Canonical · Relation to Adjacent Work

Control Theory

The science of making a system persist by supplying its control from outside — and the framework as the science of a system that is its own control.

Control theory is the mature engineering science of feedback: measure a system's output, compare it against a target, and feed the error back to shape the input so the system holds a desired behaviour despite disturbance. It is one of the framework's five named lineages — cited on the read index as “stability and feedback from control theory; Khalil, Nonlinear Systems.” So the kinship is claimed openly, and this page does the honest thing with the claim: it states what the framework takes, where the two sharply diverge, and where to locate the line. The debt is real: Lyapunov stability, input-to-state stability, feedback, and control-effort cost are inherited, not invented here. The departure is one clean question the classical apparatus does not ask — who supplies the control? In control theory the controller is a separate thing outside the plant. In the framework's terms that makes a controlled system the canonical attractlet. The framework's distinctive object is the structure that has absorbed its controller into itself. Control theory is, in one breath, a parent field and a contrast case — the same dual role the engine plays on the attractlet page.

The discipline, in brief

Khalil, H. K. (2002). Nonlinear Systems (3rd ed.). Prentice Hall. [verify edition/year/publisher against the primary source before publication] · Sontag, E. D. (1989). “Smooth stabilization implies coprime factorization,” IEEE Transactions on Automatic Control — the commonly cited origin of input-to-state stability (ISS). [verify title/journal/year/volume/pages, and that this is the appropriate ISS-origin citation] · the linear–quadratic regulator (LQR) as the standard control-effort-priced optimal-control example. [verify a canonical primary/textbook citation — e.g. Anderson & Moore, or Kalman's original — before publication]

A control system has a plant (the thing to be regulated) and a controller (the thing that regulates it). The controller measures the plant's output, compares it against a reference or setpoint, and computes an input that drives the error toward zero — the closed feedback loop. The discipline's central results are about stability: whether the closed loop returns to its target after a disturbance, and how much margin it has before it does not. Lyapunov theory proves such return by exhibiting an energy-like function that decreases along trajectories — the mathematical ancestor of a basin. Input-to-state stability (ISS) extends this to driven systems: the state stays bounded provided the disturbance stays bounded, and bounded transient disturbance is survivable rather than fatal. Optimal control, the LQR being its cleanest case, adds a price: holding the setpoint costs actuator effort, and the controller trades tracking error against control expenditure. Together these give control theory a vocabulary a reader of this framework will find eerily familiar — basin, margin, feedback, recovery, bounded disturbance, cost.

That familiarity is the reason this page is needed. A control theorist is a likely sophisticated reader, and the first thing such a reader will think is “isn't this just feedback control?” The rest of the page answers that question in three moves: what is genuinely shared, where the two part company, and where the line is drawn.

Where the framework agrees — and owes a debt

Feedback is self-reference made operational. Control theory is the mature science of the closed loop — output fed back to shape input. That is the framework's self-reference primitive made engineering-operational, and it is exactly the condition of recursion lock, the first of the four sovereignty conditions. The shape — a structure conditioning on the effects of its own prior state — is the same object in both languages. The framework does not claim to have discovered feedback; control theory is where feedback was made rigorous.

Lyapunov stability is the ancestor of the basin. A Lyapunov function proves a system returns to equilibrium after a disturbance; the set of states from which it returns is, in all but name, a basin of attraction. The framework's second sovereignty condition — identity maintained through bounded perturbation — and its stability-margin descriptor are direct descendants of Lyapunov and gain-margin thinking. The return-to-identity the framework watches in its basins, and in the Eco 1.0 simulation, is disturbance rejection under another name.

ISS is the closest formal cousin of window-based persistence. This is the strongest single technical bridge on the page, and it deserves to be named precisely. Input-to-state stability says a system stays bounded provided its disturbances stay bounded — bounded transient deficit is survivable, not fatal. That is almost exactly the framework's window-based persistence condition, RST-WIN: trajectory identity is maintained across bounded failures of pointwise support, provided the deficit intervals are bounded and the dynamics carry identity through them. Both say the same load-bearing thing — persistence is a window condition, not a pointwise one — one in control-theoretic dress, the other in the framework's sequence-based dress.

Optimal control already prices the bill. The framework's fourth sovereignty condition — that maintenance is never free — and its continuation-cost descriptors rhyme with the LQR's control-effort term: you spend actuator energy to hold the setpoint, and the controller optimizes against that spend. The idea that holding a state costs something, continuously, is not the framework's invention; optimal control foregrounded it decades ago. The framework inherits the instinct and gives it a persistence-ledger rather than a quadratic cost functional.

These are not loose analogies offered for color. They are why the framework can speak of margins, recovery, and bounded-disturbance survival without inventing new mathematics for stability itself: the stability apparatus is classical, control theory is where it was made to make systems behave, and the framework builds on it rather than around it.

Where the framework draws its line

The controller is external — and that is the attractlet line. In control theory there is always a controller: a separate thing that supplies the ordering — it builds the feedback law, feeds the reference signal, corrects the drift. Uncouple it and the plant relaxes. In the framework's terms, this makes a controlled system the canonical attractlet: it exhibits attractor-like stability precisely because something outside is doing the structuring. The framework's own paradigm attractlet — the engine — is literally a controlled thermodynamic system. So control theory mostly describes exogenous persistence, while the framework's sovereign attractors are endogenous: they are their own controller, self-produced. This is the deepest divide, and it is the same one the framework draws against every driven neighbor.

Setpoints are given from outside; sovereign attractors set their own. A thermostat's target is dialled in by a person. Control theory almost never asks where the setpoint comes from — it is a reference handed to the loop. A sovereign attractor regulates its own admissibility conditions through its own activity; the framework's boundary-retention condition is precisely about the system producing its own “what counts as me.” This is the control-theoretic face of the same endogenous-parameter question the framework raises against bifurcation theory (who turns the knob?) and Waddington (today's variable is tomorrow's parameter): control theory takes the setpoint as given; the framework asks who produces it.

