Wolfram Function Repository
Instant-use add-on functions for the Wolfram Language
Function Repository Resource:
Obtain partial probabilities of random walks on a directed graph
ResourceFunction["DirectedGraphTransferMatrix"][g] returns a matrix whose elements are partial probabilities describing random walks across directed graph g. | |
ResourceFunction["DirectedGraphTransferMatrix"][g,outs,ins] orders the rows and columns of the output matrix according to vertex lists outs and ins. |
Find the transfer matrix for a simple directed graph with one loop:
In[1]:= |
Out[1]= |
The probability for a random-walk to go from vertex 5 to vertex 3:
In[2]:= |
Out[2]= |
Determine output probabilities from input probabilities using Dot:
In[3]:= |
Out[3]= |
Label a binary tree with its output probabilities:
In[4]:= |
Out[4]= |
Adding edge weights to a graph changes the elements of the transfer matrix:
In[5]:= |
Out[5]= |
Label a binary tree with all of its partial probabilities:
In[6]:= |
Out[6]= |
Columns of the transfer matrix sum to 1:
In[7]:= |
Out[7]= |
The column sum property does not depend on edge weights:
In[8]:= |
Out[8]= |
Check the continuity constraint on each element of the transfer matrix:
In[9]:= |
Out[9]= |
On graphs with cycles, throughput vertices can have partial probabilities greater than one:
In[10]:= |
Out[10]= |
In[11]:= |
Out[11]= |
This means that, on average, the random walk will go through loops to visit the same vertex more than once:
In[12]:= |
Out[12]= |
Calculations may fail due to trapped cycles:
In[13]:= |
Out[13]= |
Adding an extra edge allows a positive result:
In[14]:= |
Out[14]= |
Find the probability of solving a logic maze through a random walk, without hitting a dead end:
In[15]:= |
Out[15]= |
Compare with brute force enumeration of 100,000 random walks:
In[16]:= |
Out[16]= |
Use WeightedAdjacencyGraph and DirectedGraphTransferMatrix to answer a Stack Exchange question:
In[17]:= |
Out[17]= |
This work is licensed under a Creative Commons Attribution 4.0 International License