Binding Constraint Thesis

• Binding Constraint Thesis is the idea that binding interactions serve as inherent constraints in formal systems, directly determining solution feasibility across various domains.
• It is applied in AdS EFT, causal modeling, Yang-Mills, optimal control, and polymer physics where intrinsic constraints replace post hoc corrections.
• Analytic and simulation methods—such as solving binding energy equations and using disconnect interventions—demonstrate its role in ensuring consistency and optimality.

Searching arXiv for the term and closely related usages to ground the article in current papers. The expression Binding Constraint Thesis is used in several technical literatures for claims that a binding relation is not an auxiliary correction but a constitutive part of the admissible theory. In Anti de-Sitter effective field theory, the phrase is proposed explicitly as the statement that any low-energy EFT in AdS$_d$ admitting a U(1) must contain at least one charged state whose total self-binding energy, including contact, gauge, graviton, and scalar exchanges, is non-negative (Andriolo et al., 2022). In causal modeling, an analogous thesis holds that non-causal binding equations must enter the model through an explicit constraint set and a distinct intervention semantics (2301.06845). In Yang-Mills theory, a related formulation states that the static heavy-quark binding energy is determined entirely by the exact non-perturbative solution of Gauss’s law in first-order formalism (Wilson-Gerow, 2020). Across these uses, the common structure is that feasibility, binding, or interaction energy is fixed by a constraint embedded inside the formal system rather than imposed after the fact.

1. Domain of use and core idea

In the optimization literature, a binding or active box constraint for a scalar control

$u(t)\in [u_{\min},u_{\max}]$

is present at time $t$ precisely when

$$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$

equivalently on subintervals where the unconstrained first-order condition would push the control outside the admissible interval and the optimizer “sticks” to the boundary (Wang, 7 Apr 2026). This definition provides a useful mathematical baseline: a binding constraint is one that directly determines the realized solution.

That baseline reappears, with different semantics, in several fields. In AdS EFT, the relevant object is a lower bound on the total self-binding energy of at least one charged state. In causal modeling, the relevant object is a global set $C$ of admissible extended states that enforces equations such as $TOT=LDL+HDL$. In Yang-Mills theory, the relevant object is the Gauss-law constraint $(D_iE^i)^a-g\rho^a=0$, whose solution fixes the binding potential. In confined-polymer physics, a severe configurational constraint can induce binding that is absent in bulk (Andriolo et al., 2022, 2301.06845, Wilson-Gerow, 2020, Fraser et al., 2019).

Domain	Constraint or binding object	Claimed role
AdS EFT	$E_{\rm bind}\ge 0$ for some charged state	consistency criterion and Swampland test
Causal models	constraint set $C$ and disconnect intervention	admissible states and interventions
Yang-Mills	Gauss law $(D_iE^i)^a-g\rho^a=0$	full source of static $u(t)\in [u_{\min},u_{\max}]$0 binding energy
Optimal control	active box constraint	determines boundary arcs of the optimizer
Polymer confinement	narrow-tube configurational restriction	stimulates chain binding

A plausible unifying implication is that the thesis is less a single theorem than a family of formalisms asserting that a constraint must be enforced intrinsically, at the level of the governing equations.

2. AdS effective field theory and the positive self-binding condition

In AdS, the explicit “Binding Constraint Thesis” is formulated as an extension of Weak Gravity Conjecture reasoning. The Positive Binding Conjecture states: in any consistent theory of gravity in AdS with a U(1) gauge symmetry there must exist at least one charged particle whose self-binding energy is non-negative (Andriolo et al., 2022). Physically, the net effect of all long-range interactions between two identical copies of that particle in AdS must therefore be either repulsive or exactly marginal. The paper further states that this is conjecturally necessary both for quantum-gravity consistency and to ensure convexity properties of charged operator dimensions in the dual CFT.

For a complex scalar field $u(t)\in [u_{\min},u_{\max}]$1 in global AdS$u(t)\in [u_{\min},u_{\max}]$2, with gravity, gauge fields, a neutral scalar, and contact interactions, the self-binding energy is defined by

$u(t)\in [u_{\min},u_{\max}]$3

Equivalently,

$u(t)\in [u_{\min},u_{\max}]$4

At leading order in weak couplings,

$u(t)\in [u_{\min},u_{\max}]$5

The decomposition is operational rather than merely schematic: each contribution is obtained by solving the linearized equations of motion for the mediator sourced by $u(t)\in [u_{\min},u_{\max}]$6, substituting back into the action, and performing the overlap integral.

