添加EID，唉

calc_avg_tr.py

@@ -0,0 +1,24 @@
+from island.match import Match
+from island.matches import Matches
+import numpy as np
+import csv
+
+
+def calc(mode):
+    matches = Matches.from_profile_expr(lambda a: mode in a)
+    trs = 0
+    acts = 0
+    for m in matches.data:
+        maxr = int(m.query('game', 'created').first()['info']['game_end_at'])
+        for r in range(1, maxr+1):
+            rows = m.query('action', 'done').where(lambda x: x['rno'] == r).raw_data
+            for row in rows:
+                trs += row['tr']
+            acts += len(rows)
+    return trs, acts, trs / acts
+
+
+if __name__ == '__main__':
+    print(calc('CLASSIC'))
+    print(calc('SURVIVE'))
+

calc_coopr_rate.py

@@ -0,0 +1,35 @@
+from island.match import Match
+from island.matches import Matches
+import numpy as np
+import csv
+
+
+def calcAndSave(mode):
+    matches = Matches.from_profile(mode)
+    times = np.zeros(15)
+    cs = np.zeros(15)
+    for m in matches.data:
+        maxr = int(m.query('game', 'created').first()['info']['game_end_at'])
+        for r in range(1, maxr+1):
+            rows = m.query('action', 'done').where(lambda x: x['rno'] == r).raw_data
+            coop = 0
+            for row in rows:
+                if row['act_a'] == 'C':
+                    coop += 1
+                if row['act_b'] == 'C':
+                    coop += 1
+            if rows:
+                times[r-1] += len(rows) * 2
+                cs[r-1] += coop
+    for i in range(15):
+        if(times[i] == 0):
+            times[i] = 1
+    cs /= times
+    with open("outputs/CR_%s.csv"%mode, 'w') as f:
+        csv.writer(f).writerow(cs)    
+
+if __name__ == '__main__':
+    calcAndSave('CLASSIC')
+    calcAndSave('SURVIVE')
+
+

coopr_per_round.py

@@ -1,50 +0,0 @@
-import json
-from matplotlib import pyplot as plt
-from island.match import Match
-from island.matches import Matches
-import numpy as np
-
-
-matches = Matches.from_profile_expr(lambda r: 'CLASSIC' in r)
-max_round = 15
-
-coopr = []
-yerr_min = []
-yerr_max = []
-x = np.arange(1, max_round+1)
-bx = []
-
-survivals = {}
-with open('survivals.json','r') as f:
-    survivals = json.load(f)
-
-for i in range(max_round):
-    co = []
-    for j in range(len(matches.data)):
-        coop = 0
-        rows = matches.data[j].query('action', 'done').where(lambda x: x['rno']==i+1).raw_data
-        for row in rows:
-            if row['act_a'] == 'C' and row['act_b'] == 'C':
-                coop += 1
-
-        if rows:
-            co.append(float(coop) / float(len(rows)))
-
-    bx.append(co)
-
-    if co:
-        coopr.append(sum(co) / len(co))
-
-        yerr_min.append(coopr[-1] - min(co))
-        yerr_max.append(max(co) - coopr[-1])
-        print("%f, %f, %f"%(yerr_min[-1], yerr_max[-1], coopr[-1]))
-    else:
-        coopr.append(0)
-
-
-plt.figure()
-# plt.errorbar(x, coopr, yerr=[yerr_min, yerr_max], fmt='o', capsize=4)
-# plt.boxplot(bx, showmeans=True, meanline=True)
-plt.plot(x, coopr)
-plt.show()
-# plt.savefig('graph/co_per_round.png')

