Abstract
The Tutte polynomial of a graph is a 2-variable polynomial which is quite important in both Combinatorics and Statistical physics. It contains various numerical invariants and polynomial invariants, such as the number of spanning trees, the number of spanning forests, the number of acyclic orientations, the reliability polynomial, chromatic polynomial and flow polynomial. In this paper, we study and obtain a recursive formula for the Tutte polynomial of pseudofractal scale-free web (PSFW), and thus logarithmic complexity algorithm to calculate the Tutte polynomial of the PSFW is obtained, although it is NP-hard for general graph. By solving the recurrence relations derived from the Tutte polynomial, the rigorous solution for the number of spanning trees of the PSFW is obtained. Therefore, an alternative approach to determine explicitly the number of spanning trees of the PSFW is given. Furthermore, we analyze the all-terminal reliability of the PSFW and compare the results with those of the Sierpinski gasket which has the same number of nodes and edges as the PSFW. In contrast with the well-known conclusion that inhomogeneous networks (e.g., scale-free networks) are more robust than homogeneous networks (i.e., networks in which each node has approximately the same number of links) with respect to random deletion of nodes, the Sierpinski gasket (which is a homogeneous network), as our results show, is more robust than the PSFW (which is an inhomogeneous network) with respect to random edge failures.
Similar content being viewed by others
References
Tutte, W.T.: A ring in graph theory. Proc. Camb. Philos. Soc. 43, 26–40 (1947)
Tutte, W.T.: A contribution to the theory of chromatic polynomials. Can. J. Math. 6, 80–91 (1954)
Tutte, W.T.: On dichromatic polynomials. J. Comb. Theory 2, 301–320 (1967)
Ellis-Monaghan, J., Merino, C.: Graph polynomials and their applications I: the Tutte polynomial. In: Dehmer, M. (ed.) Structural Analysis of Complex Networks. http://arxiv.org/abs/0803.3079v2 (in press)
Chang, S.-C., Chen, L.-C., Yang, W.-S.: Spanning trees on the Sierpiński gasket. J. Stat. Phys. 126(3), 649–667 (2007)
Zhang, Z.Z., Wu, B., Lin, Y.: Counting spanning trees in a small-world Farey graph. Phys. A 391, 3342–3349 (2012)
Chang, S.-C., Chen, L.-C.: Number of connected spanning subgraphs on the Sierpiński gasket. Discrete Math. Theor. Comput. Sci. 11(1), 55–77 (2009)
Chang, S.-C., Chen, L.-C.: Spanning forests on the Sierpiński gasket. Discrete Math. Theor. Comput. Sci. 10(2), 55–76 (2008)
Chang, S.-C.: Acyclic orientations on the Sierpiński gasket. Int. J. Mod. Phys. B 05 (2010)
Chang, S.-C., Shrock, R.: Flow polynomials and their asymptotic limits for lattice strip graphs. J. Stat. Phys. 112, 815–879 (2003)
Chang, S.-C., Shrock, R.: Reliability polynomials and their asymptotic limits for families of graphs. J. Stat. Phys. 112, 1019–1077 (2003)
Rocek, M., Shrock, R., Tsai, S.H.: Chromatic polynomials for families of strip graphs and their asymptotic limits. Phys. A 252, 505–546 (1998)
Shrock, R., Xu, Y.: Chromatic polynomials of planar triangulations, the Tutte upper bound, and chromatic zeros. J. Phys. A Math. Theor. 45, 0552122012 (2012)
Welsh, D.J.A., Merino, C.: The Potts model and the Tutte polynomial. J. Math. Phys. 41(3), 1127–1152 (2000)
Shrock, R.: Exact Potts–Tutte polynomials for polygon chain graphs. J. Phys. A Math. Theor. 44, 145002 (2011)
Fortuin, C.M.: On the random-cluster model II. The percolation model. Physica 58(3), 393–418 (1972)
Oxley, J., Welsh, D.J.A.: The Tutte polynomial and percolation. In: Bondy, J.A., Murty, U.S.R. (eds.) Graph Theory and Related Topics (Proceeding of Conference, University of Waterloo, Waterloo, Ontario, 1977), pp. 329–339. Academic Press, New York, London (1979)
