Ask Your Question

Pairs of graphs with same spectral radius but with different diameter and different number of edges

asked 2018-04-12 00:47:21 -0500

Deepak Sarma gravatar image

I know that graphs.cospectral_graphs() gives all pairs of graphs with same eigenvalues. But I need graphs with same spectral radius i.e. largest eigenvalue (the other eigenvalues may be different) and differing in diameter and number of edges. What to do for this? Thank you in advance.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2018-04-15 11:29:30 -0500

tmonteil gravatar image

updated 2018-04-15 12:38:22 -0500

Here is a quick and dirty no-brain method:

First, we can iterate over all connected graphs of a given size up to isomorphism, see:

sage: graphs.nauty_geng?

Regarding the spectral radius, note that the spectral radius method is numerical, so there will be some noise preventing us to detect graphs with the same eigenvalue. So, we will use the largest eigenvalue of the adjacency matrix, wich is an exact algebraic number:

sage: G = graphs.RandomGNP(10,1/2)
sage: G.spectral_radius()
(4.885867917434971, 4.885867917912396)
sage: max(G.adjacency_matrix().eigenvalues())

So, let us pick a not-so-small class of connected graphs, say graphs with 8 vertices. And let us classify each of them according to their spectral radius : we make a (lazy) dictionary that maps each spectral radius to the list of graphs with that spectral radius:

sage: from collections import defaultdict
sage: d = defaultdict(list
sage: for G in graphs.nauty_geng('8c'):
....:     d[max(G.adjacency_matrix().eigenvalues())] += [G]

We can see that there are 10350 possible eigenvalues:

sage: len(d)

So let us restrict to the eigenvalues with more than one corresponding graph:

sage: dd = {i:d[i] for i in d if len(d[i])>1}
sage: len(dd)

Nicer. Just to get an insight, let us look at the maximal classes:

sage: max(len(dd[i]) for i in dd)
sage: ddd = {i:d[i] for i in d if len(d[i])==23}
sage: len(ddd)

So, there is exactly one eigenvalue for which there are 23 graphs with that spectral radius. Let us inspect it further:

The eigenvalue is:

sage: e = next(iter(ddd)) ; e
sage: e.radical_expression()
sqrt(5) + 1

The corresponding list of graphs is:

sage: L = ddd[e]

Now we can see if there are good candidates:

sage: for G in L:
....:     print(G.diameter(), G.num_edges()) 
(+Infinity, 8)
(+Infinity, 10)
(+Infinity, 9)
(+Infinity, 9)
(3, 11)
(4, 11)
(+Infinity, 9)
(+Infinity, 9)
(+Infinity, 10)
(3, 12)
(+Infinity, 11)
(+Infinity, 10)
(3, 12)
(+Infinity, 11)
(3, 11)
(3, 12)
(5, 11)
(3, 12)
(3, 12)
(+Infinity, 10)
(3, 12)
(+Infinity, 11)
(4, 12)

We can see that with the same spectral radius e there are graphs on 8 vertices with diameter and number of edges (3,11) and (4,12) respectively:

sage: G1,G2 = L[4], L[-1]
sage: G1.diameter()
sage: G2.diameter()
sage: G1.num_edges()
sage: G2.num_edges()
sage: G1.spectral_radius()
(3.2360679773376506, 3.2360679776151673)
sage: G2.spectral_radius()
(3.236067977288281, 3.2360679776071106)

Here are the pictures of the candidates:



We are lucky ! If no such example were found, we would have to inspect all elements of dd.

EDIT Actualy i did a typo in the use of graphs.nauty_geng('8c'), which explains why we got disconnected graphs (infinite diameter). The correct syntax is graphs.nauty_geng('8 -c'). Instead we get 9820 different spectral radius, 926 spectral radius with at least two graphs, the spectral radius with the largest number of graphs is 1/2*sqrt(17) + 3/2, and there are 12 such graphs. In that list, the pairs (diameter, num_edges) is (3,12), (4,12), (3,13), (4,13), (2,14), (3,14).

edit flag offensive delete link more


Thank you for the detail explanation. But what I see here we will get the graphs with same spectral radius and then we have check ourselves for graphs with different diameter and no of edges. But what to do if I want to write the program in such a way that it checks everything itself and finally give me a pair (or all such pairs) of my requirement?

Deepak Sarma gravatar imageDeepak Sarma ( 2018-04-16 22:46:54 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools


Asked: 2018-04-12 00:47:21 -0500

Seen: 45 times

Last updated: Apr 15