eid_calc.py

@@ -0,0 +1,168 @@
+"""
+计算E_{i,D}
+E_{i,D}=所有邻居的剩余时间 - (背叛对象的剩余时间 + 与背叛对象的上轮交互时间)
+       =除去背叛对象的剩余时间 - 背叛对象的上一轮交互时间
+输出：
+1. json格式(详细)
+{
+    "GID": {
+        "RID": {
+            "PID": EID,
+        }
+    }
+}
+2. CSV格式(按轮平均值, CLASSIC/SURVIVE分别对应一个文件)
+"""
+import csv
+import json
+from functools import reduce
+import numpy as np
+from island.match import Match
+from island.matches import Matches
+
+
+class Eid:
+    def __init__(self):
+        self.survivals = {}
+        with open('outputs/survivals.json','r') as f:
+            self.survivals = json.load(f)
+        self.neighbors = {}
+        with open('outputs/neighborhood.json', 'r') as f:
+            self.neighbors = json.load(f)
+        self.seasonSurvive = Matches.from_profile('SURVIVE')
+        self.seasonClassic = Matches.from_profile('CLASSIC')
+
+
+    def getNeighborTR(self, m, r, p, s, d):
+        """
+        获取该玩家所有邻居中，除去背叛对象的剩余时间
+        
+        :param m: Match
+        :param r: Round ID
+        :param p: PID
+        :param s: Survivals list(neighborhood)
+        :param d: list of victim
+        :returns: E_{i}
+        """
+        truenb = [i for i in s if i not in d]
+        trs = dict.fromkeys(truenb, 1440)
+        req = m.query('action', 'request').where(lambda x: x['rno'] == r and x['from'] in truenb and x['to'] != p)
+        for d in req.raw_data:
+            trs[d['from']] -= d['tr']
+
+        for d in m.query('action', 'approve').where(lambda x: x['rno'] == r and x['from'] in truenb and x['to'] != p).raw_data:
+            trs[d['from']] -= d['tr']
+
+        for d in m.query('action', 'cancel').where(lambda x: x['rno'] == r and x['from'] in truenb and x['to'] != p).raw_data:
+            trs[d['from']] += req.where(lambda x: x['from'] == d['from'] and x['to'] == d['to'] and x['log_id'] < d['log_id']).orderby('log_id').raw_data[-1]['tr']
+
+        for d in m.query('action', 'deny').where(lambda x: x['rno'] == r and x['to'] in truenb and x['from'] != p).raw_data:
+            trs[d['to']] += req.where(lambda x: x['from'] == d['to'] and x['to'] == d['from'] and x['log_id'] < d['log_id']).orderby('log_id').raw_data[-1]['tr']
+
+        return reduce(lambda x,y: x + max(0,min(y,1440)), trs.values(), 0)
+
+    def getNeighborhood(self, m, r, p):
+        """
+        获取该玩家当轮存活邻居
+        
+        :param m: Match
+        :param r: Round ID
+        :param p: PID
+        :returns: Survivals list(neighborhood)
+        """
+        if str(p) not in self.neighbors[m.name]:
+            print("Alone(%d)!" % p)
+            return []
+        return [i for i in self.survivals[m.name][str(r)] if i in self.neighbors[m.name][str(p)]]
+
+    def getVictims(self, m, r, p):
+        """
+        获取该玩家当轮的背叛对象，以及与他们的博弈时间资源之和
+
+        :param m: Match
+        :param r: Round ID
+        :param p: PID
