Graph product

In graph theory, a graph product is a binary operation on graphs. Specifically, it is an operation that takes two graphs $G 1$ and $G 2$ and produces a graph $H$ with the following properties:

The vertex set of $H$ is the Cartesian product $V (G 1) \times V (G 2)$ , where $V (G 1)$ and $V (G 2)$ are the vertex sets of $G 1$ and $G 2$ , respectively.
Two vertices $(a 1, a 2)$ and $(b 1, b 2)$ of $H$ are connected by an edge, iff a condition about $a 1, b 1$ in $G 1$ and $a 2, b 2$ in $G 2$ is fulfilled.

The graph products differ in what exactly this condition is. It is always about whether or not the vertices $a n, b n$ in $G n$ are equal or connected by an edge.

The terminology and notation for specific graph products in the literature varies quite a lot; even if the following may be considered somewhat standard, readers are advised to check what definition a particular author uses for a graph product, especially in older texts.

Even for more standard definitions, it is not always consistent in the literature how to handle self-loops. The formulas below for the number of edges in a product also may fail when including self-loops. For example, the tensor product of a single vertex self-loop with itself is another single vertex self-loop with $E = 1$ , and not $E = 2$ as the formula $E_{G \times H} = 2 E_{G} E_{H}$ would suggest.

Overview table

The following table shows the most common graph products, with $\sim$ denoting "is connected by an edge to", and $\sim̸$ denoting non-adjacency. While $\sim̸$ does allow equality, $≄$ means they must be distinct and non-adjacent. The operator symbols listed here are by no means standard, especially in older papers.

Name	Condition for $(a_{1}, a_{2}) \sim (b_{1}, b_{2})$		Number of edges $\begin{matrix} v_{1} = \| V (G_{1}) \| & v_{2} = \| V (G_{2}) \| \\ e_{1} = \| E (G_{1}) \| & e_{2} = \| E (G_{2}) \| \end{matrix}$	Example
Name		with $a_{n} rel b_{n}$ abbreviated as ${rel}_{n}$		Example
Cartesian product (box product) $G_{1} ◻ G_{2}$	$a_{1} = b_{1} \land a_{2} \sim b_{2}$ $\lor$ $a_{1} \sim b_{1} \land a_{2} = b_{2}$	$=_{1} \sim_{2}$ $\lor$ $\sim_{1} =_{2}$	$v_{1} e_{2} + e_{1} v_{2}$
Tensor product (Kronecker product, categorical product) $G_{1} \times G_{2}$	$a_{1} \sim b_{1} \land a_{2} \sim b_{2}$	$\sim_{1} \sim_{2}$	$2 e_{1} e_{2}$
Strong product (Normal product, AND product) $G_{1} ⊠ G_{2}$ $= (G_{1} \times G_{2}) \cup (G_{1} ◻ G_{2})$	$a_{1} = b_{1} \land a_{2} \sim b_{2}$ $\lor$ $a_{1} \sim b_{1} \land a_{2} = b_{2}$ $\lor$ $a_{1} \sim b_{1} \land a_{2} \sim b_{2}$	$=_{1} \sim_{2}$ $\lor$ $\sim_{1} =_{2}$ $\lor$ $\sim_{1} \sim_{2}$	$v_{1} e_{2} + e_{1} v_{2} + 2 e_{1} e_{2}$
Lexicographical product $G_{1} \cdot G_{2}$ or $G_{1} [G_{2}]$	$a_{1} \sim b_{1}$ $\lor$ $a_{1} = b_{1} \land a_{2} \sim b_{2}$	$\sim_{1}$ $\lor$ $=_{1} \sim_{2}$	$v_{1} e_{2} + e_{1} v_{2}^{2}$
Co-normal product (disjunctive product, OR product) $G_{1} * G_{2}$ or $G_{1} \lor G_{2}$ $= G_{1} [G_{2}] \cup G_{2} [G_{1}]$ $= \overline{\overline{G_{1}} ⊠ \overline{G_{2}}}$	$a_{1} \sim b_{1}$ $\lor$ $a_{2} \sim b_{2}$	$\sim_{1}$ $\lor$ $\sim_{2}$	$v_{1}^{2} e_{2} + e_{1} v_{2}^{2} - 2 e_{1} e_{2}$
Modular product $G_{1} ⋄ G_{2}$ $= (G_{1} ⊠ G_{2}) \cup (\overline{G_{1}} \times \overline{G_{2}})$	$a_{1} \sim b_{1} \land a_{2} \sim b_{2}$ $\lor$ $a_{1} ≄ b_{1} \land a_{2} ≄ b_{2}$	$\sim_{1} \sim_{2}$ $\lor$ $≃_{1} ≃_{2}$
Rooted product	see article		$v_{1} e_{2} + e_{1}$
Zig-zag product	see article		see article	see article
Replacement product
Homomorphic product $G_{1} ⋉ G_{2}$	$a_{1} = b_{1}$ $\lor$ $a_{1} \sim b_{1} \land a_{2} \sim̸ b_{2}$	$=_{1}$ $\lor$ $\sim_{1} \sim_{2}$	$v_{1} v_{2} (v_{2} - 1) / 2 + e_{1} (v_{2}^{2} - 2 e_{2})$

