---
title: Desnanot–Jacobi–Dodgson Identity
url: https://www.emergentmind.com/topics/desnanot-jacobi-dodgson-identity
type: topic
---

# Desnanot–Jacobi–Dodgson Identity

The Desnanot–Jacobi–Dodgson Identity is a quadratic determinantal relation that connects the determinant of a square matrix with those of its principal minors. This identity underlies many fundamental arguments in linear algebra, combinatorics, and statistical mechanics, appearing both in classical matrix theory and in graph-theoretical interpretations via the Matrix–Tree theorem and spanning forest polynomials. It further generalizes to weighted and $q$-analog contexts, and has deep connections with enumerative combinatorics and applications in physics.

## 1. Canonical Formulation and Statement

Let $M$ be an $n \times n$ matrix. For any $k$ and index sequences $i_1, \ldots, i_k$ and $j_1, \ldots, j_k$, let $M(i_1,\ldots,i_k; j_1,\ldots,j_k)$ denote the submatrix formed by deleting rows $i_1,\ldots,i_k$ and columns $j_1,\ldots,j_k$. The Desnanot–Jacobi–Dodgson identity (often called Dodgson condensation when $k=2$) is:
\[
\det M \cdot \det M(1,2;1,2) = \det M(1;1)\cdot \det M(2;2) - \det M(1;2)\cdot \det M(2;1)
\]
This relation admits full generality for any four indices $i,j,k,l$ via
\[
\det A \cdot \det(A_{ij}^{kl}) = \det(A_{i}^{k}) \cdot \det(A_{j}^{l}) - \det(A_{i}^{l}) \cdot \det(A_{j}^{k})
\]
where $A_{ij}^{kl}$ denotes the submatrix deleting rows $i,j$ and columns $k,l$ [1106.2869, 1902.06234, 2110.01026].

## 2. Combinatorial and Graph-Theoretic Interpretations

The identity admits a powerful reformulation using the all-minors Matrix–Tree theorem. For a graph $G$ with Kirchhoff matrix $M_G$, each minor $\det M_G(I;J)$ can be expanded as a signed sum indexed by spanning forests, constraining the placement of marked vertices among trees. In the three-vertex ($a,b,c$) case, one obtains a quadratic forest polynomial identity:
\[
\Phi_G^{\{a,b,c\}}\Phi_G^{\{a\},\{b\},\{c\}} = \Phi_G^{\{a,b\},\{c\}}\Phi_G^{\{a\},\{b,c\}} + \Phi_G^{\{a,c\},\{b\}} \Phi_G^{\{a\},\{b,c\}}
\]
where the splitting of partitions determines contributions from different spanning forest configurations [1106.2869].

More generally, the identity is a $k=2$ case of the Dodgson–Muir (or Lewis Carroll) family, which is interpreted combinatorially via forests, $k$-tuples of spanning tree decompositions, and involution/edge-swapping principles for cancellation of overcounted structures [2110.01026].

## 3. Algebraic Generalizations and $q$-Analogs

Kobayashi introduced a $q$-deformation, the bigrassmannian determinant:
\[
\bdet(A) = \sum_{w \in S_n} (-1)^{\ell(w)} q^{\beta(w)} \prod_{i=1}^n a_{i,w(i)}
\]
with a weighted condensation formula mirroring the classical identity but introducing factors of $q^{n-1}$ in cross terms:
\[
\bdet(A)\bdet(A_{1n}^{1n}) = \bdet(A_1^1)\bdet(A_n^n) - q^{n-1}\bdet(A_1^n)\bdet(A_n^1)
\]
Setting $q=1$ reduces immediately to the classical case. This $q$-analogue is linked to combinatorial statistics on permutations and is recoverable via entrywise deformation and application of the usual condensation on the resulting matrix [1902.06234].

## 4. Higher-Order Identities and Forest Polynomial Extensions

The identity admits higher-level analogues. Vlasev–Yeats constructed a quadratic identity involving spanning forest polynomials for graphs with four marked vertices. Denote $A_i, B_j$ as specific three-part partitions indexed by coloring and set-partitioning of the vertices, with $P=(1,1,1,1)$ (all in a single tree) and $(1,2,3,4)$ (all separated). The four-vertex generalization is:
\[
P(1,2,3,4) = \sum_{\text{linear in $x_1,\ldots,x_8$}} (\text{combinations of }A_i B_j),
\]
where only $8$ linear parameters remain unfixed due to classical Dodgson-type relations—far less than the naive count of $6 \times 7$ possible $A_i B_j$ products. The right-hand side is ultimately independent of the choice of $x_i$, due to a hierarchy of identities among the $A_i$ and $B_j$. Computations with five and six marked vertices support the conjecture that, for $n$ marked points, $n(n-2)$ linear degrees of freedom survive [1106.2869].

## 5. Applications in Enumerative and Statistical Combinatorics

The identity is fundamental in log-concavity arguments for determinantal arrays, as in the exact factorization for Pascal minors:
\[
L(PD_k)_{i,j} = PD_k(i,j)^2 - PD_k(i,j-1)PD_k(i,j+1) = PD_{k-1}(i,j)PD_{k+1}(i,j)
\]
where $L$ is the row-wise log-concavity operator and $PD_k(i,j)$ is the $k\times k$ minor of the lower-triangular Pascal matrix. This structural factorization directly proves the infinite log-concavity of the determinantal arrays and yields sharp determinantal Hadamard-type inequalities [2512.06414].

In the double-dimer model, the partition function for tripartite node configurations on planar bipartite graphs satisfies a four-term condensation directly parallel to the Dodgson identity—establishing a determinant formula for dimer partition functions and enabling generalizations to random tiling theory and cluster algebras [1911.04079].

## 6. Inductive and Structural Proof Methodologies

The identity is indispensable for inductive arguments. For determinants $D(n,d)$ of matrices of rising powers of second-order linear recurrences,
\[
D(n,d)D(n+2,d)-D(n+1,d)^2 = D(n,d-1)D(n+2,d-1),
\]
the condensation step provided by the identity allows closed forms and precise exponential and product structure on the determinants, with minor adjustments for sign and parameterization per the recurrence coefficients [1608.00096].

Proofs of the identity can utilize Laplace expansions (row and column removals), determinantal recurrences (Robbins–Rumsey $X$-determinants), or direct elementary matrix operations. Combinatorial bijections can be constructed through edge-color-swapping involutions (the generalized Red Hot Potato algorithm), which provide a structural understanding of why signed cancellations occur in the expansion of determinants in terms of forests [2110.01026].

## 7. Influence, Open Questions, and Prospects

The Desnanot–Jacobi–Dodgson identity and generalizations occupy a central role in the algebraic analysis of determinants, non-intersecting path enumeration, and the graph-theoretic description of electrical networks and Feynman diagram periods. Open questions include providing a conceptual proof (rather than brute-force algebraic verification) for the count of degrees of freedom in the general $n$-vertex quadratic spanning forest polynomial identities [1106.2869]. Further, the interplay between the forest polynomial framework and applications to field theory integrals, cluster varieties, and log-concavity in combinatorial geometry continues to motivate new research directions across algebraic combinatorics, mathematical physics, and algebraic geometry.

Source: https://www.emergentmind.com/topics/desnanot-jacobi-dodgson-identity