+        :returns: sum of tr and a list of victim
+        """
+        victim = []
+        ans = 0
+        for d in m.query('action', 'done').where(lambda x: x['rno'] == r and (x['a'] == p or x['b'] == p)).raw_data:
+            if d['a'] == p and d['act_a'] == 'D':
+                victim.append(d['b'])
+                ans += d['tr']
+            elif d['b'] == p and d['act_b'] == 'D':
+                victim.append(d['a'])
+                ans += d['tr']
+        return (ans, victim)
+
+    def getVictims2(self, m, r, p):
+        """
+        获取该玩家当轮的背叛对象，以及与他们的博弈时间资源之和
+        ## 悲观假设背叛上一轮所有博弈对象
+
+        :param m: Match
+        :param r: Round ID
+        :param p: PID
+        :returns: sum of tr and a list of victim
+        """
+        victim = []
+        ans = 0
+        for d in m.query('action', 'done').where(lambda x: x['rno'] == r-1 and (x['a'] == p or x['b'] == p)).raw_data:
+            victim.append(d['b'] if d['a'] == p else d['a'])
+            ans += d['tr']
+        return (ans, victim)
+
+    def calcRoundData(self, m, r):
+        """
+        计算某场比赛，某一轮的Eid值
+        
+        :param m: Match
+        :param r: Round ID
+        :param p: PID
+        :returns: an average value and a detail dict
+        """
+        ans = {}
+        sigma = 0.0
+        for p in self.survivals[m.name][str(r)]:
+            u, v = self.getVictims2(m, r+1, p)
+            e = max(0, min(1440, self.getNeighborTR(m, r, p, self.getNeighborhood(m, r, p), v))) - u
+            ans[p] = e
+            sigma += e
+        return (sigma, ans)
+
+    def calc(self):
+        detail = {}
+        avgC = np.zeros(15)
+        avgS = np.zeros(15)
+        cnt = np.zeros(15)
+        for m in self.seasonClassic.data:
+            d = {}
+            maxr = int(m.query('game', 'created').first()['info']['game_end_at'])
+            for r in range(1, maxr+1):
+                sigma, ans = self.calcRoundData(m, r)
+                d[r] = ans
+                avgC[r-1] += sigma
+                cnt[r-1] += len(ans)
+            detail[m.name] = d
+        avgC /= cnt
+        cnt = np.zeros(15)
+        for m in self.seasonSurvive.data:
+            d = {}
+            maxr = int(m.query('game', 'created').first()['info']['game_end_at'])
+            for r in range(1, maxr+1):
+                sigma, ans = self.calcRoundData(m, r)
+                d[r] = ans
+                avgS[r-1] += sigma
+                cnt[r-1] += len(ans)
+            detail[m.name] = d
+        avgS /= cnt
+        with open('outputs/EID_CLASSIC.csv', 'w') as f:
+            csv.writer(f).writerow(avgC)
+        with open('outputs/EID_SURVIVE.csv', 'w') as f:
+            csv.writer(f).writerow(avgS)
+        with open('outputs/EID_detail.json', 'w') as f:
+            json.dump(detail, f)
+        
+if __name__ == '__main__':
+    e = Eid()
+    e.calc()