In general, a graph product is determined by any condition for $(a_{1}, a_{2}) \sim (b_{1}, b_{2})$ that can be expressed in terms of $a_{n} = b_{n}$ and $a_{n} \sim b_{n}$ .

Mnemonic

Let $K_{2}$ be the complete graph on two vertices (i.e. a single edge). The product graphs $K_{2} ◻ K_{2}$ , $K_{2} \times K_{2}$ , and $K_{2} ⊠ K_{2}$ look exactly like the graph representing the operator. For example, $K_{2} ◻ K_{2}$ is a four cycle (a square) and $K_{2} ⊠ K_{2}$ is the complete graph on four vertices.

The $G_{1} [G_{2}]$ notation for lexicographic product serves as a reminder that this product is not commutative. The resulting graph looks like substituting a copy of $G_{2}$ for every vertex of $G_{1}$ .

Notes

↑ Graph Products Revisited: Tight Approximation Hardness of Induced Matching, Poset Dimension and More, Parinya Chalermsook, Bundit Laekhanukit, Danupon Nanongkai, 2012
1 2 Roberson, David E.; Mancinska, Laura (2012). "Graph Homomorphisms for Quantum Players". Journal of Combinatorial Theory, Series B. 118: 228–267. arXiv:1212.1724. doi:10.1016/j.jctb.2015.12.009.
↑ Bačík, R.; Mahajan, S. (1995). "Semidefinite programming and its applications to NP problems". Computing and Combinatorics. Lecture Notes in Computer Science. Vol. 959. p. 566. doi:10.1007/BFb0030878. ISBN 978-3-540-60216-3.
↑ The hom-product of is the graph complement of the homomorphic product of.

References

Imrich, Wilfried; Klavžar, Sandi (2000). Product Graphs: Structure and Recognition. Wiley. ISBN 978-0-471-37039-0.

[1] Graph Products Revisited: Tight Approximation Hardness of Induced Matching, Poset Dimension and More, Parinya Chalermsook, Bundit Laekhanukit, Danupon Nanongkai, 2012

[rm12-2] 1 2 Roberson, David E.; Mancinska, Laura (2012). "Graph Homomorphisms for Quantum Players". Journal of Combinatorial Theory, Series B. 118: 228–267. arXiv:1212.1724. doi:10.1016/j.jctb.2015.12.009.

[3] Bačík, R.; Mahajan, S. (1995). "Semidefinite programming and its applications to NP problems". Computing and Combinatorics. Lecture Notes in Computer Science. Vol. 959. p. 566. doi:10.1007/BFb0030878. ISBN 978-3-540-60216-3.

[4] The hom-product of is the graph complement of the homomorphic product of.

Graph product

Overview table

Mnemonic

See also

Notes

References