For edge weights , form the out-LaplacianThe directed matrix-tree theorem states that the cofactor , obtained by deleting row and column , equals the sum of over directed spanning trees oriented towards .
Expand the determinant by permutations, and in each diagonal entry expand the sum of outgoing edge weights. A term chooses one outgoing edge at every vertex other than . If the resulting functional digraph contains a directed cycle, sign-reversing inclusion-exclusion over its cycles cancels the term. The surviving choices are precisely the acyclic ones; every vertex then reaches , so they are rooted directed spanning trees, each with positive sign and its product weight. This proves the theorem.
Give every edge weight one and delete the row and column for from the out-Laplacian. Expanding the resulting banded determinant along its last available row givesThe characteristic roots are and , and the initial values giveBy the directed matrix-tree theorem, this determinant is exactly the number of directed spanning trees rooted towards .
The Stirling number of the second kind counts partitions of an -element set into nonempty unlabeled blocks. Distinguishing the block containing the last element gives
For a positive integer , count functions by the number of nonempty fibres. Their fibres form a -block partition in ways, and the blocks receive distinct images in ways. HenceBoth sides are polynomials of degree agreeing at every positive integer, so this is a polynomial identity.
Partitions of the vertices into independent blocks are the unlabeled colour-class partitions counted by the Graphical Stirling number. ThereforeSince , the ordinary Stirling identity applied to givesUniqueness in the falling-factorial basis proves the claim.
In an independent-block partition of , either lie in different blocks, giving a partition valid for , or they lie in the same block. Contracting those endpoints in the second case gives an independent-block partition of . This bijection proves the recurrence. Multiplying by and summing givesUsing and induction from yields
A proper colouring using exactly colours first partitions the vertices into nonempty independent colour classes and then injectively assigns of the named colours to those classes. These choices numberSumming over counts every proper colouring exactly once; the expression is the chromatic polynomial .
A nonempty Dyck path decomposes uniquely as an up-step, a Dyck path, a down-step, and another Dyck path. Marking each matched outer pair by givesThe solution with constant term one is
Deleting the first up-step and last down-step of a strictly positive walk of semilength and lowering the remainder by one gives a Dyck path of semilength , bijectively. Hence and
A strictly negative primitive excursion of semilength has all steps negative, so weight . Reflection in the axis identifies it with a strictly positive excursion, givingEvery bridge has a unique decomposition at successive returns to the axis into positive or negative primitive excursions. The sequence construction therefore has generating functionIts exponent of is half the number of negative steps, proving the asserted interpretation of .
Since ,Rationalizing and using the Catalan generating function givesThus the coefficient of every , , is . The number of bridges with exactly negative steps is consequently independent of .
Let be the incidence vector of . Thenfor . If , taking the dot product with gives . The vectors are linearly independent, so .
The required prime-power form of the Frankl-Wilson theorem is: if , no is divisible by , and every intersection of distinct members has size divisible by , thenIt follows by associating to each its incidence vector augmented by a constant coordinate and applying the Frankl-Wilson polynomial independence lemma to the layers of multilinear intersection polynomials. The hypotheses make the diagonal evaluations nonzero modulo and every -fold off-diagonal evaluation zero; independence leaves at most polynomials. Applying the theorem gives the desired bound.
When , the argument works directly over without the constant-coordinate lift and yields the stronger boundTherefore a family of size does not exist.
The Alon-Tarsi lemma says that if and consists of distinct field elements, thenThis follows by applying univariate Lagrange interpolation successively in each variable.
If the displayed coefficient is nonzero, at least one summand has . This is the coefficient form of the Combinatorial Nullstellensatz.
Write each hyperplane as , normalized so that , and putThen and vanishes at every other point of . Reduce modulo in every variable. This preserves its function on , does not increase total degree, and gives the unique representative with each variable degree at most .
The unique reduced polynomial for the delta function at zero iswhose total degree is . HenceUniqueness follows equally from the Alon-Tarsi lemma on the grids .
Articles by others on the same topic
There are currently no matching articles.