eid_plot.py

@@ -0,0 +1,99 @@
+import numpy as np
+from matplotlib import pyplot as plt
+import scipy as sp
+from scipy.stats import pearsonr
+from matplotlib import markers
+blue = '#0984e3'
+red = '#d63031'
+
+def error(f,x,y):
+    return sp.sum((f(x)-y)**2)
+def p1(x, coopr, e, postfix, show=True):
+    fig = plt.figure(figsize=(4.5, 3))
+    ax = fig.gca()
+    ax.plot(x, coopr, color=blue, linewidth=2, label="$f_c$")
+    ax.set_ylim(0, 1)
+    ax.set_yticks(sp.linspace(0, 1, 5))
+    ax2 = ax.twinx()
+    ax2.plot(x, e, color=red, linewidth=2, label=r"$E_{i,D}$")
+    # ax2.set_ylim(0, 1440)
+    # ax2.set_yticks(sp.linspace(0, 1440, 5))
+    ax2.tick_params(labelsize=18)
+    ax.tick_params(labelsize=18)
+    ax.set_xlim(1, 15)
+    ax.set_xlabel("Rounds", size=22)
+    ax.set_ylabel(r"$f_c$", family='sans-serif', color=blue, size=22)
+    ax.tick_params(axis='y', labelcolor=blue)
+    ax2.set_ylabel(r"$E_{i,D}$", family='sans-serif', color=red, size=22)
+    ax2.tick_params(axis='y', labelcolor=red)
+
+    plt.tight_layout()
+    if show:
+        plt.show()
+    else:
+        plt.savefig("graph/eid_co_plot_%s.eps" % postfix)
+
+def p2(e, coopr, postfix, show=True):
+    # p2散点图
+    fig = plt.figure(figsize=(4, 3))
+    ax = fig.gca()
+    fp1,residuals,rank,sv,rcond = sp.polyfit(e, coopr, 1, full=True)
+    print("残差：",residuals)
+    print('Model parameter:',fp1)
+    print("Other parameters: rank=%s, sv=%s, rcond=%s"%(str(rank), str(sv), str(rcond)))
+    f1 = sp.poly1d(fp1)
+    print("error= %f" % error(f1, e, coopr))
+    fx = sp.linspace(np.min(e), np.max(e), 2)
+
+    plt.plot(fx,f1(fx),linewidth=2,color=red, ls='--', zorder=0)
+
+
+    plt.scatter(e, coopr, color='white', edgecolors=blue, linewidths=2, zorder=101)
+    ax.set_xlabel(r'$E_{i,D}$', family='sans-serif', size=20)
+    ax.set_ylabel(r'$f_{c}$', family='sans-serif', size=20)
+    # ax.set_xlim(0, 1440)
+    # ax.set_xticks(sp.linspace(0, 1440, 5))
+    ax.tick_params(labelsize=14)
+    ax.set_ylim(0.5, 1)
+    ax.set_yticks(sp.linspace(0.5, 1, 5))
+    plt.tight_layout()
+    if show:
+        plt.show()
+    else:
+        plt.savefig("graph/eid_co_sca_%s.eps" % postfix)
+
+    # 皮尔逊相关系数
+    print("pearson: %f, p-value: %f" % pearsonr(e, coopr))
+
+if __name__ == '__main__':
+    mode = 'CLASSIC'
+    mode = 'SURVIVE'
+
+    show = False
+    show = True
+
+    coopr = np.loadtxt("outputs/CR_%s.csv"%mode, delimiter=',')
+    x = np.arange(1, 16)
+    e = np.loadtxt("outputs/EID_%s.csv"%mode, delimiter=',')
+
+    p1(x,coopr, e, mode, show)
+    # p2(e, coopr, mode, show)
+    
+
+
+    """
+SURVIVE
+残差： [ 0.10769174]
+Model parameter: [  1.33390242e-04   7.29761895e-01]
+Other parameters: rank=2, sv=[ 1.37254846  0.34075025], rcond=3.33066907388e-15
+error= 0.107692
+pearson: 0.709599, p-value: 0.003045
+
+
+CLASSIC
+残差： [ 0.00992721]
+Model parameter: [  3.14661037e-05   7.35284315e-01]
+Other parameters: rank=2, sv=[ 1.41231835  0.07319068], rcond=3.33066907388e-15
+error= 0.009927
+pearson: 0.309326, p-value: 0.261915
+    """

gen_neighborhood.py

@@ -0,0 +1,25 @@
+import json
+from island.matches import Matches
+
+
+matches = Matches.from_profile_expr(lambda r:True)
+
+neighbors = {}
+
+for i in range(len(matches.data)):
+    m = matches.data[i]
+    n = {}
+    for r in m.query('neighbor', 'create').raw_data:
+        if r['a'] in n:
+            n[r['a']].append(r['b'])
+        else:
+            n[r['a']] = [r['b']]
+
+        if r['b'] in n:
+            n[r['b']].append(r['a'])
+        else:
+            n[r['b']] = [r['a']]    
+    neighbors[matches.names[i]] = n
+
+with open('outputs/neighborhood.json', 'w') as f:
+    json.dump(neighbors, f)

