Interests: Incidence Geometry, Combinatorics, Spectral Graph Theory, and Polynomial Methods.
Inspired by a famous open problem posed by Jacques Tits on existence of semi-finite generalized polygons, for which no progress has been made so far in the case of generalized hexagons, we solve a specific case of an easier version of the problem where non-existence is proved after assuming that the generalized hexagon contains a particular subhexagon. We show that (a) no generalized hexagon contains as a full proper subgeometry, (b) every near hexagon containing is a finite generalized hexagon, and hence isomorphic to or the triality hexagon via the classification result of Cohen and Tits.
The following computer code constructs Table 1 and 2, and verifies Lemma 3.1: GSplit2.g.
Bart and I discovered a new near octagon corresponding to the finite simple group in July 2014. Here we give its construction, prove several of its properties, and find a “Suzuki tower of near polygons” corresponding to the Suzuki tower of finite simple groups. We also give geometric constructions of some well known strongly regular graphs using this new near octagon and its subgeometries. Moreover, we construct another new near octagon as a subgeometry, related to the group .
We prove a generalization of a result of Alon and Füredi, and show how this elementary result on zeros of polynomials is connected to various other results in Coding Theory, Finite Geometry and Theoretical Computer Science. This in combination with the earlier work of Clark, Forrow and Schmitt suggest that much like Combinatorial Nullstellensatz, the Alon-Füredi Theorem is a fundamental result on polynomials with connections to various important topics in mathematics. Here is a video of a talk I gave on this paper to a general scientific audience.
We prove uniqueness results for the near polygons lying in the Suzuki tower, that we described in . In particular, we prove a characterization of the Hall-Janko near octagon as the unique near octagon of order containing the dual split Cayley hexagon as a subgeometry. The following computer codes construct Table 1, 2, 3 and 4, and verify Lemmas 4.12, 5.1, 5.2, 5.3, 5.4: Suz1.g, Suz2.g, HallJankoHyp.g. Also see this for an independent verification by Bart.
We show that the dual split Cayley generalized hexagon does not have any distance- ovoids (which are equivalent to exact hitting sets in the corresponding -regular -uniform hypergraph on vertices), and that the Ree-Tits octagon does not have any distance- ovoids, thus resolving the last remaining case for existence of distance- ovoids in known finite generalized octagons. The computational techniques we use, which combine Knuth’s Dancing Links, Linton’s SmallestImageSet, Integer Linear Programming, along with a nice trick involving subgeometries, might be useful in small cases of other finite geometrical problems that have a similar flavour. Our complete computer code is available here.
We extend the work done in  by proving that there is no semi-finite hexagon containing any of the known generalized hexagons of order and as full subgeometry. Moreover, we show that the split Cayley hexagon is not contained in any generalized hexagon as a full subgeometry. The code in main.g constructs computer models of small generalized hexagons that we use in our computations and the code in main.sage performs all the computations mentioned in Section 4.
We give an alternate direct construction of one of the near octagons discovered in  using the projective special linear group . This construction is used to derive geometric and group theoretic properties of this near octagon, and we propose a new family of near octagons to which both this near octagon and the near octagon discovered in  belong. So far, these are the only two nontrivial members of this family that we know (we define and classify the trivial ones in the paper).
Using the expander mixing lemma, we prove the first non-trivial upper bound on the size of a minimal t-fold blocking set in a finite projective plane. This generalises a classical result of Bruen and Thas. The techniques used also give new proofs of some old and new results in finite geometry. We talk about the sharpness of our result, some constructions, and generalizations.
The cage problem asks for the smallest graph with a given degree and girth . The number of vertices in the smallest graph is denoted by , and graphs meeting this bound are called cages. The only known non-trivial infinite families of cages are the generalized polygons, which tell us that for , whenever is a prime power. We improve the known upper bounds on and for infinitely many values of by constructing new small regular induced subgraphs of known generalized quadrangles and hexagons. We also give a lower bound on the smallest number of vertices in a regular induced subgraph of a generalized polygon, using the expander mixing lemma, which gives a limit to our methods in improving the upper bounds on cage numbers.
 A generalization of the theorems of Chevalley-Warning and Ax-Katz via polynomial substitutions. With Ioulia N. Baoulina and Pete L. Clark. To appear in Proceedings of the AMS. arXiv.
We prove a new generalization of the classical Chevalley-Warning theorem, Ax-Katz theorem and several related results.
 Non-intersecting Ryser hypergraphs, with Valentina Pepe. arXiv
Ryser’s conjecture is a longstanding open problem in combinatorics, which states that every -partite, -uniform hypergraph has a vertex cover of size at most times the matching number of the hypergraph. For , this is just Konig’s theorem, and besides that value of the conjecture is only known to be true for . In recent years, extremal hypergraphs with respect to this conjecture, known as Ryser hypergraphs, have been studied extensively. We contribute to their study by giving new infinite families of non-intersecting Ryser hypergraphs, for uniformity at least , that cannot be built by taking disjoint copies of intersecting Ryser hypergraphs (and possibly adding some new edges), as was known to be true for -uniform hypergraphs. In this sense, it is the first really non-intersecting infinite family of Ryser hypergraphs.
- On semi-finite hexagons of order (2,t) containing a subhexagon of order 2, at the Fourth Irsee Conference (2014), slides.
- The Alon-Füredi bound, at the British Combinatorial Conference 2015, slides.
- Computing hyperplanes of near polygons, at COCOA 2015, slides.
- On zeros of a polynomial in a finite grid: the Alon-Füredi bound, at Combinatorics 2016 and at Discrete Mathematics Days 2016, slides.
- Zeros of polynomials over a finite grid, at the polynomial method workshop in HUJI organised by Jordan Ellenberg and Gil Kalai, 2016.
- Zeros of polynomials over a finite grid, at Caltech combinatorics seminar and UCLA combinatorics seminar, 2017.
- Expander mixing lemma in finite geometry, at UWA combinatorics seminar, 2017.
- The cage problem and finite geometry, at Fq13, 2017.
- Minimal multiple blocking sets, at Irsee, 2017, slides.
- Extremal problems in finite geometry, at BMS-BGSMath Junior Meeting 2017, Institut d’Estudis Catalans, Barcelona, slides.
- Improved bounds on cage numbers, at Combinatorics 2018, Facets of Complexity (colloquium talk), and the Berlin-Poznań-Hamburg Seminar.
- Extremal and near extremal constructions for Ryser’s conjecture, at the 1st Southwestern German Workshop on Graph Theory, 2018.