(i);
ns = graph.newreach(realize, seed);
seed2 = graph.findlarge(ns, count);
cd2 = new CoordinateDescent(graph, bg, ns, seed2, cns, type, 10, alpha, mh);
List <double> cnsi = cd2.IterativeMinimize();
double qnsi = cd2.expectation();
sumnsi += qnsi;
realize.Remove(i); realize.Add(j);
ns = graph.newreach(realize, seed);
seed2 = graph.findlarge(ns, count);
cd2 = new CoordinateDescent(graph, bg, ns, seed2, cns, type, 10, alpha, mh);
List <double> cnsj = cd2.IterativeMinimize();
double qnsj = cd2.expectation();
sumnsj += qnsj;
realize.Add(i);
ns = graph.newreach(realize, seed);
seed2 = graph.findlarge(ns, count);
if (qnsi > qnsj)
{
cd2 = new CoordinateDescent(graph, bg, ns, seed2, cnsi, type, 10, alpha, mh);
}
else
{
cd2 = new CoordinateDescent(graph, bg, ns, seed2, cnsj, type, 10, alpha, mh);
}
List <double> cnsij = cd2.IterativeMinimize();
double qnsij = cd2.expectation();
sumnsij += qnsij;
}
cd1.PairMinimize(i, j, sumnsi - sumns, sumnsi - sumns, sumns + sumnsij - sumnsi - sumnsj, bp, left, right);
}
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Hyper_start = DateTime.Now;
seed1 = cd1.Realize();
List <int> nx = graph.newreach(seed1, seed);
List <int> s = graph.findlarge(nx, count);
CoordinateDescent cd = new CoordinateDescent(graph, bg, nx, s, 0.6667, type, 10, alpha, mh);
cd.IterativeMinimize();
List <double> prob = new List <double>();
for (int i = 0; i < nx.Count; i++)
{
prob.Add(cd.SeedProb(nx[i], cd.C[nx[i]]));
}
Tuple <double, double> results = icm.InfluenceSpread(graph, nx, prob, 200);
Hyper_end = DateTime.Now;
FileStream outfile = new FileStream(filepath + "_4o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + (b1 + b2) + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b1 += 2;
}
alpha += 0.0;
}
}
public static Tuple<List<int>, double> UnifiedCGreedy(Graph graph, Bipartite bg, List<int> Type, double c, double B, double alpha)
{
List<double> P = new List<double>();
for (int id = 0; id < Type.Count; ++id)
{
int type = Type[id];
double p;
if (type == 1)
p = c * c;
else if (type == 2)
p = c;
else
p = (2 - c) * c;
P.Add(p);
}
int k = (int)(B / c);
Tuple<List<int>, double> tup = bg.Greedy(k, P);
Console.WriteLine("seeds have been selected");
return tup;
}
public static Tuple<List<int>, double> UnifiedCGreedy(Graph graph, List<int> Type, double c, double B, double alpha)
{
List<double> P = new List<double>();
for (int id = 0; id < Type.Count; ++id)
{
int type = Type[id];
double p;
if (type == 1)
p = c * c;
else if (type == 2)
p = c;
else
p = (2 - c) * c;
P.Add(p);
}
ICModel icm = new ICModel(alpha);
int mH = GlobalVar.mH;
List<List<int>> RR = new List<List<int>>();
for (int r = 0; r < mH; ++r)
{
List<int> rSet = icm.RR(graph).ToList();
RR.Add(rSet);
}
Bipartite bg = new Bipartite(RR, graph.numV);
Console.WriteLine("Hyper-graph has been built");
int k = (int)(B / c);
Tuple<List<int>, double> tup = bg.Greedy(k, P);
Console.WriteLine("seeds have been selected");
return tup;
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader initial = new StreamReader(filepath + "_ini1000.txt");
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <int> initNodes = new List <int>();
for (int i = 0; i < 1000; i++)
{
initNodes.Add(int.Parse(initial.ReadLine()));
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 1.0; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 1.0)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
int b = 10;
while (b <= 50)
// Build a random hyper graph with mH random hyper edges.
