Schema-Gated Orchestration

• The paper proves every nice graph is total weight (1,5)-choosable, reducing the bound from (1,17) to (1,5).
• It leverages the Combinatorial Nullstellensatz and permanents to guarantee a sufficient exponent vector with entries at most 4.
• The approach uses structural decompositions, including good subsets and covering families, to orchestrate algebraic reductions effectively.

The paper studies a list-weighting version of the classical 1-2-3 conjecture, but in the stronger “total weighting” framework where both vertices and edges receive weights.

A total weighting of a graph $G=(V,E)$ is a map

$\phi:V\cup E\to \mathbb{R}.$

It is proper if for every edge $\{u,v\}\in E$,

$$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$

So the “weighted degree” at each endpoint of every edge must be different.

A list version is defined as follows. For a function $\psi:V\cup E\to \mathbb{N}^+$, a $\psi$-list assignment gives each vertex/edge $z$ a list $L(z)$ of $\psi(z)$ allowable real weights. A graph is total weight $\psi$-choosable if every such assignment admits a proper $\phi:V\cup E\to \mathbb{R}.$0-total weighting $\phi:V\cup E\to \mathbb{R}.$1 with $\phi:V\cup E\to \mathbb{R}.$2.

In the special uniform case, $\phi:V\cup E\to \mathbb{R}.$3 is total weight $\phi:V\cup E\to \mathbb{R}.$4-choosable if every vertex has list size $\phi:V\cup E\to \mathbb{R}.$5 and every edge has list size $\phi:V\cup E\to \mathbb{R}.$6. The paper focuses on the case $\phi:V\cup E\to \mathbb{R}.$7: one available weight at each vertex and five at each edge.

A graph is called nice if it has no isolated edges. This condition is essential: if $\phi:V\cup E\to \mathbb{R}.$8 is an isolated edge, then the weighted degrees at $\phi:V\cup E\to \mathbb{R}.$9 and $\{u,v\}\in E$0 are forced to be equal, since both vertices see only that edge, so no proper total weighting can exist. Thus isolated edges are immediate obstructions.

Main theorem

The paper’s main result is:

$\{u,v\}\in E$1

Equivalently, for any assignment of one real number to each vertex and five real numbers to each edge, there is a proper total weighting choosing one allowed weight from each list.

This improves the prior bound of Cao, who proved every nice graph is $\{u,v\}\in E$2-choosable.

Relation to earlier conjectures and results

The paper places this in the context of several major conjectures/results:

1. Wong and Zhu’s conjecture: every nice graph is $\{u,v\}\in E$3-choosable. This would strengthen the 1-2-3 conjecture in a list-total-weight form, but remains open.
2. Cao’s result: every nice graph is $\{u,v\}\in E$4-choosable. This was the first proof that there is a universal constant $\{u,v\}\in E$5 such that every nice graph is $\{u,v\}\in E$6-choosable.
3. This paper improves the universal constant from $\{u,v\}\in E$7 to $\{u,v\}\in E$8:

$\{u,v\}\in E$9

The paper also explains the relationship to edge-weight choosability: for connected non-bipartite graphs, $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$0-choosability is equivalent to edge-weight $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$1-choosability by shifting vertex weights into edge weights along an odd cycle. But in general, $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$2-choosability is stronger, especially for bipartite graphs.

Algebraic reformulation

A major tool is the Combinatorial Nullstellensatz via a polynomial encoding of proper total weightings.

For a graph $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$3, define

$$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$4

Then a weighting $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$5 is proper iff

$$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$6

For the $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$7 setting, the vertex variables can be normalized away, and one studies

$$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$8

A monomial $$\sum_{e\in E(u)}\phi(e)+\phi(u)\neq \sum_{e\in E(v)}\phi(e)+\phi(v).$$9 is called sufficient for $\psi:V\cup E\to \mathbb{N}^+$0 if some monomial $\psi:V\cup E\to \mathbb{N}^+$1 with $\psi:V\cup E\to \mathbb{N}^+$2 appears in $\psi:V\cup E\to \mathbb{N}^+$3 with nonzero coefficient. By Nullstellensatz, if $\psi:V\cup E\to \mathbb{N}^+$4 for all edges and $\psi:V\cup E\to \mathbb{N}^+$5 is sufficient, then $\psi:V\cup E\to \mathbb{N}^+$6 is $\psi:V\cup E\to \mathbb{N}^+$7-choosable.

The paper’s strategy is to prove that every nice graph has a sufficient exponent vector $\psi:V\cup E\to \mathbb{N}^+$8 with all entries at most $\psi:V\cup E\to \mathbb{N}^+$9. That is the algebraic form of $\psi$0-choosability.

Permanent / matrix formulation

The coefficients of these polynomials are expressed using permanents of certain matrices. Define

$\psi$1

Then

$\psi$2

where $\psi$3 is the permanent of the matrix formed by repeating columns according to $\psi$4.

For a graph $\psi$5, the relevant coefficient of $\psi$6 can be written as

$\psi$7

where $\psi$8 and $\psi$9 are the signed incidence-type matrices defined in the paper.

The paper then introduces an inner-product perspective: $z$0 where $z$1 is essentially

$z$2

and $z$3 is

$z$4

This is the algebraic backbone of the proof.

