Cell array with adjacents elements by a “from” vector and a “to” vector

I have 2 vectors, A and B of equal dimension. Each value of A indicates the source element of the element given by its position. Each value of B indicates the target element of the element given by its position. Some values of A and B are NaN, but they have the same indices (they occupy the same positions in A and B). Some values of A and B may be repeated.

For example:

A=[NaN NaN NaN NaN NaN NaN 3 2 4 4 1 5 ... ]; 
B=[NaN NaN NaN NaN NaN NaN 1 6 2 8 9 1 ... ];

I would like to get a cell array, C, that is an adjacency list of these elements. That is, a cell array where each cell is a vector with the target elements, and its position is the source element. In this case it would be like this:

source () -> target (find())

3 -> 7 ()

2 -> 8 ()

4 -> 9 ()

4 -> 10 ()

1 -> 11 ()

5 -> 12 ()

On the other hand:

source (find()) -> target ()

7 () -> 1

8 () -> 6

9 () -> 2

10 () -> 8

11 () -> 9

12 () -> 1

Therefore:

C={[11], [8], [7], [9, 10], [12], [1], [6], [2], [8], [9], [1], ...};

I have managed to do it with “for”, but A and B are very large and it takes too long to calculate. Could it be done in batch?

Thanks in advance

Answer 1

[Steven Lord]

What are you hoping to do with this list? Depending on the specific operations you're looking to perform you may want to create a graph or digraph from your list of sources and targets and then use the functions available for Graph and Network Algorithms to manipulate the graph or digraph.

[Sergio Rojas Blanco]

Hello, thanks for your response Steven Lord . The elements of A and B are roads. I will use the adjacency list of roads to know the possible paths a vehicle can take when on a given road.

[Steven Lord]

So the shortestpath or allpaths functions?

[Sergio Rojas Blanco]

I just need to know the paths that a vehicle can take knowing its position.

Also, I have a list of adjacencies of traffic lights

[Steven Lord]

A=[NaN NaN NaN NaN NaN NaN 3 2 4 4 1 5];
source = A(~isnan(A))

source = 1x6

3 2 4 4 1 5

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

target = find(~isnan(A))

target = 1x6

7 8 9 10 11 12

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

D = graph(source, target)

D = 
  graph with properties:

    Edges: [6x1 table]
    Nodes: [12x0 table]

plot(D)

dist = distances(D, 4)

dist = 1x12

Inf Inf Inf 0 Inf Inf Inf Inf 1 1 Inf Inf

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

reachable = find(isfinite(dist))

reachable = 1x3

4 9 10

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

So you can get from 4 to 4, 9, or 10.

allpaths(D, 4, 10)

ans = 1x1 cell array
    {[4 10]}

In this simple case the list of all paths is short. Let's add a few more edges.

D = addedge(D, [4, 10], 8)

D = 
  graph with properties:

    Edges: [8x1 table]
    Nodes: [12x0 table]

plot(D)

Now there are more ways to get from 4 to 10.

allpaths(D, 4, 10)

ans = 2x1 cell array
    {[4 8 10]}
    {[  4 10]}

[Sergio Rojas Blanco]

Stephen, thank you very much for your answer, but I must consider the vector B to obtain all the adjacency relations.

I use VISSIM to get network information. A road (link) can be a normal link or a connector. A connector joins 2 adjacent links. VISSIM provides information about the source link (A), the destination link (B) and whether it is a connector.

It seems like it could be done with findgroups and splitapply, but I can't figure it out.