default
plain-language theorem explainer
Canonical Gibbs policy for the RS→RL bridge: recognition temperature fixed at 1 and action cost taken from the virtue energy functional, independent of state. Anyone wiring thermodynamic action selection (softmax over virtues) cites this as the neutral baseline. Construction is a structure instance: positivity of temperature by norm_num, cost field a direct projection of VirtueAction.energyCost.
Claim. The default Gibbs policy is the triple $(T_R,\,J)$ with recognition temperature $T_R=1$ (hence $T_R>0$) and cost $J(s,a)$ equal to the energy cost of the virtue-coefficient action $a$, independent of the moral state $s$. The induced distribution is $p(a\mid s)\propto\exp(-J(s,a)/T_R)$.
background
The RS→RL bridge treats Recognition Science as control theory: moral states carry ledger, bonds, skew, and energy; admissible moves are the 14 virtue generators; hard feasibility is σ=0 via LACompletion; selection is lexicographic then thermodynamic.
A Gibbs policy packages the RS thermodynamic rule $p(a\mid s)\propto\exp(-J(s,a)/T_R)$. Here $T_R$ is the recognition temperature (strictness of the Boltzmann factor) and $J$ is any real cost on (MoralState, VirtueAction). The module equates this form with soft actor-critic / maximum-entropy RL, but with $J$ grounded in RS rather than an arbitrary reward.
Upstream, recognition cost is consistently the J-cost on ratios or events (ObserverForcing, MultiplicativeRecognizer, RungCoarsen). The sibling energyCost supplies the concrete $J$ used here: cost of a virtue action alone, so the default policy is state-blind in its energy term.
proof idea
Definitional structure instance, not a proof. Set temp_R := 1. Discharge temp_pos : 0 < temp_R by norm_num. Set cost to ignore the state and return VirtueAction.energyCost a. No lemmas beyond positivity of the constant 1.
why it matters
Gives the neutral thermodynamic baseline for the bridge: unit temperature and pure virtue energy cost, so downstream RL machinery can sample or score actions without inventing a temperature or a cost ad hoc.
It sits under the module's GibbsPolicy design (thermodynamic learning item 5) and feeds any consumer that needs a concrete policy instance. Downstream uses span certificate displays, external anchors, Aczél regularity kernels, and CPM defect identities that import a default of this shape as a fill-in parameter rather than as physics content.
Framework landmarks: the exponential is the RS Boltzmann form tied to J (T5 uniqueness of J); eight-tick cadence elsewhere in the module is T6/T7, but this default does not itself encode the octave. Cooler or hotter policies are sibling constructors; this one is the reference $T_R=1$ point.
Switch to Lean above to see the machine-checked source, dependencies, and usage graph.