Fair and Efficient Balanced Allocations for Additive Valuations

Benjamin Cookson, Nisarg Shah, Paritosh Verma

For any number of indivisible goods and agents with nonnegative additive valuations, an allocation exists with bundle sizes differing by at most one, envy-free up to one good, and no balanced fractional reallocation that Pareto improves it.

Sources for this record

What was already known

EF1 and fractional Pareto optimality were known together without balancedness. Balanced results had weaker guarantees or restricted valuation classes.

AI involvement AI-led

Cookson, Shah and Verma supplied research direction, connections and strategies. They credit Codex GPT-5.6 Sol with deriving all proofs and counterexamples, then verified and simplified them.

The authors explicitly credit Codex with all proofs and counterexamples, while crediting themselves with strategy and review.

Importance Notable

Extends the combined existence guarantee to general nonnegative additive valuations.

Limitations

The result proves existence, not a polynomial-time allocation algorithm.

Technical verification Formal (human)

The three human authors explicitly report verifying all mathematical details of this paper, including the main balanced EF1/fPO theorem; this is a reported complete human check.

Formal checking

Mathematical details of the balanced EF1 and fPO existence theorem. author proof verification and simplification

Checker
Benjamin Cookson, Nisarg Shah, and Paritosh Verma (authors)
Scope
Mathematical details of the balanced EF1 and fPO existence theorem
Basis
Reported by the linked checker
Evidence source

Checking a formal statement does not establish that it matches the original problem.

Human understanding Full

The human authors give a technical overview of weighted-welfare faces, price interlacing, KKM selection, and the order-statistic correction, and report simplifying the proofs.

Correction and publication history