This paper is a survey of recent progress in graph theory that used a weak but useful form of connectivity called sublinear expansion. In plain terms, an expander is a network that is well connected despite having relatively few edges. Sublinear expansion is a weaker version of that idea. Over the last decade it has helped resolve several long-standing questions about sparse graphs — those with far fewer edges than the number of possible connections.

The author, Richard Montgomery, collects and explains recent results where researchers found or worked inside subgraphs that have these weaker expansion properties. The paper recalls the basic idea of an α-expander: any not-too-large set of vertices has a comparatively large set of neighbours outside it. For very sparse graphs one cannot hope for a fixed positive α. Komlós and Szemerédi introduced a more flexible notion, the (ε,k)-expander, which lets the required neighbourhood size shrink slowly as the set grows — roughly losing a factor of about log^2 in the size scale.

Montgomery describes how this notion has been applied. One major area is subdivisions, where edges of a small graph are replaced by paths inside a larger graph. Classical results show that a graph with large enough average degree must contain subdivided complete graphs. Work by Bollobás and Thomason, and by Komlós and Szemerédi, led to the conclusion that the threshold density d(t) for forcing a subdivision of K_t grows on the order of t^2. The survey quotes concrete bounds from the literature: an upper bound of the form (1023+o(1)) t^2 and a lower-bound construction of the form (964+o(1)) t^2, with a conjecture that the lower bound may be closer to the true value.

The survey also outlines many other problems where sublinear expansion plays a central role. These include questions about graph minors, the presence and packing of cycles, cycle lengths in sparse graphs, rainbow cycles (where edges have distinct colours), and problems in Ramsey theory. The common theme is that passing to a subgraph with guaranteed (weak) expansion gives enough connectivity to carry out constructions and proofs that were hard by other means.