{
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
int count = Convert.ToInt16(1.5 * Convert.ToDouble(b));
List <int> degrees = new List <int>();
for (int i = 0; i < 1000; i++)
{
degrees.Add(graph.adj[initNodes[i]].Count);
}
List <int> seedNodes = new List <int>();
while (count > 0)
{
int flag = 0;
for (int i = 1; i < 1000; i++)
{
if (degrees[i] > degrees[flag])
{
flag = i;
}
}
seedNodes.Add(initNodes[flag]);
degrees[flag] = -1;
count--;
}
CoordinateDescent cd = new CoordinateDescent(graph, bg, initNodes, seedNodes, 1 / 1.5, type, 50, alpha, mh);
List <double> allocation = cd.IterativeMinimize();
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
FileStream outfile = new FileStream(filepath + "_3l.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
List <double> prob = new List <double>();
for (int i = 0; i < initNodes.Count; i++)
{
prob.Add(cd.SeedProb(initNodes[i], allocation[initNodes[i]]));
}
Hyper_start = DateTime.Now;
Tuple <double, double> results = icm.InfluenceSpread(graph, initNodes, prob, 20000);
Hyper_end = DateTime.Now;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + b + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "\n");
writer.Flush();
writer.Close();
b += 10;
}
alpha += 0.2;
}
}
public static void CoordinateDescentAlgCommonHyperGraphOneAlpha(Graph graph)
{
StreamReader nodetype = new StreamReader(filepath + "_typeo.txt");
StreamReader threshold = new StreamReader(filepath + "_tu.txt");
List <double> thresh = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
thresh.Add(double.Parse(threshold.ReadLine()));
}
List <int> type = new List <int>();
for (int i = 0; i < graph.numV; i++)
{
int flag = int.Parse(nodetype.ReadLine());
if (flag == 0)
{
type.Add(0);
}
else if (flag == 1)
{
type.Add(2);
}
else
{
type.Add(1);
}
}
List <double> d = new List <double> {
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0
};
List <double> cu = new List <double>();
for (int i = 0; i < graph.numV; i++)
{
double t = thresh[i];
double p = 0.0;
if (type[i] == 1)
{
p = System.Math.Pow(t, 0.5);
}
else if (type[i] == 2)
{
p = t;
}
else
{
p = 1 - System.Math.Pow(1 - t, 0.5);
}
foreach (double point in d)
{
if (point < p)
{
continue;
}
cu.Add(point);
break;
}
}
int mh = 0;
if (filepath.Contains("Wiki"))
{
mh = 250000;
}
else if (filepath.Contains("CA"))
{
mh = 2000000;
}
else if (filepath.Contains("dblp"))
{
mh = 20000000;
}
else
{
mh = 40000000;
}
double alpha = 1.0; // Step of c of searching the best discount in th Unified Discount Algorithm
while (alpha <= 1.0)
{
Bipartite bg = new Bipartite(filepath, alpha, graph.numV);
double btotal = 30.0;
List <double> ratio = new List <double> {
1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0
};
while (btotal <= 30.0)
// Build a random hyper graph with mH random hyper edges.
{
foreach (double r in ratio)
{
double b1 = btotal / (1.0 + r);
StreamReader initial = new StreamReader(filepath + "_ini100.txt");
List <int> seed = new List <int>();
for (int i = 0; i < 100; i++)
{
seed.Add(int.Parse(initial.ReadLine()));
}
DateTime Hyper_start = DateTime.Now;
ICModel icm = new ICModel(alpha);
CoordinateDescent cd = new CoordinateDescent(graph, bg, seed, 0.0, type, 10, alpha, mh);
double b1used = 0.0;
while (b1used < b1)
{
int flag = seed[0];
double maxE = 0.0;
List <double> maxC = new List <double>();
foreach (int u in seed)
{
List <int> nrlist = graph.newreach(new List <int> {
u
}, cd.x);
double b2cub1 = r * cu[u];
int count = Convert.ToInt16(Math.Ceiling(1.5 * b2cub1));
double init_c = b2cub1 / Convert.ToDouble(count);
List <int> choose = graph.findlarge(nrlist, count);
foreach (int v in choose)
{
cd.ChangeAllocation(v, init_c);
}
cd.initNodes = nrlist;
cd.IterativeMinimize();
double nowE = cd.expectation();
if (nowE / cu[u] >= maxE / cu[flag])
{
flag = u;
maxE = nowE;
maxC = new List <double>();
for (int pos = 0; pos < cd.C.Count; pos++)
{
maxC.Add(cd.C[pos]);
}
}
foreach (int v in nrlist)
{
cd.ChangeAllocation(v, 0.0);
}
}
Console.WriteLine(cd.C.Sum());
b1used += cu[flag];
if (b1used > b1)
{
break;
}
seed.Remove(flag);
List <int> maxnr = graph.newreach(new List <int> {
flag
}, cd.x);
foreach (int u in maxnr)
{
cd.ChangeAllocation(u, maxC[u]);
cd.x.Add(u);
}
}
DateTime Hyper_end = DateTime.Now;
double Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
Hyper_start = DateTime.Now;
List <double> prob = new List <double>();
foreach (int u in cd.x)
{
prob.Add(cd.SeedProb(u, cd.C[u]));
}
Tuple <double, double> results = icm.InfluenceSpread(graph, cd.x, prob, 200);
Hyper_end = DateTime.Now;
FileStream outfile = new FileStream(filepath + "_6o.txt", FileMode.Append);
StreamWriter writer = new StreamWriter(outfile);
writer.Write("Choose time:" + Hyper_time + "\t");
Hyper_time = (Hyper_end - Hyper_start).TotalMilliseconds;
string mem = Convert.ToString(Process.GetCurrentProcess().WorkingSet64 / 8 / 1024 / 1024);
writer.Write("Propagation time:" + Hyper_time + "\t");
writer.Write("a:" + alpha + "\tb:" + btotal + "\tave:" + results.Item1 + "\tstd:" + results.Item2 + "\tmemory:" + mem + "ratio" + Convert.ToString(r) + "\n");
writer.Flush();
writer.Close();
}
btotal += 10.0;
}
alpha += 0.2;
}
}