outputs/CR_CLASSIC.csv

@@ -0,0 +1 @@
+0.771505376344,0.8,0.825980392157,0.848258706468,0.828431372549,0.829383886256,0.837320574163,0.847087378641,0.842592592593,0.842723004695,0.820754716981,0.80487804878,0.783018867925,0.766666666667,0.788461538462

outputs/CR_SURVIVE.csv

@@ -0,0 +1 @@
+0.820512820513,0.784722222222,0.842465753425,0.879581151832,0.939024390244,0.949324324324,0.97265625,0.995833333333,0.987288135593,0.973214285714,0.972602739726,0.872549019608,0.518518518519,0.857142857143,1.0

outputs/EID_CLASSIC.csv

@@ -0,0 +1 @@
+180.0,-41.8604651163,-124.186046512,-69.7674418605,-44.6511627907,-164.651162791,-69.7674418605,-92.0930232558,-97.6744186047,-101.860465116,-209.032258065,-213.71024735,-104.939271255,-165.755395683,-649.411764706

outputs/EID_SURVIVE.csv

@@ -0,0 +1 @@
+-58.293956044,-788.740365112,-833.236533958,-1329.43820225,-1278.73417722,-1369.30909091,-1370.60240964,-1343.59832636,-1427.42358079,-1392.89719626,-1440.0,-1440.0,-1390.34482759,-1440.0,-1152.0

outputs/winner.json

@@ -0,0 +1 @@
+{"G153": [], "G203": [], "G254": [], "G272": [3170], "G285": [], "G295": [], "G299": [3558, 3559, 3563, 3575], "G302": [3649], "G307": [3896, 3901], "G318": [3965, 3967, 3970, 3973], "G337": [], "G341": [4462, 4463, 4464, 4476], "G349": [4558, 4559], "G354": [4614, 4615, 4616, 4618, 4623, 4624], "G54": [], "G68": [], "G72": [], "G75": [], "G86": []}

tau_f_co.py

@@ -33,7 +33,7 @@ def p1(x, coopr, tau, postfix, show=True):
     ax2.set_ylabel(r"$\tau_{f}$", family='sans-serif', color=red, size=22)
     ax2.tick_params(axis='y', labelcolor=red)
 
-    fig.legend(fontsize=18)
+    # fig.legend(fontsize=18)
     plt.tight_layout()
     if show:
         plt.show()
@@ -154,7 +154,8 @@ if __name__ == '__main__':
                         print("[%s(%d)] alone: %d" % (matches.names[j], i+1, k))
                     else:
                         t = 1440 * len(neighbors[matches.names[j]][k])
-                    tp.append(t if t < 1440 else 1440)
+                    # tp.append(t if t < 1440 else 1440)
+                    tp.append(t)
                     mwTau["%s-%d"%(j,i)] = tp[-1]
             else:
                 if str(i) not in survivals[matches.names[j]]:
@@ -167,8 +168,9 @@ if __name__ == '__main__':
                         trs = matches.data[j].get_tr(i, k, neighbors[matches.names[j]][k], survivals[matches.names[j]][str(i)])
                         for l in neighbors[matches.names[j]][k]:
                             if l in trs and trs[l] > 0:
-                                t += trs[l]
+                                t += trs[l] if trs[l] < 1440 else 1440
                     tp.append(t if t < 1440 else 1440)
+                    # tp.append(t)
                     mwTau["%s-%d"%(j,i)] = tp[-1]
 
         if tp:
@@ -182,10 +184,10 @@ if __name__ == '__main__':
 
     # p1折线图
     # p1(x, coopr, tau, 'classic', False)
-    # p1(x, coopr, tau, 'survive', False)
+    # p1(x, coopr, tau, 'survive', True)
 
-    p2(tau[:12], coopr[:12], 720, 'survive', False)
+    p2(tau[:12], coopr[:12], 1322, 'survive', True)
-    # p2(tau, coopr, 720, 'classic', False)
+    # p2(tau, coopr, 1399, 'classic', True)
 
 
     '''