2012-11-12

Two Ways To Understand Reachable Matrix

Last week we have learned how to use adjacency matrix to represent a social network. One question is why the power of matrix can indicate the reachability of social networks. Actually, the matrix mentioned in the class is named reachable matrix.

Let's see how it works. Here, I assume all of us have known the way to do matrix multiplication.

1. The First Way
One simple way to calculate reachable matrix is to regard it as logical operations,(And, Or, Not). We regard 0 and 1 represents false and true respectively.(you can also understand in this way: 0 represents two nodes are not connected; 1 represents two nodes are connected)

Another thing that should be declared is that in logic algebra, numeric operations of multiplication x*y, addition x + y are equal to the respective logical operations of conjunction x∧y, disjunction x∨y. The relevant logic operation rules are shown as below:

Let's review the example, here is a matrix:

Graph below is generated by matrix above: 
The two steps matrix within logic calculation:(which means the reachability going through two steps)


We can see that in X, n6 and n3 are not connected but in X^2 they are. So let's analyze the elements m'63(i=6, j=3).(using m for the elements in X matrix; using m' for the elements in X^2 matrix)

  • m'63

    m'63=(m61∧m13)∨(m62∧m23)∨(m63∧m33)∨(m64∧m43)∨(m65∧m53)∨(m66∧m63)
    We have know that n6 is not connected with n3 within one step, So if n6 is connected with n3 within two steps, there must be at least one node nx which connects n3 and is connected by n6.

    Now look at the calculation process. It means if n6 n1 are connected and at the same time n1 n3 are connected; OR n6 n2 are connected and at the same time n2 n3 are connected;OR....., then n6 n3 are connected.

    In this process, only (m62∧m23) is equal to 1, others are all zero. So the path should be 6->2->3.
The thinking way is the same with the situation of more steps.

2. The Second Way
Another way to think this is to multiple matrix in normal way. With the foundation above, we can deepen our understanding.

The two steps matrix within normal way calculation:

With the knowledge in FIRST PART, we have understood the meaning of non-zero elements. Now let's analyze the elements m'16 and m'22

  • m'16
    m'16=m11*m16+m12*m26+m13*m36+m14*m46+m15*m56+m16*m66
    We can see that only :
    m12*m26 =1
    m15*m56 =1
    Others are all equal to zero. So 
    m'16=m12*m26+m15*m56=2
    And you must find that 2 represents that there are two paths which make n1 n6 connected. Obviously, the paths can be 1->2->6 and 1->5->6

  • m'22
    It is easy to find that two paths are 2->3->2 and 2->6->2

The thinking way is the same with the situation of more steps.

If you have any question, pls feel free to discuss with me.
Hope you enjoy this!

13 条评论:

  1. The two methods to explain the Reachable Matrix are so interesting, expecially the first one, which different with the professor's way, discussed in the class. You're are so clever to figure out the alternative way to analyse the matrix. Using your methods, I can understand the matrix multiplication easily.

    回复删除
  2. This blog really do helpful for me to understand the Reachable Matrix.

    回复删除
  3. The article has analyzed the Reachable Matrix explicitly and answered the questions mentioned in class perfectly. I am very sorry to say that I am confused the expressions in the first picture and the relationship between them and elements in the matrix. Could you explain it in a clear way? Moreover, actually, the main idea of the double ways is the same thing, although the methods are a little bit different. To hold the core thought, we could understand the process smoothly.

    回复删除
  4. Your explanation about the two ways to calculate the Reachable Matrix is so impressive and helps us a lot to solve similar problems.To make a good social network analysis, it's important to do matrix multiplication well!

    回复删除
  5. Your explanation about Matrix multiplication is excellent. In the class, you also give us a good suggestion about the meaning of that. You are so smart! One more thing really puzzled me is that why you use logical operation to illustrate the problem. In X^2, every element is either 1 or 0. However, in X^3 or a higher power matrix, the element can be more than 1. So I think logical operations are less useful than normal operation, + or *.

    回复删除
  6. Cool~Thank you for sharing the explanation about matrix multiplication.But, I am confused about the logical operation that one or one equal one plus one, it should equal to two,not one, so why the answer is one in your post?

    回复删除
  7. Hi, Luo. I'm amazed by the way you solved the question. You do really got the whole understanding of relationships between the matrix and sociograms. But I agree with Cui Helei's idea that in the matrix produced by your logical operation, we can't see how many walks on earth between one node and the other, because the "0" and "1" in this matrix only indicate yes or no. It's not very intuitive. As the result in your second way, it shows different numbers with different power which indicating the total walks. But anyway, it's still a good way to know if one node can reach the other node with the steps as its powers.

    回复删除
  8. Thanks. It lets me have a clear concept about calculating reachability.

    回复删除
  9. As a fundamental factor of SNA arithmetic, reachable matrix plays a significant role in SNA structure. You give me a clear way to understand the two ways of reachable matrix. Besides, I really want to discuss with you about some contents which I do not understand deeply.

    回复删除
  10. This is a good essay illustrating the relationship between matrix and the sociograms. This helps me figure out how to create the diagram from a matrix by knowing the mechanism behind.

    回复删除
  11. These two methods you mentioned about the reachable matrix is really impressive. Actually, since we mainly focused on social psychology in the early stage, this reachable matrix shows me a new way to analyze the social network in quantitative way. This arouse my interest in learning social network again and thank you for sharing these two excellent way to understand the reachable matrix.

    回复删除
  12. Actually, I don't understand how you could get your matrix (power 2) from method 1 if it's using computer... Can you explain more?

    回复删除
  13. In the class to learned SNA theory and use adjacency matrix to represent a social network.Prof Rosana had proposed a very basic question. However I really forgot the detail way to do matrix multiplication. The two methods to calculate reachable matrix have recalled me a lot the knowledge we studied in the university.Your method is so clever to calculate the reachable matrix. The mathematical background is really important to support the SNA theory. I need to review something...

    回复删除