Structural decomposition: good subsets and covering families

The combinatorial part of the proof is built around a carefully chosen subset $z$5, called a good subset. The graph is decomposed relative to $z$6 into four edge classes:

• $z$7: edges in $z$8 that are not isolated there,
• $z$9: edges in $L(z)$0 that are isolated there,
• $L(z)$1: edges with exactly one endpoint in $L(z)$2,
• $L(z)$3: edges with both endpoints in $L(z)$4.

The “good subset” conditions ensure a controlled structure:

• $L(z)$5 has maximum degree at most 1,
• $L(z)$6 has no isolated edges,
• each $L(z)$7 has at most one private neighbour,
• each edge in $L(z)$8 has an associated external witness vertex $L(z)$9,
• each isolated edge in $\psi(z)$0 has both endpoints adjacent into $\psi(z)$1.

A key lemma proves that every nice graph has such a good subset.

The next idea is a $\psi(z)$2-covering family $\psi(z)$3, which consists of:

• an $\psi(z)$4-covering family,
• an $\psi(z)$5-covering family,
• an $\psi(z)$6-covering family.

These are families of paths/edges/odd closed walks designed to encode the edge structure algebraically. The crucial target is to make each edge appear in the covering family at most 4 times if it is not in $\psi(z)$7, and 0 times if it is in $\psi(z)$8.

Key lemmas and proof strategy

1. Key lifting lemma

If $\psi(z)$9 is sufficient for the subgraph $\psi$0, then adding the incidence pattern of a $\psi$1-covering family preserves sufficiency for the whole graph: $\psi$2 This is the main reduction lemma.

Intuitively, one proves sufficiency for the “core” $\psi$3, then lifts it to $\psi$4 using the paths and odd walks in $\psi$5, which contribute the needed algebraic factors.

2. Existence of a good subset

A maximum independent set $\psi$6 is chosen with minimal number of isolated edges in $\psi$7. A sequence of local adjustments eliminates problematic private-neighbour configurations and produces a good subset.

3. Construction of a covering family

The paper first constructs a covering family giving a bound of 5, then refines it to 4.

• In the first stage, the paths in $\psi$8-covering families are built from vertices in $\psi$9 with at least two neighbours in $\phi:V\cup E\to \mathbb{R}.$00, using short paths of length 2 or 3.
• The $\phi:V\cup E\to \mathbb{R}.$01-part uses triangles $\phi:V\cup E\to \mathbb{R}.$02, one for each edge $\phi:V\cup E\to \mathbb{R}.$03.
• The $\phi:V\cup E\to \mathbb{R}.$04-part uses a selected adjacent edge from $\phi:V\cup E\to \mathbb{R}.$05 for each isolated edge of $\phi:V\cup E\to \mathbb{R}.$06.

This first construction gives the bound 5.

4. Refinement to reach 4

To improve 5 to 4, the paper identifies “special neighbours” and proves structural lemmas showing that vertices with such neighbours have degree at most 2 in $\phi:V\cup E\to \mathbb{R}.$07. It also proves some forbidden configurations:

• no degree-1 vertex adjacent to a degree-2 vertex,
• no two adjacent degree-2 vertices.

These are used to guarantee a more delicate choice of the $\phi:V\cup E\to \mathbb{R}.$08-covering family and the $\phi:V\cup E\to \mathbb{R}.$09-covering edges so that the multiplicity of every edge is reduced to at most 4.

A key new ingredient is a good assignment

$\phi:V\cup E\to \mathbb{R}.$10

choosing, for each $\phi:V\cup E\to \mathbb{R}.$11, a non-private neighbour $\phi:V\cup E\to \mathbb{R}.$12, with a compatibility condition that avoids conflicts with the isolated edges in $\phi:V\cup E\to \mathbb{R}.$13. The existence of such an assignment is proved by orienting an auxiliary labelled multigraph and extracting sinks/transitive tournaments.

This assignment allows the authors to control which paths are removed or modified when an edge in $\phi:V\cup E\to \mathbb{R}.$14 would otherwise force an edge in $\phi:V\cup E\to \mathbb{R}.$15 to appear 5 times.

Analytic ingredient in the final lemma

The proof of the main algebraic lemma ultimately reduces to showing nonvanishing of certain inner products of polynomials. At the end, the argument uses a real-analysis positivity trick.

One defines

$\phi:V\cup E\to \mathbb{R}.$16

and proves that for any polynomial $\phi:V\cup E\to \mathbb{R}.$17, there is a monomial $\phi:V\cup E\to \mathbb{R}.$18 such that

$\phi:V\cup E\to \mathbb{R}.$19

The proof uses an auxiliary inner product represented by an integral over the torus: $\phi:V\cup E\to \mathbb{R}.$20 and shows that a certain matrix is nonsingular by proving its quadratic form is nonzero via positivity of

$\phi:V\cup E\to \mathbb{R}.$21

This is the same real-analytic style introduced by Cao.

Final consequences

The immediate consequence is the main theorem:

$\phi:V\cup E\to \mathbb{R}.$22

This yields a strong universal bound for total weight list choosability with one vertex option and five edge options. It also implies corresponding algebraic choosability, since the proof is entirely via a monomial existence criterion in the polynomial $\phi:V\cup E\to \mathbb{R}.$23.

No further corollaries are emphasized beyond the theorem itself, but the result sharpens the best known universal constant for the list-total-weight version of the 1-2-3 conjecture from 17 to 5, and it strengthens the framework for future improvements toward the conjectured 3.