Jaeger, F., Vertigan, D.L., Welsh, D.J.A.: On the computational complexity of the Jones and Tutte polynomials. Math. Proc. Camb. Philos. Soc. 108, 35–53 (1990)
Oxley, J., Welsh, D.: Chromatic, flow and reliability polynomials: the complexity of their coefficients. Comb. Probab. Comput. 11(4), 403–426 (2002)
Donno, A., Iacono, D.: The Tutte polynomial of the Sierpinski and Hanoi graphs. Adv. Geom. 13(4), 663–694 (2013)
Chang, S.-C., Shrock, R.: Exact Potts model partition functions for strips of the honeycomb lattice. Phys. A 296, 183–233 (2001)
Chang, S.-C., Shrock, R.: Exact Potts model partition function on strips of the triangular lattice. Phys. A 286, 189–238 (2000)
Salas, J., Chang, S.-C., Shrock, R.: Exact Potts model partition function for strips of the square lattice. J. Stat. Phys. 107, 1207–1253 (2002)
Shrock, R.: Exact Potts Model Partition Functions on Ladder Graphs. Physica A 283, 388–446 (2000)
Barabási, A.-L., Albert, R.: Emergence of scaling in random networks. Science 286, 509–512 (1999)
Watts, D.J., Strogatz, H.: Collective dynamics of ’small-world’ networks. Nature (London) 393, 440–442 (1998)
Dorogovtsev, S.N., Goltsev, A.V., Mendes, J.F.F.: Pseudofractal scale-free web. Phys. Rev. E 65, 066122 (2002)
Zhang, Z.Z., Rong, L.L., Zhou, S.G.: A general geometric growth model for pseudofractal scale-free web. Phys. A 377, 329–339 (2007)
Zhang, Z.Z., Zhou, S.G., Chen, L.C.: Evolving pseudofractal networks. Eur. Phys. J. B. 58, 337–344 (2007)
Zhang, Z.Z., Liu, H.X., Wu, B., Zhou, S.G.: Enumeration of spanning trees in a pseudofractal scale-free web. Europhys. Lett. 90, 68002 (2010)
Zhang, Z.Z., Qi, Y., Zhou, S.G., Xie, W.L., Guan, J.H.: Exact solution for mean first-passage time on a pseudofractal scale-free web. Phys. Rev. E 79, 021127 (2009)
Albert, R., Jeong, H., Barabási, A.-L.: Error and attack tolerance of complex networks. Nature 406, 378–482 (2000)
Albert, R., Barabási, A.-L.: Statistical mechanics of complex networks. Rev. Mod. Phys. 74, 47–97 (2002)
Motter, A.E., Lai, Y.C.: Cascade-based attacks on complex networks. Phys. Rev. E 66, 065102 (2002)
Bollt, E.M., ben-Avraham, D.: What is special about diffusion in scale-free networks. New J. Phys. 7, 26 (2005)
Chang, S.-C., Chen, L.-C., Yang, W.-S.: Spanning trees on the Sierpiński gasket. J. Stat. Phys. 126, 649–667 (2007)
Mandelbrot, B.: The Fractal Geometry of Nature, pp. 193–195. Freeman, Francisco (1982)
Hilfer, R., Blumen, A.: Renormalisation on Sierpinski-type fractals. J. Phys. A Math. Gen. 17, L537–L545 (1984)
Callaway, D.S., Newman, M.E.J., Strogatz, S.H., Watts, D.J.: Network robustness and fragility: percolation on random graphs. Phys. Rev. Lett. 85(25), 5468–5471 (2000)
Cohen, R., Erez, K., ben-Avraham, D., Havlin, S.: Breakdown of the Internet under intentional attack. Phys. Rev. Lett. 86, 3682 (2001)
Zeng, A., Liu, W.P.: Enhancing network robustness against malicious attacks. Phys. Rev. E 85, 066130 (2012)
Acknowledgments
The authors are grateful to the anonymous referees for their valuable comments and suggestions. This work was supported by the scientific research program of Guangzhou municipal colleges and universities under Grant No. 2012A022 and the science and technology project of Guangzhou under Grant No. 2014J4100245.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1: A Proof of Theorem 1
In order to derive recursive formula of \(T_{n}(x,y)\), we must analyze the relation of spanning subgraphs between \(G(n+1)\) and its three subgraphs \(\Gamma _i(i=1,2,3)\) which are a copy of \(G(n)\) respectively.
