Answers are grounded in this paper and cite their sources.
The study classifies admissible cost functions and establishes existence of unique combiners, suggesting new theoretical insights.
We classify the continuous reciprocally symmetric cost functions J:(0,∞)→R with J(1)=0 and strictly convex log-substitution G(t):=J(et) (admissible costs) for which the symmetric compound J(xy)+J(x/y) depends on (x,y) only through (J(x),J(y)). We first prove that this dependence is automatic: for every admissible J, there exists a unique continuous combiner (the auxiliary function P that encodes the compound) P:[0,∞)2→R with J(xy)+J(x/y)=P(J(x),J(y)) for all x,y>0 (Theorem 1); P is symmetric, non-negative, satisfies P(u,0)=2u, and inherits monotonicity and coercivity from admissibility. When P is required to be a polynomial, a growth rate comparison between two recursions for G forces degP≤2 (Theorem 4), so P(u,v)=cuv+2u+2v with c≥0, and the corresponding admissible costs are exhausted by two explicit families (Theorem 8)—the hyperbolic family J(x)=c−1(xλ+x−λ)−2c−1 (c,λ>0) and the degenerate quadratic family J(x)=a(lnx)2 (a>0)—with the latter arising as the Inönü–Wigner contraction λ→0+, λ2/c→a of the former (Theorem 9). Two regularity extensions are obtained: a Lebesgue-measurable cost satisfying explicit regularity hypotheses admits a continuous representative (Theorem 5), and in the entire finite-order regime, the diagonal combiner Q(u):=P(u,u), when polynomial of degree d′, obeys the sharp bound d′≥2ρ (Theorem 6), attained with equality in both classified families. The normalisations P(1,1)=6 and G′′(0)=1 single out the canonical representative Jcost(x)=12(x+x−1)−1.
No takes yet. Share an insight, caveat, or question.
Pardo-Guerra et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: