TINA

This repository contains the manuscript and complete reproduction package for A Theory of Information Architecture for Networked Decisions: Freshness, Locality, and Coordination, by Scott Moeller and Bhaskar Krishnamachari.

The paper studies decision-relevant information scope and freshness in an exogenously evolving quadratic static-team model. Its principal design family uses synchronized neighborhood snapshots: a broader snapshot arrives at a greater age. The repository includes the exact manuscript inputs, archived raw and processed numerical results, and deterministic build/test entry points.

Repository map

• main.tex, references.bib: canonical manuscript sources.
• figures/: the six PDF figures consumed by main.tex.
• tina.pdf: canonical compiled preprint.
• numerics/config/: fixed experiment parameters and seeds.
• numerics/experiments/: six experiment drivers and the all-experiment runner.
• numerics/src/: covariance, process, regret, theory, graph, and plotting code.
• numerics/results/: archived raw arrays, processed CSVs, and run metadata.
• docs/: numerical specification, parameter inventory, audit notes, and a clean-room reproduction checklist.
• tests/: closed-form, covariance, team-optimality, and package-integrity tests.
• scripts/: clean manuscript build and source/reference audits.

Reproduce the numerical artifacts

Python 3.12 is recommended. From the repository root:

python -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install -r requirements-lock.txt
python numerics/experiments/run_all.py
python -m unittest discover -s tests -v

To preserve the archived reference results while testing a clean run:

python numerics/experiments/run_all.py --output-root tmp/reproduction --no-paper-copy

The reference run takes about one minute on a contemporary laptop. The runner records configurations, software versions, seeds, diagnostics, and headline errors in numerics/results/metadata.

Build the paper

A TeX installation containing pdflatex, bibtex, the standard packages used by main.tex, and IEEEtran.bst is required.

python scripts/audit_sources.py
python scripts/build_paper.py

The build script copies only the declared source closure to a temporary build directory, runs pdflatex/bibtex/pdflatex/pdflatex, rejects undefined citations or references, and writes output/pdf/tina.pdf.

Scope and license

The numerical work verifies or illustrates results inside the paper's common-rate Gaussian-affine model. It does not test mixed-age architectures, heterogeneous temporal rates, endogenous dynamics, nonlinear decision maps, or action constraints. Code and repository content are released under the BSD 3-Clause License. See CITATION.cff for citation metadata.