← All researchMF-PRISM-DLT-2026-05 / interactive guide

INTERACTIVE RESEARCH GUIDE

Stop the controller from chasing noise.

A real advantage can be obscured by noisy telemetry. Test how a switching margin changes behavior.

Research reportAQR-C: Convergence of Independently Adaptive Federated Quorum Controllers

Calculated illustration · change the inputs to inspect the mechanism.

01 / 04Guided chapter

Separate safety from convergence

Two configurations can both be structurally admissible while a controller oscillates between them. A safe choice set does not automatically create stable adaptive behavior.

Telemetry error amplitude

0.4

Switching margin

0.25

True B advantage: 0.20
Profile switches: 1
Final profile: B

This bounded-noise trace makes 1 switch and ends on B. A higher switching margin trades responsiveness for resistance to oscillation.

Switch only when measured gain over the current profile > margin.

TRY THIS

Compare margin zero with margin 0.60. A quiet controller is not automatically an optimal one: too much hysteresis can leave it on the worse profile.

What this experiment represents. Synthetic single-controller trace with fixed true advantage and bounded deterministic measurement error. Not a simulation of complete federated consensus or proof of the multi-controller theorem.

Source manuscript & release ↗Read the full explanation ↓

JavaScript is off. The initial figure above and the complete written explanation below remain available. Interactive controls require JavaScript.

1. Separate safety from convergence

Two configurations can both be structurally admissible while a controller oscillates between them. A safe choice set does not automatically create stable adaptive behavior.

2. Observe a noisy score

In this synthetic example, profile B has a fixed true advantage of 0.20. A deterministic bounded oscillation is added to the measured advantage.

3. Require a switching margin

The controller switches only when the measured gain over its current choice exceeds the selected margin. A larger margin filters noise but may also prevent a useful switch.

4. Read the tradeoff

The paper gives convergence conditions under explicit assumptions. This one-controller trace illustrates hysteresis; it does not prove convergence of independently adapting network controllers.

Distributed Systems · Research Paper

AQR-C: Convergence of Independently Adaptive Federated Quorum Controllers

Potential-guided selection, robust hysteresis, conflict-colored activation, and failover rendezvous

Research reportMF-PRISM-DLT-2026-05 / v0.2

Separate structurally safe configurations from adaptive-controller convergence. The paper studies oscillation, robust switching margins, conflict-free parallel changes and recovery rendezvous under explicit assumptions.

Read the paper Release & files

Cover of AQR-C: Convergence of Independently Adaptive Federated Quorum Controllers — Current public editionv0.2 · 2026-09-12

Identifier: MF-PRISM-DLT-2026-05
Series: Distributed Systems
Edition: Version 0.2
Length: 26 pages
Reserved DOI: 10.5281/zenodo.22728322 (record reserved)

Plain-language guide

3 minute explanation

A plain-language summary.

The paper separates configuration safety from controller stability and gives conditions under which independently adapting validators settle instead of oscillating.

adaptive controlfederated consensusconvergence

Even if every selectable quorum configuration is safe, local controllers can still keep switching back and forth, react to noisy telemetry, or activate incompatible changes at the same time.

AQR-C introduces a near-potential objective, robust hysteresis, a conflict graph for parallel changes, and a shared failover rendezvous. Together these mechanisms aim to make distributed adaptation converge under explicit assumptions.

How it works

The argument, without the notation.

Guide choices with a potential

Require local improvements to align approximately with a global scalar objective despite bounded telemetry error.

Add robust hysteresis

Demand a sufficient improvement margin before switching so noise cannot trigger endless reversals.

Coordinate compatible changes

Color a conservative conflict graph for parallel activation and use a common rendezvous state during failover.

What it adds

Main findings.

Structural quorum safety and adaptive-controller convergence are distinct proof obligations.
Hysteresis and bounded-error margins prevent small telemetry fluctuations from causing thrashing.
Conflict-aware activation allows some changes in parallel while preserving the convergence argument.

Scope & limits

Evidence and limitations.

Current status

Convergence conditions and synthetic controller experiments. Structural safety, controller convergence, and complete consensus correctness remain separate obligations.

This paper gives conditions under which independently adapting quorum controllers settle instead of oscillating. The arguments hold under the stated assumptions and are illustrated with synthetic controller experiments.

Controller convergence does not by itself prove consensus safety or liveness.
The argument assumes the environment eventually stabilizes within stated bounds.
Nodes must identify shared recovery scenarios consistently for the rendezvous mechanism to work.

Why it matters

Why this research matters.

A set of safe configurations is not enough: controllers can still thrash between them. The paper isolates the additional assumptions needed for stable adaptation.

Key terms

Terms used in this paper.

These are the terms needed to understand the claim. The full paper uses them more precisely.

Convergence

Reaching a stable configuration rather than switching indefinitely.

Hysteresis

A switching margin that prevents immediate reversal after small input changes.

Potential function

A scalar quantity used to show that each permitted update makes global progress.

Conflict graph

A graph marking which changes cannot safely activate at the same time.

Independent review

Where scrutiny should concentrate.

01
Bounded telemetry error and near-potential certificates.
02
Conservative conflict graph and commuting activations.
03
Stabilized environment and shared recovery-scenario detection.

Publication record

Read the paper and examine the source files.

This page is a reading guide, not a substitute for the manuscript. The public record links the explanation to the paper, source package, review materials, and persistent identifier.

GitHub release Source folder

Document: Research Paper
Review status: Open for independent review
Published: 2026-09-12
DOI status: Reserved · 10.5281/zenodo.22728322
Canonical record: MF-PRISM-DLT-2026-05

Metriq PRISM Laboratory, AQR-C: Convergence of Independently Adaptive Federated Quorum Controllers, Metriq PRISM Laboratory Research Paper MF-PRISM-DLT-2026-05, Version 0.2, 2026. Corresponding contributor: Daniel H. Jeffery, ORCID 0009-0001-1200-6042. DOI: 10.5281/zenodo.22728322.