Control theory keeps time; the framework removes it. Control is built on t — transfer functions in s, differential equations, step responses over time. The framework runs on sequence, not time: the time index never appears in an attractor's logic, and time is a non-primitive downstream consequence of recursion, never a background axis. This is not cosmetic — it is what lets the framework speak of persistence in systems that have no external clock, where control theory's core apparatus presupposes one.

Identity-bearing boundary and non-inherited loss have no clean control-theory analogue. Control theory has stability and instability, but it carries no notion of the system's identity-bearing boundary whose violation is irreversible loss of the thing itself. A controlled system that goes unstable can usually just be re-stabilized; there is no “identity not inherited across the break.” The framework's account of rupture is stronger: past a terminal break, identity is not inherited back, and any successor must ignite from scratch. That mortality is genuinely extra structure the control apparatus does not contain.

Control theory answers: how do you make a system hold its target, given a controller to supply the control? The framework's question is one the classical apparatus does not ask: what if the system is its own controller? For a driven system the honest answer is that the control comes from outside — that is the attractlet. For a sovereign one, the framework's distinctive claim is that the system is both the plant and the hand on the loop.

What the departure buys

It turns a controlled plant into a self-controlling one — the attractlet-to-sovereign step. Read control theory's picture and it is a plant plus an external controller. Absorb the controller into the plant — let the system produce its own feedback law, its own setpoint, and its own boundary — and the object stops being an attractlet and becomes a candidate sovereign attractor. This is the exact step the framework is built to describe, and it is the same step that, on the Eco 2.0 harness, carries a system that pays its own bill (Condition 4, met) toward one that produces its own boundary (Condition 3, the gap). Control theory supplies the rigorous account of the loop; the framework asks what happens when the loop closes on itself with nothing left outside.

Where this points, left open. The ISS ≈ RST-WIN correspondence is the most promising formal bridge on this page, and it is deliberately left as a bridge to be pressure-tested, not a theorem claimed. ISS is a specific, provable input–output property with precise gain conditions; RST-WIN is a substrate-general admissibility statement. The shape matches; whether the machinery can be made to correspond precisely — and what an ISS-style gain function would become once the disturbance and the state are no longer separated by an external boundary — is open work, and a natural place for a formal pass.

What the framework does not claim. It does not claim to improve on control theory as engineering — for the exogenous-controller case, control theory is the more complete and more quantitative account, and nothing here competes with it. It does not assert that ISS and RST-WIN are the same theorem; they rhyme, and the page says so rather than smuggling in a correspondence it has not established. And it does not present “a controlled system is an attractlet” as settled canon: it is a strong structural reading, offered for the adversarial pass it deserves. The falsifiability tell keeps it honest — if there were a controlled system that is genuinely endogenous, producing its own controller, setpoint, and boundary with nothing supplied from outside, then the “controlled ⇒ attractlet” reading would break, and that system would be sovereign, not controlled. The claim lives or dies on whether the controller is exogenous.

A boundary the framework is careful not to overstep. It would be too much to say that RST-WIN is ISS, or that every controlled system is an attractlet as a matter of proof. ISS is a description in the language of gains and bounds; the framework's window-persistence is a structural admissibility statement; and “controlled ⇒ attractlet” is a reading that holds exactly as far as the controller is genuinely exogenous. They rhyme — control theory is the closest classical formalization of the framework's stability and window instincts — but they are not the same objects, and asserting an identity would import a formal correspondence the framework has not yet earned. The honest statement is the modest one: the stability the framework relies on is classically real, control theory is where it was made to make systems behave, and the framework's addition is the who-supplies-the-control question, not a claim to have re-derived control theory.

Where this sits, and what it leaves open. This page treats a neighbor discipline rather than a phenomenon: it records the framework's debt to control theory as the mature science of feedback and stability, and its one departure — the who-supplies-the-control question. Threads deliberately left open:

ISS ≈ RST-WIN, made precise. The strongest technical bridge on the page is left as a correspondence to be pressure-tested, not a theorem. Turning the rhyme into a statement — or finding exactly where it fails once the external boundary between disturbance and state is removed — is a natural formal pass, and a good candidate for a framework memo before anything is asserted.
Controlled ⇒ attractlet, adversarially tested. The cleanest divergence is also the strongest claim; it deserves an adversarial search for a genuinely endogenous “controlled” system, which would break it. Framed here, resolved elsewhere.
Companion neighbors. This page sits beside the framework's other discipline neighbors — autopoiesis (the closest), bifurcation theory (who turns the knob?), and viability theory (survival under constraint) — and beside the phenomenon pages that share its instincts: the dynamical-systems canon for the stability lineage, the attractlet page for the exogenous-control contrast, and sequence, not time for the demotion of the time index that control theory keeps.

None of this touches the framework's fixed foundations; the six substrate conditions are closed and are not at issue on this page. The control-theory citations are marked [verify] and must be confirmed against the primary sources before publication.

Adjacent-work assessments state where Principia Attractum agrees with and departs from neighboring frameworks. They introduce no constructs and modify no canon; they locate the framework relative to its field. The reading of control theory as the mature science of exogenous feedback, and of the framework's departure as the who-supplies-the-control question, are downstream applications of the framework's existing distinctions, not additions to it. The framework asserts no formal identity between input-to-state stability and its window-persistence condition, and offers “a controlled system is an attractlet” as a structural reading to be pressure-tested rather than settled canon. Citations of the control-theory literature are marked [verify] and must be confirmed against the primary sources before publication.

← Back to Relation to Adjacent Work