mirror of
https://github.com/hanzoai/papers.git
synced 2026-11-03 02:27:30 +00:00
zeekay
f9db4f5737
Bylines, paper authorship, correspondence addresses and the founder blurb under each post named an individual. They name the organisation that published the work instead: the brand on posts and pages, the research byline on papers, and research@ in place of a personal address. Both are what these repositories already used elsewhere — nothing new was coined. Co-authors are untouched. Where a paper named someone else alongside, they stay named; only the one byline changed. Bibliography entries, running heads, PDF metadata and affiliations move with it, so a citation still resolves to whoever the cited paper names. Sample and fixture data uses the placeholder name these tests already use elsewhere, so no assertion changed meaning.
108 lines
4.0 KiB
TeX
108 lines
4.0 KiB
TeX
% Hanzo AI Chain (AIVM): The AI Virtual Machine on Lux A-Chain
|
|
% Working draft: 2026-08-18
|
|
\documentclass[11pt,a4paper]{article}
|
|
\usepackage{../shared/hanzocover}
|
|
\usepackage[utf8]{inputenc}
|
|
\usepackage[T1]{fontenc}
|
|
\usepackage{amsmath,amssymb,amsfonts,amsthm}
|
|
\usepackage{mathtools}
|
|
\usepackage{graphicx}
|
|
\usepackage{booktabs}
|
|
\usepackage{hyperref}
|
|
\usepackage{algorithm}
|
|
\usepackage{algpseudocode}
|
|
\usepackage{enumitem}
|
|
\usepackage{xcolor}
|
|
\usepackage{listings}
|
|
\input{../shared/lstlang}
|
|
\usepackage{tikz}
|
|
\usetikzlibrary{shapes,arrows,positioning,fit,backgrounds}
|
|
|
|
\hypersetup{colorlinks=true,linkcolor=black,citecolor=blue,urlcolor=blue}
|
|
|
|
\newtheorem{theorem}{Theorem}[section]
|
|
\newtheorem{lemma}[theorem]{Lemma}
|
|
\newtheorem{corollary}[theorem]{Corollary}
|
|
\newtheorem{proposition}[theorem]{Proposition}
|
|
\newtheorem{definition}{Definition}[section]
|
|
\newtheorem{remark}{Remark}[section]
|
|
|
|
\lstset{
|
|
basicstyle=\ttfamily\small,
|
|
breaklines=true,
|
|
frame=single,
|
|
numbers=left,
|
|
numberstyle=\tiny,
|
|
xleftmargin=2em,
|
|
}
|
|
|
|
\title{Hanzo AI Chain (AIVM)\\[4pt]
|
|
\large Useful-Work Mining, the Native AI Coin, and\\
|
|
Post-Quantum Omnichain Settlement on Lux}
|
|
|
|
\author{
|
|
Hanzo AI Research\thanks{Corresponding author: [email protected]}\\
|
|
\textit{Hanzo Industries, Inc.\ (Techstars '17)}\\
|
|
\texttt{[email protected]}
|
|
}
|
|
|
|
\date{Working draft: 18 August 2026}
|
|
|
|
\begin{document}
|
|
|
|
\hanzocoverpage
|
|
|
|
\begin{abstract}
|
|
We present Hanzo Proof of AI (PoAI), the useful-work consensus of the Lux
|
|
A-Chain and the economic substrate of Hanzo Network. A real AI job is admitted
|
|
before mining and binds its consumer, model, input, deterministic execution
|
|
policy, proof policy, reward, and payout destination. A CPU or GPU provider
|
|
executes that job and commits its integer output, graph transcript, and
|
|
metering. A-Chain PoAI consensus validates that the bound useful work was
|
|
actually performed, consumes one global nullifier, and finalizes one mining
|
|
receipt. Z-Chain batches finalized receipt transitions through P3Q, a
|
|
Plonky3-derived, pairing-free STARK/FRI settlement rollup. Lux Quasar orders and
|
|
finalizes the A-Chain and Z-Chain states under the activated post-quantum
|
|
validator policy. Neither Z-Chain nor another destination may validate raw work
|
|
or create independent AI supply.
|
|
|
|
The public execution policy is a versioned deterministic kernel profile rather
|
|
than a hardware mandate. It fixes operator semantics, tensor layout, accumulator
|
|
and overflow rules, reduction, fixed-point scaling, rounding, and canonical
|
|
serialization. Conforming CPU and GPU implementations emit the same canonical
|
|
transcript binding the job, model, input, graph, output, metering, provider, and
|
|
destination. Cross-backend parity is a release conformance test for that profile,
|
|
not another proof system or consensus lane.
|
|
|
|
The native coin \textbf{AI} has a hard maximum nominal supply of two trillion:
|
|
one trillion allocated permanently to the Hanzo DAO and one trillion reserved
|
|
for proof-earned miner and validator rewards under a Bitcoin-shaped geometric
|
|
halving schedule. Existing AI clouds can keep their normal workload revenue and
|
|
earn additional AI by attaching proof evidence to work they already perform;
|
|
decentralized operators use the same interface for open jobs. Exact-integer
|
|
commitments permit ordinary CPUs to validate challenged matrix products more
|
|
cheaply than GPU execution, while complete protocol security additionally
|
|
requires graph binding, challenge coverage, transcript availability, verifiable
|
|
metering, and finality. TEE evidence is optional for confidential workloads,
|
|
and succinct P3Q settlement authority remains activation- and audit-gated.
|
|
\end{abstract}
|
|
|
|
\tableofcontents
|
|
\newpage
|
|
|
|
% ============================================================================
|
|
|
|
\input{sections/01-introduction}
|
|
\input{sections/02-architecture}
|
|
\input{sections/03-ai-coin}
|
|
\input{sections/04-hmm-registry}
|
|
\input{sections/05-compute-markets}
|
|
\input{sections/06-agent-registry}
|
|
\input{sections/07-cross-chain}
|
|
\input{sections/08-privacy}
|
|
\input{sections/09-security}
|
|
\input{sections/10-deployment}
|
|
\input{sections/99-conclusion}
|
|
|
|
\end{document}
|