FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3141567130> ?p ?o ?g. }

• W3141567130 abstract "Sidorenko's conjecture for blow-ups, Discrete Analysis 2021:2, 13 pp. Let $G$ be a bipartite graph with finite vertex sets $X$ and $Y$. If $G$ has density $alpha$, then the average degree of the vertices in $X$ is $alpha|Y|$, so the mean-square degree is at least $alpha^2|Y|^2$. This is easily seen to be equivalent to the statement that if two vertices $y_1,y_2$ are selected independently and uniformly at random from $Y$, then the average number of neighbours they have in common in $X$ is at least $alpha^2|X|$. It follows that the mean-square number of common neighbours is at least $alpha^4|X|^2$, and this translates into the statement that if $(x_1,x_2,y_1,y_2)$ is a random element of $X^2times Y^2$, then the probability that all four pairs $x_iy_j$ are edges of $G$ is at least $alpha^4$. This can be interpreted as saying that the 4-cycle density of a graph $G$ is always at least as large as it would be for a typical random graph of the same density. The above argument can be straightforwardly generalized from 4-cycles to general complete bipartite graphs $K_{r,s}$. The famous Sidorenko conjecture (also asked by Erdős and Simonovits) states that the conclusion holds for _all_ bipartite graphs. That is, if $H$ is a bipartite graph with vertex sets $U$ and $V$ of size $r$ and $s$, $G$ is as above, and $phi:Uto X$ and $psi:Vto Y$ are random functions, then the probability that $phi(u)psi(v)$ is an edge of $G$ for every $uvin E(H)$ is at least $alpha^{|E(H)|}$. One might think that this conjecture would either have a simple counterexample or would follow fairly easily from known inequalities, but this appears not to be the case. A good way to get a feel for the difficulty is to try to prove it for paths of length 3. It is known to be true in this case, but a certain amount of ingenuity is required to prove it. In fact, it is known to be true for quite a large class of bipartite graphs, with results typically taking the form that a bipartite graph satisfies the conjecture if it can be built up in a certain way from graphs that belong to a particularly simple class. The achievement of this paper is that it proves the conjecture for a class of graphs that _cannot_ be built up in that way. In other words, it obtains a proof that is not just a refinement of existing methods but a genuine extension of the available techniques. A particular class of graphs for which they obtain a positive result is that of _blow-ups_. These are bipartite graphs obtained as follows: let $H$ be a bipartite graph with vertex sets $A$ and $B$ and let $p$ be a positive integer, take $p$ disjoint copies $H_1,dots,H_p$ of $H$ and identify corresponding vertices that belong to $A$. (For example, if $H$ is a single edge, then the blow-up is the complete bipartite graph $K_{1,p}$ -- that is, a star with $p$ edges.) The authors show that for every bipartite graph $H$ there is some $p$ such that the blow-up satisfies Sidorenko's conjecture. The rough idea behind the proof is that blow-ups have a property that allows one to replace the graph by a simpler graph to which known techniques apply. Perhaps the simplest case for which Sidorenko's conjecture is still unknown is the graph obtained from the complete bipartite graph $K_{5,5}$ by removing a 10-cycle. While the authors do not resolve this question, their result does imply that the square of this graph (that is, the blow-up when $p=2$) satisfies the conjecture. Another consequence of their results is that for every bipartite graph $H$ there is a bipartite graph $H'$ such that the disjoint union of $H$ and $H'$ satisfies the conjecture. So is Sidorenko's conjecture true? The experts on the problem are reluctant to take a strong view either way. The results proved so far place significant restrictions on what a counterexample could be like (though these leave open relatively small possibilities such as the $K_{5,5}setminus C_{10}$ example), which perhaps points in the positive direction. On the other hand, a natural analogue of the conjecture for 3-uniform hypergraphs, which places a restriction on what a proof could be like -- it cannot be too generalizable. This paper is a valuable addition to our understanding of a tantalizing question. <iframe width=560 height=315 src=https://www.youtube.com/embed/D2NeNSdUICo title=YouTube video player frameborder=0 allow=accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture allowfullscreen></iframe>" @default.
• W3141567130 created "2021-04-13" @default.
• W3141567130 creator A5042722711 @default.
• W3141567130 creator A5070979968 @default.
• W3141567130 date "2021-03-30" @default.
• W3141567130 modified "2023-09-27" @default.
• W3141567130 doi "https://doi.org/10.19086/da.21472" @default.
• W3141567130 hasPublicationYear "2021" @default.
• W3141567130 type Work @default.
• W3141567130 sameAs 3141567130 @default.
• W3141567130 citedByCount "5" @default.
• W3141567130 countsByYear W31415671302018 @default.
• W3141567130 countsByYear W31415671302021 @default.
• W3141567130 countsByYear W31415671302023 @default.
• W3141567130 crossrefType "journal-issue" @default.
• W3141567130 hasAuthorship W3141567130A5042722711 @default.
• W3141567130 hasAuthorship W3141567130A5070979968 @default.
• W3141567130 hasBestOaLocation W31415671301 @default.
• W3141567130 hasConcept C114614502 @default.
• W3141567130 hasConcept C118615104 @default.
• W3141567130 hasConcept C132525143 @default.
• W3141567130 hasConcept C197657726 @default.
• W3141567130 hasConcept C2780990831 @default.
• W3141567130 hasConcept C33923547 @default.
• W3141567130 hasConcept C80899671 @default.
• W3141567130 hasConceptScore W3141567130C114614502 @default.
• W3141567130 hasConceptScore W3141567130C118615104 @default.
• W3141567130 hasConceptScore W3141567130C132525143 @default.
• W3141567130 hasConceptScore W3141567130C197657726 @default.
• W3141567130 hasConceptScore W3141567130C2780990831 @default.
• W3141567130 hasConceptScore W3141567130C33923547 @default.
• W3141567130 hasConceptScore W3141567130C80899671 @default.
• W3141567130 hasLocation W31415671301 @default.
• W3141567130 hasLocation W31415671302 @default.
• W3141567130 hasOpenAccess W3141567130 @default.
• W3141567130 hasPrimaryLocation W31415671301 @default.
• W3141567130 hasRelatedWork W2990042548 @default.
• W3141567130 hasRelatedWork W3040670847 @default.
• W3141567130 hasRelatedWork W3201366099 @default.
• W3141567130 hasRelatedWork W4285056899 @default.
• W3141567130 hasRelatedWork W4286978554 @default.
• W3141567130 hasRelatedWork W4297676365 @default.
• W3141567130 hasRelatedWork W4301344670 @default.
• W3141567130 hasRelatedWork W4318975185 @default.
• W3141567130 hasRelatedWork W755952011 @default.
• W3141567130 hasRelatedWork W810375382 @default.
• W3141567130 hasVolume "2021" @default.
• W3141567130 isParatext "false" @default.
• W3141567130 isRetracted "false" @default.
• W3141567130 magId "3141567130" @default.
• W3141567130 workType "paratext" @default.