For any \(H\in D_{n+1}\), it is a spanning subgraphs of \(G(n+1)\). Let
We find that \(H_i=(V_i,E_i)\) is a spanning subgraphs of \(\Gamma _i\) which is a copy of \(G(n)\), thus \(H_i \in D_n\), for \(i=1,2,3\).
On the contrary, given three spanning subgraphs \(H_1\), \(H_2\) and \(H_3\) of \(\Gamma _1, \Gamma _2\) and \(\Gamma _3\). Let
We find that \(H=(V, E)\) is a spanning subgraphs of \(G(n+1)\) and there exists a bijection between spanning subgraphs of \(G(n+1)\) and spanning subgraphs of \(\Gamma _1, \Gamma _2, \Gamma _3\) inside \(G(n+1)\). Therefore
where the relation between \(H=(V,E)\) and \(H_i=(V_i,E_i)\) (\(i=1,2,3\)) are shown in Eqs. (30) and (31). We denote this relation always holds when they appear in the following paragraphs.
We first derive the relations between \(\varPhi _{n+1}(H)\) and \(\varPhi _{n}(H_i)\) (\(i=1,2,3\)), and then the recursive formula of \(T_{n}(x,y)\).
1.1 Relations Between \(\varPhi _{n+1}(H)\) and \(\varPhi _{n}(H_i)\)
Now, we first derive recursive formula of \(r(G(n))\), and the relations between \(r(H)\) and \(r(H_i)\), \(n(H)\) and \(n(H_i)\). Then we can obtain relations between \(\varPhi _{n+1}(H)\) and \(\varPhi _{n}(H_i)\).
Because the PSFW is a connected graph, therefore \(k(G(n))=1\), for any \(n\ge 0\). Note that \(G(n+1)\) is composed of three copies of \(G(n)\) denoted as \(\Gamma _1\), \(\Gamma _2\), \(\Gamma _3\), we have
Thus
According to the second construction algorithm of the PSFW, we have
As for the relation between \(r(H)\) and \(r(H_i)\), \(n(H)\) and \(n(H_i)\), they can be divided into two cases. Thus relations between summation term of \(T_{n+1}(x,y)\) and that of \(T_n(x,y)\) can also be divided into two cases.
Case 1 The hub nodes of \(G({n+1})\) belong to the same connected component in the spanning subgraph \(H_i\) of \(G_i\), for any \(i=1, 2, 3\) (i.e., hub nodes \(A,C\) of \(G({n+1})\) belong to the same connected component in \(H_1\), hub nodes \(B, C\) of \(G({n+1})\) belong to the same connected component in \(H_2\), hub nodes \(A, B\) of \(G({n+1})\) belong to the same connected component in \(H_3\)).
According to the second construction algorithm of the PSFW, we have
Substituting \(|V(H)|\), \(|k(H)|\) from Eqs. (34) and (36) respectively in Eq. (1),
Substituting \(|E(H)|\), \(|r(H)|\) from Eqs. (35) and (37) respectively in Eq. (2),
Let Eq. (33) minus Eq. (37), we obtain
Hence
Case 2 The conditions for case \(1\) are not satisfied (i.e., for certain \(i=1, 2, 3\), the hub nodes of \(G({n+1})\) do not belong to the same connected component in the spanning subgraph \(H_i\)), we have
Substituting \(|V(H)|\), \(|k(H)|\) from Eqs. (34) and (39) respectively in Eq. (1), we have
Thus
and
Hence
1.2 Recursive Formulas of \(T_{1, n}(x,y)\) , \(T_{2, n}(x,y)\) and \(T_{3, n}(x,y)\)
Here, we proof \(T_{1,n}(x,y)\), \(T_{2,n}(x,y)\),\(T_{3,n}(x,y) \) satisfy the following recurrence relations:
with initial conditions
where \(T_{1,n}\), \(T_{2,n}\), \(T_{3,n} \) are the abbreviations of \(T_{1,n}(x,y)\), \(T_{2,n}(x,y)\), \(T_{3,n}(x,y)\) respectively.
