A matroid is, roughly speaking, a finite set together with a generalization of a concept from linear algebra that satisfies a natural set of properties for that concept. For example, the finite set could be the rows of a matrix, and the generalizing concept could be linear dependence and independence of any subset of rows of the matrix.
Formally, a matroid consists of a finite set of elements together with a family
of nonempty subsets
of
, called circuits,
which satisfy the axioms
1. No proper subset of a matroid circuit is a matroid circuit,
2. If and
,
then
contains a matroid
circuit.
(Harary 1994, p. 40).
An equivalent definition considers a matroid as a finite set of elements together with a family
of subsets of
, called independent sets, such that
1. The empty set is independent,
2. Every subset of an independent set is independent,
3. For every subset of
, all inclusion-maximal independent
sets contained in
have the same number of elements.
(Harary 1994, pp. 40-41).
For a finite matroid, a cyclic basis ordering is a cyclic order of its ground set in which every window of matroid rank length is a matroid basis. The Kajitani-Ueno-Miyano conjecture asserts that these orderings exist exactly for uniformly dense matroids (Kajitani et al. 1988).
Up to isomorphism, the numbers of simple matroids (or combinatorial geometries) with , 1, ... points are 1, 1, 1, 2, 4, 9, 26, 101, 950,
... (OEIS A002773), and the numbers of matroids
on
, 1, ... points are 1, 2, 4, 8, 17, 38, 98, 306, 1724,
... (OEIS A055545; Oxley 1993, p. 473).
(The value for
given by Oxley 1993, p. 42, is incorrect.)
A basic example is the uniform matroid ,
whose independent sets are precisely the subsets
of an
-element ground
set of size at most
. Parallel thickenings of these
matroids provide explicit counterexamples to the multiplicative Merino-Welsh
conjecture for matroids (Chen et al. 2026).
A Helly-type property concerns bases adapted to several matroid flats. A matroid basis is adapted to a matroid flat
when
is a matroid basis
of
. Jafari and Kaveh (2026) proved that a family
of matroid flats in a matroid of matroid
rank
has a common adapted basis iff each subfamily of at most
flats does. The number
is sharp for
. The result also allows
infinite ground sets and families when the rank is
finite. Jafari and Kaveh (2026) report AI assistance with the proof.
For an integer , a finite matroid is
-entropic if there is a family of random
variables indexed by its ground set, taking values
in a common
-element
alphabet, such that the joint Shannon entropy
for each subset equals its matroid
rank times
. Kalantari and Khazaei (2026,
Theorem 1.1) proved that the 5-entropic matroids are precisely those representable
over the finite field
.
Weighted negative dependence leads to the class of Rayleigh matroids. Every matroid of matroid rank or corank
at most 3 satisfies the stronger -Rayleigh inequalities for
(Zhao and Zhou 2026), where corank is the matroid
rank of the dual matroid.