eris-econ

Geometric economics: multi-dimensional decision manifolds, A* pathfinding, and Bond Geodesic Equilibrium.

Overview

Classical economics reduces human decisions to scalar utility maximization. This library implements the geometric economics framework from Bond (2026), which models economic decisions as pathfinding on a 9-dimensional decision manifold.

The core insight: Homo economicus is not wrong because humans are irrational — it's incomplete because it computes on a projected subspace of the actual decision manifold. Projecting from 9 dimensions to 1 (monetary utility) destroys information that is mathematically irrecoverable.

The Nine Dimensions

Every economic state is a point in R^9:

Dimensions d₁–d₄ are transferable (conserved in bilateral exchange). Dimensions d₅–d₉ are evaluative (not conserved, enabling mutual gains from trade).

Key Concepts

flowchart TB
    subgraph R9[R9 decision manifold]
      direction LR
      T[Transferable d1-d4<br/>consequences rights<br/>fairness autonomy]
      E[Evaluative d5-d9<br/>privacy social virtue<br/>legitimacy epistemic]
    end

    STATE[Current state]
    GOALS[Candidate goals]

    subgraph ASTAR[A-star pathfinding]
      G[g System 2<br/>deliberate cost]
      H[h System 1<br/>moral intuition]
      M[Mahalanobis + boundary penalties]
    end

    PATH[Bond Geodesic<br/>optimal path]
    BGE[Bond Geodesic Equilibrium<br/>multi-agent fixed point]

    subgraph EMRG[Emergent phenomena]
      LA[Loss aversion lambda 2.25]
      RD[Reference dependence]
      EE[Endowment effect]
      FR[Framing effects]
    end

    T --> STATE
    E --> STATE
    STATE --> G
    GOALS --> G
    M --> G
    M --> H
    G --> PATH
    H --> PATH
    PATH --> BGE
    M --> LA
    M --> RD
    M --> EE
    M --> FR

Bond Geodesic

The optimal path on the decision manifold from current state to goal, minimizing Mahalanobis distance + boundary penalties. This is computed via A* search, where:

• g(n) = accumulated cost (System 2: deliberate calculation)
• h(n) = heuristic estimate (System 1: moral intuition)

Bond Geodesic Equilibrium (BGE)

Generalization of Nash equilibrium to multi-dimensional manifolds. Each agent minimizes behavioral friction on their own decision complex. Reduces to Nash when all non-monetary dimensions vanish.

Emergent Behavioral Phenomena

These are not ad-hoc biases — they emerge geometrically:

• Loss aversion (λ ≈ 2.25): losses traverse more dimensions than gains
• Reference dependence: distance measured from current state on manifold
• Endowment effect: ownership activates rights + identity dimensions
• Framing effects: gauge transformations on the description basis

Installation

pip install eris-econ

Or for development:

git clone https://github.com/ahb-sjsu/eris-econ.git
cd eris-econ
pip install -e ".[dev]"

Quick Start

from eris_econ.games import ultimatum_game
from eris_econ.pathfinding import astar

# Build the proposer's decision complex
E = ultimatum_game(stake=10.0)

# Find the optimal offer (Bond Geodesic)
goals = {f"offer_{p}" for p in [0, 10, 20, 30, 40, 50]}
result = astar(E, "start", goals)

print(f"Optimal choice: {result.path[-1]}")  # ~offer_40 (not offer_0!)
print(f"Total behavioral friction: {result.total_cost:.3f}")

from eris_econ.behavioral import compute_loss_aversion
import numpy as np

# Compute emergent loss aversion from metric structure
sigma = np.eye(9)
sigma[0, 6] = 0.3  # consequences-identity coupling
sigma[6, 0] = 0.3
lambda_ratio = compute_loss_aversion(sigma)
print(f"Loss aversion λ = {lambda_ratio:.2f}")  # > 1.0

Reproducing Table VII of Bond (2026, IEEE TCSS)

The sixteen-target benchmark of Geometric Prediction of Economic Behavior: Cross-Domain Validation Across Game Theory and Prospect Theory is recomputed, not stored. One command prints the table with the paper's rounding, the 16/16 pass count, the overall MAE of 2.70%, the Ruggeri lottery-subset MAE of 3.95%, and the game and non-game figures quoted in the text:

eris-econ-table7                      # plain text
eris-econ-table7 --format latex       # booktabs rows for the manuscript
eris-econ-table7 --format json        # machine-readable, with the summary

from eris_econ.paper import table_vii, selected_sigma

t = table_vii()
t.summary["overall_mae"]              # 2.702...
selected_sigma()                      # the paper's diagonal covariance

tests/test_paper.py asserts every printed value of Table VII, so the test suite fails if the package ever stops reproducing the paper.

Modules

Citation

If you use this library in academic work, please cite:

@article{bond2026geometric,
  title={Geometric Economics: Multi-Dimensional Decision Manifolds and Bond Geodesic Equilibrium},
  author={Bond, Andrew H.},
  journal={Journal of Economic Theory},
  year={2026}
}

License

MIT