The initial conditions are easy to be verified according to the definition of \(T_{1,n}(x,y)\), \(T_{2,n}(x,y)\) and \(T_{3,n}(x,y)\). As for the recurrence relations, the strategy of proof is to study all the possible structure of spanning subgraphs \(H\) of \(G(n+1)\) and analyze which kind of contribution they give to \(T_{1,n+1}(x,y)\), \(T_{2,n+1}(x,y)\) and \(T_{3,n+1}(x,y)\).
As defined in Sect. 3, the spanning subgraphs of \(G(n)\) has five different structures (i.e., \(D_{1,n}, D_{2,n}^A, D_{2,n}^B, D_{2,n}^C\), \(D_{3,n}\)), thus \(H_i\) also has five different structures, for any \(i=1, 2, 3\), and \(H\) has \(5^3\) kinds of structure.
In order to depict the relations between the structure of \(H_i\) and \(H\) clearly, we introduce five corresponding notations for the five different structures of \(H_i\) which are shown in Fig. 5. Each structure is denoted by a triangle whose three vertices denote the three hub nodes of \(G(n)\). The hub nodes which are in the same connected component are connected by a solid line, and the hub nodes which are not in the same connected component connected by a dotted line.
Now, we will study all the \(5^3\) kinds of structures of \(H\), and analyze which kind of contributions they give to \(T_{1,n+1}(x,y)\), \(T_{2,n+1}(x,y)\) and \(T_{3,n+1}(x,y)\).
First, we analyze the possible structures of \(H\) which give contribution to \(T_{1,n+1}(x,y)\). According to the definition of \(T_{1,n+1}(x,y)\), the condition \(H\in D_{1,n+1}\) holds. We find it has ten kinds of possible structures. The first three of them are shown in Fig. 6, the rest seven kinds of structures are shown in Fig. 7. For the first structure (left one of Fig. 6), \(H_i\in D_{1,n}\) for any \(i=1,2,3\), the hub nodes \(A, B\) of \(G_{n+1}\) are connected by a solid line in \(H_3\), the hub nodes \(A, C\) of \(G_{n+1}\) are connected by a solid line in \(H_1\), the hub nodes \(B, C\) of \(G_{n+1}\) are connected by a solid line in \(H_2\). Thus the conditions for the case \(1\) of Appendix 1 are satisfied. According to Eq. (38), we have
Thus, it contributes to \(T_{1,n+1}(x,y)\) by a term \((y-1)T_{1,n}^3\).
For the second structure (center one of Fig. 6), \(H_1\in D_{1,n}\), \(H_2\in D_{1,n}\) and \(H_3\in D_{2,n}^C\), the hub nodes \(A, B\) of \(G_{n+1}\) are connected by a solid line in \(H_3\), the hub nodes \(A, C\) of \(G_{n+1}\) are connected by a solid line in \(H_1\), the hub nodes \(B, C\) of \(G_{n+1}\) are connected by a solid line in \(H_2\). Thus the conditions for the case \(1\) of Appendix 1 are also satisfied. According to Eq. (38), we have
Noticing the possible rotations, we find it has \(3\) equivalent structures. Thus this kind of spanning subgraphs contributes to \(T_{1,n+1}(x,y)\) by a term \(3(y-1)T_{1,n}^2T_{2,n}\). Computing the contributions of all possible structures and adding them together, we obtain Eq. (42).
