public static void InsertVReward(string state, string stra, bool win)
{
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
int sno = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
int ano = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == stra).Single().ANO;
int times;
if (win) times = 1;
else times = 0;
double vs = whiteQlearning.QSTATEs.Where(p => p.SNO == sno && p.ANO == ano).Single().VALUE;
if (!whiteQlearning.VREWARDs.Any(p => p.SNO == sno && p.ANO == ano))
{
VREWARD tmps = new VREWARD();
tmps.SNO = sno;
tmps.ANO = ano;
tmps.TIMES = times;
tmps.TOTAL = 1;
tmps.REWARD = times;
whiteQlearning.VREWARDs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
}
else
{
VREWARD tmps = whiteQlearning.VREWARDs.Where(p => p.SNO == sno && p.ANO == ano).Single();
tmps.TIMES += times;
tmps.TOTAL++;
tmps.REWARD = vs * Convert.ToDouble(times) / tmps.TOTAL;
whiteQlearning.SubmitChanges();
}
}
public static void InsertStrategy(string strategy)
{
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
if (!whiteQlearning.ASTRATEGies.Any(p => p.STRATEGY == strategy))
{
ASTRATEGY tmps = new ASTRATEGY();
if (whiteQlearning.ASTRATEGies.Count() > 0) tmps.ANO = whiteQlearning.ASTRATEGies.Max(p => p.ANO) + 1;
else tmps.ANO = 0;
tmps.STRATEGY = strategy;
whiteQlearning.ASTRATEGies.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
}
}
public static void InsertState(string state)
{
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
if (!whiteQlearning.STATEs.Any(p => p.MSTATE == state))
{
STATE tmps = new STATE();
if (whiteQlearning.STATEs.Count() > 0) tmps.SNO = whiteQlearning.STATEs.Max(p => p.SNO) + 1;
else tmps.SNO = 0;
tmps.MSTATE = state;
whiteQlearning.STATEs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
}
}
public static void InsertQState(string state, string stra, double value)
{
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
int sno = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
int ano = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == stra).Single().ANO;
if (!whiteQlearning.QSTATEs.Any(p => p.SNO == sno && p.ANO == ano))
{
QSTATE tmps = new QSTATE();
tmps.SNO = sno;
tmps.ANO = ano;
tmps.VALUE = value;
whiteQlearning.QSTATEs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
}
else
{
whiteQlearning.QSTATEs.Where(p => p.SNO == sno && p.ANO == ano).Single().VALUE = value;
whiteQlearning.SubmitChanges();
}
}
int SelectWinTimes(string state, string strategy)
{
int times;
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
var tmp = from c in whiteQlearning.STATEs
from o in whiteQlearning.ASTRATEGies
from e in whiteQlearning.VREWARDs
where c.SNO == e.SNO && o.ANO == e.ANO && c.MSTATE == state && o.STRATEGY == strategy
select e.TIMES;
if (tmp.Count() <= 0)
{
VREWARD tmps = new VREWARD();
tmps.SNO = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
tmps.ANO = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == strategy).Single().ANO;
tmps.TIMES = 0;
tmps.REWARD = 1.0;
whiteQlearning.VREWARDs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
times = 0;
}
else times = tmp.Single();
return times;
}
public static double SelectReward(string state, string strategy, bool isKing)
{
double reward;
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
var tmp = from c in whiteQlearning.STATEs
from o in whiteQlearning.ASTRATEGies
from e in whiteQlearning.VREWARDs
where c.SNO == e.SNO && o.ANO == e.ANO && c.MSTATE == state && o.STRATEGY == strategy
select e.REWARD;
if (tmp.Count() <= 0)
{
VREWARD tmps = new VREWARD();
tmps.SNO = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
tmps.ANO = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == strategy).Single().ANO;
tmps.TIMES = 0;
if (isKing) tmps.REWARD = 100.0;
else tmps.REWARD = 1.0;
whiteQlearning.VREWARDs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
reward = tmps.REWARD;
}
else reward = tmp.Single();
return reward;
}
public static double SelectVState(string state)
{
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
var tmp = from c in whiteQlearning.STATEs
from e in whiteQlearning.QSTATEs
where c.MSTATE == state && c.SNO == e.SNO
select e.VALUE;
double maxQ = 0.0;
foreach (var c in tmp)
if (c > maxQ) maxQ = c;
return maxQ;
}
//public static void InsertProb(string state, string stra, string nstate)
//{
// WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
// int sno = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
// int ano = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == stra).Single().ANO;
// int nsno = whiteQlearning.STATEs.Where(p => p.MSTATE == nstate).Single().SNO;
// if (!whiteQlearning.PROBs.Any(p => p.SNO == sno && p.ANO == ano && p.NSNO == nsno))
// {
// PROB tmps = new PROB();
// tmps.SNO = sno;
// tmps.ANO = ano;
// tmps.NSNO = nsno;
// tmps.TIMES = 1;
// whiteQlearning.PROBs.InsertOnSubmit(tmps);
// whiteQlearning.SubmitChanges();
// }
// else
// {
// whiteQlearning.PROBs.Where(p => p.SNO == sno && p.ANO == ano && p.NSNO == nsno).Single().TIMES++;
// whiteQlearning.SubmitChanges();
// }
// }
public static double SelectQState(string state, string stra)
{
double value;
WhiteQlearningDataContext whiteQlearning = new WhiteQlearningDataContext();
int sno = whiteQlearning.STATEs.Where(p => p.MSTATE == state).Single().SNO;
int ano = whiteQlearning.ASTRATEGies.Where(p => p.STRATEGY == stra).Single().ANO;
var tmp = from c in whiteQlearning.QSTATEs
where c.SNO == sno && c.ANO == ano
select c.VALUE;
if (tmp.Count() <= 0)
{
QSTATE tmps = new QSTATE();
tmps.SNO = sno;
tmps.ANO = ano;
tmps.VALUE = 1.0;
whiteQlearning.QSTATEs.InsertOnSubmit(tmps);
whiteQlearning.SubmitChanges();
value = 1.0;
}
else value = tmp.Single();
return value;
}
