added mpi randomized values

Random/MPI.py

+#Uses MPI to dispatch random number generation on multiple nodes.
+#files are saved into .npy files which can be loaded later with numpy
+from mpi4py import MPI
+from generators import *
+from datetime import datetime
+
+comm = MPI.COMM_WORLD
+size = comm.Get_size() # we suppose > 1
+rank = comm.Get_rank()
+
+#Wrapper for outside call
+def MPIBatchRandomNormalized(generator,elmPerNode):
+    name = type(generator).__name__
+    N = elmPerNode
+
+    if rank == 0:
+        print("processing "+name+"...")
+
+    #set seed with rank offset
+    generator.seed(rank + datetime.now().timestamp())
+    randomNumbers = generator.batchRandomNormalized(N)
+    
+    #receives all the results. 
+    gathered = comm.gather(data, root=0)
+    if rank==0:
+        data = np.array()
+        for l in gathered:
+            data += np.array(l)
+        print("Saving to "+name+".npz ...")
+        with open(name+".npy", "wb") as f:
+            np.save(f, data)
+
+if __name__ == "__main__":
+    N = 1000000
+    generators = [ParkMiller(), KnuthLewis(), Marsaglia()
+                  , LavauxJenssens(), Haynes(), MitchelMoore()
+                  , MersenneTwister(), BlumBlumShub()]
+    
+    for e in generators
+        MPIBatchRandomNormalized(e,N)

Random/generators.py

@@ -4,6 +4,7 @@ import random
 
 class RandomNumberGenerator:
     def __init__(self):
+        self.max = 0 #for normalization
         return NotImplementedError
 
     def random(self):
@@ -28,7 +29,7 @@ class RandomNumberGenerator:
         """
         return n normalized values
         """
-        return [self.randomNorm() for i in range(n)]
+        return [self.random() for i in range(n)] / self.max
 
 
 class LCG(RandomNumberGenerator):
@@ -42,6 +43,7 @@ class LCG(RandomNumberGenerator):
         self.c = c
         self.a = a
         self.x = 0  # default seed
+        self.max = self.m + 1
 
     def random(self):
         self.x = (self.a * self.x + self.c) % self.m
@@ -51,7 +53,7 @@ class LCG(RandomNumberGenerator):
         self.x = seed
 
     def randomNorm(self):
-        return self.random() / (self.m + 1)
+        return self.random() / self.max
 
 
 class ParkMiller(LCG):
@@ -110,6 +112,7 @@ class LaggedFibonnaci(RandomNumberGenerator):
         self.m = m
         self.ParkMiller = ParkMiller()
         self.values = []
+        self.max = self.m + 1
 
     def seed(self, seed):
         self.ParkMiller.seed(seed)
@@ -124,7 +127,7 @@ class LaggedFibonnaci(RandomNumberGenerator):
         return self.values[-1]
 
     def randomNorm(self):
-        return self.random() / (self.m + 1)
+        return self.random() / self.max 
 
 
 class MitchelMoore(LaggedFibonnaci):
@@ -164,6 +167,7 @@ class CNS(RandomNumberGenerator):
         super().__init__()
         self.M = M
         self.x = x0  # must verify pgcd(M,x0) = 1
+        self.max = M + 1
 
     def seed(self, seed):
         self.x = seed
@@ -173,7 +177,7 @@ class CNS(RandomNumberGenerator):
         return self.x
 
     def randomNorm(self):
-        return self.random() / (self.M + 1)
+        return self.random() / self.max
 
 
 class BlumBlumShub(CNS):

Random/main.py

@@ -2,30 +2,43 @@ from generators import *
 from datetime import datetime
 import matplotlib.pyplot as plt
 import numpy as np
-
+import seaborn as sns
 
 if __name__=="__main__":
 
+    ##INIT PARAMS##
+    N = 1000000
+    RESOLUTION = 100
+    #set seaborn params
+    sns.set_theme(style="darkgrid")
+    
+    #bandwidth for smooth density
+    #https://en.wikipedia.org/wiki/Kernel_density_estimation#Bandwidth_selection
+    BW = 0.337
+    
+    #init generators
     generators = [ParkMiller(), KnuthLewis(), Marsaglia()
                   , LavauxJenssens(), Haynes(), MitchelMoore()
                   , MersenneTwister(), BlumBlumShub()]
     
+    #init seed on timesystem 
     seed = datetime.now().timestamp()
-    N = 1000000
-    RESOLUTION = 100
-
+    
+    #generate each diagrams for each random func
     for e in generators:
         name = type(e).__name__
+        
         print("processing "+name+"...")
         e.seed(seed)
         res = e.batchRandomNormalized(N)
+        results.append(results)
+        
         print("building histograms..") 
-        #matplotlib prep
-        #see https://matplotlib.org/stable/gallery/statistics/histogram_features.html#sphx-glr-gallery-statistics-histogram-features-py
         fig, ax = plt.subplots()
-        n, bins, patches = ax.hist(res, RESOLUTION, density=True)
+        sns.displot(res, kde=True)
         ax.set_xlabel('Value')
         ax.set_ylabel('Probability density')
         ax.set_title(type(e).__name__)
-        #plt.show()
         plt.savefig(type(e).__name__+'.png')
+
+    #generate diagrams with all randoms