Next, we analyze the possible structures of \(H\) which give contribution to \(T_{2,n+1}(x,y)\). By symmetry, we only study \(T_{2,n+1}^C(x,y)\). Then the condition \(H\in D_{2,n+1}^C\) holds. We find it has \(6\) possible structures which are shown in Fig. 8. For the first structure, \(H_3\in D_{1,n}\), \(H_1\in D_{2,n}^B\) and \(H_2\in D_{2,n}^A\). Therefore the hub nodes \(A, C\) of \(G_{n+1}\) are connected by a dotted line in \(H_1\) which shows that they are not in the same connected component. Thus the conditions for case \(2\) are satisfied. According to Eq. (41), it contributes to \(T_{2,n+1}^C(x,y)\) by a term \(\frac{4}{x-1}T_{1,n}T_{2,n}^2\) (we have consider \(4\) equivalent structures: \(H_1\in D_{2,n}^C\) (or \(D_{2,n}^B\)), \(H_2\in D_{2,n}^A\) (or \(D_{2,n}^C\)) ).
Computing the contributions of all the \(6\) possible structures and adding them together, we obtain Eq. (43).
Finally, we analyze the possible structures of \(H\) which give contribution to \(T_{3,n+1}(x,y)\). Then the condition \(H\in D_{3,n+1}\) holds. We find it has \(4\) possible structures which is shown in Fig. 9. For the first structure, \(H_1\in D_{2,n}^B\) (or \(D_{2,n}^C)\), \(H_2\in D_{2,n}^A\) (or \(D_{2,n}^C\)), \(H_3\in D_{2,n}^A\) (or \(D_{2,n}^B)\). It contributes to \(T_{3,n+1}(x,y)\) by a term \(\frac{8}{x-1}T_{2,n}^3\) according to Eq. (41). Computing the contributions of all the \(4\) possible structures and adding them together, we obtain Eq. (44).
Appendix 2: Proof of Lemma 1
In this appendix, we prove the following result by mathematical induction.
For any \(n \ge 0\), \(T_{2,n}(x,y),T_{3,n}(x,y)\) can be factored as
where \(P_n(x,y)\), \(Q_n(x,y)\) are polynomials of \(x\) and \(y\).
Initial step: for \(n=0\), \(T_{2,0}(x,y)=x-1, T_{3,0}(x,y)=(x-1)^2\), let \(P_0(x,y)=Q_0(x,y)=1\), we know Eq. (47) holds.
Inductive step: assuming Eq. (47) holds for certain \(n \ge 0\), we prove Eq. (47) holds for \(n+1\) as follows.
For convince, we abbreviate \(P_{n}(x,y)\), \(Q_{n}(x,y)\) as \(P_{n}\) and \(Q_{n} \) respectively. Substituting \(T_{2,n}(x,y),T_{3,n}(x,y)\) from Eq. (47) in Eqs. (43) and (44), we have
and
Let
We find
Thus Eq. (47) holds for \(n+1\) which end the proof.
Appendix 3: Comparison of \(R_{s}(n)\) and \(R(n)\)
For the Sierpinski gasket \(SG(n)\) (\(n\ge 0\)), Alfredo Donno [20] found that the all-terminal reliability polynomial is given by
and \(T_{1,n}\left( 1,\frac{1}{1-p}\right) \) satisfy the following recurrence relation
with initial conditions
Let
We obtain the following recurrence relation from Eqs. (51), (52), (53) and (54).
with initial conditions
Now, we compare all-terminal reliability between the Sierpinski gasket and the PSFW while \(p\in (0,1)\).
For \(n=0\), the Sierpinski gasket and the PSFW have the same topology (i.e., a triangle). Thus
According to Eqs. (25), (55), (26) and (56), we have
But for \(n \ge 2\), we have
which is proved by mathematical induction as follows.
Initial step: for \(n=2\),
and
Thus Eq. (57) holds for \(n=2\).
Inductive step: assuming that Eq. (57) holds for certain \(n \ge 2\), we prove Eq. (57) holds for \(n+1\).
Let Eq. (55) minus Eq. (25) and Eq. (56) minus Eq. (26), we have
and
Thus Eq. (57) holds for \(n+1\) which end the proof.
Rights and permissions
About this article
Cite this article
Peng, J., Xiong, J. & Xu, G. Tutte Polynomial of Pseudofractal Scale-Free Web. J Stat Phys 159, 1196–1215 (2015). https://doi.org/10.1007/s10955-015-1225-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10955-015-1225-x