The paper gives analytic expressions in AdS$u(t)\in [u_{\min},u_{\max}]$7 and AdS$u(t)\in [u_{\min},u_{\max}]$8. In AdS$u(t)\in [u_{\min},u_{\max}]$9, with $t$0, the contact, photon, and graviton contributions are written explicitly, and the scalar exchange is solved analytically for the massless and BF-bound cases. In AdS$t$1, with $t$2, an analytic formula is likewise given for the massless scalar contribution. The flat-space limit $t$3, with $t$4 fixed, reproduces the known Minkowski-space self-binding formulas exactly.

Two features are especially significant. First, unlike the flat-space case, even a massive scalar contributes significantly to the binding energy in AdS, so scalar exchange is an essential part of the constraint. Second, in explicit supersymmetric AdS$t$5 and AdS$t$6 examples, the total binding vanishes for BPS states, while a non-BPS scalar need not saturate the conjecture. For generic scalar mass $t$7, the scalar contribution is obtained numerically by shooting from the AdS center to the boundary; for fixed $t$8, the resulting $t$9 is smooth and monotonically increasing, interpolating between the BF-bound value and a decoupling regime at large $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$0 (Andriolo et al., 2022).

Within this AdS usage, the thesis is therefore a concrete EFT consistency condition: $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$1 Violations are proposed to signal a would-be EFT in the Swampland.

3. Extended causal models and non-causal binding equations

In causal inference, the corresponding thesis is that causal domains often contain relations that are not themselves causal equations but are nonetheless binding. Standard structural causal models do not permit such relations straightforwardly, because one can intervene simultaneously on all endogenous variables, including variables linked by a part-whole identity such as

$$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$2

The extension proposed in “Causal Models with Constraints” replaces this limitation by adding an explicit constraint set $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$3 to the model (2301.06845).

A standard SCM over a signature $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$4 is $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$5. An extended causal model with constraints is instead

$$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$6

where $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$7 is only a partial set of structural functions and

$$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$8

is the set of admissible extended states $$u^*(t)=u_{\min}\quad\text{or}\quad u^*(t)=u_{\max},$$9. Constraint formulas may be written as

$C$0

with the intended semantics that

$C$1

whenever $C$2.

The formal novelty is the disconnect intervention. Given $C$3 and $C$4, the disconnected model $C$5 removes any structural functions for variables in $C$6. One then combines this with ordinary interventions. If $C$7 and $C$8 are disjoint and $C$9, then

$TOT=LDL+HDL$0

means: disconnect all $TOT=LDL+HDL$1, set $TOT=LDL+HDL$2, and evaluate $TOT=LDL+HDL$3 in the resulting constrained model. Semantically,

$TOT=LDL+HDL$4

iff for all $TOT=LDL+HDL$5 such that $TOT=LDL+HDL$6 and $TOT=LDL+HDL$7 satisfies the modified structural equations, $TOT=LDL+HDL$8 is true in $TOT=LDL+HDL$9.

The LDL–HDL–TOT example makes the role of the binding constraint explicit. With

$(D_iE^i)^a-g\rho^a=0$0

the intervention

$(D_iE^i)^a-g\rho^a=0$1

disconnects $(D_iE^i)^a-g\rho^a=0$2, sets $(D_iE^i)^a-g\rho^a=0$3, retains the equation for $(D_iE^i)^a-g\rho^a=0$4, and enforces the binding equation $(D_iE^i)^a-g\rho^a=0$5, so that

$(D_iE^i)^a-g\rho^a=0$6

This is not representable by ordinary simultaneous do-operations on all three variables.

The paper also supplies an axiomatization. Let $(D_iE^i)^a-g\rho^a=0$7 be $(D_iE^i)^a-g\rho^a=0$8–$(D_iE^i)^a-g\rho^a=0$9, $E_{\rm bind}\ge 0$0–$E_{\rm bind}\ge 0$1, $E_{\rm bind}\ge 0$2, plus the new axiom

$E_{\rm bind}\ge 0$3

together with modus ponens. The theorem of Beckers–Halpern–Hitchcock states that $E_{\rm bind}\ge 0$4 is a sound and complete axiomatization of the modal language with both standard do-interventions and disconnect-interventions over all finite causal models with constraints (2301.06845).

A common misconception addressed by this framework is that adding constraints is a superficial extension of SCMs. The paper’s position is stronger: some interventions become impossible because no solutions exist, and faithful intervention semantics therefore requires an explicit distinction between causal equations and global binding conditions.

4. Yang-Mills Gauss law as a binding constraint on heavy-quark energy

In non-Abelian gauge theory, the “Binding Constraint Thesis” is formulated as the statement that the full static $E_{\rm bind}\ge 0$5 binding energy follows from the exact non-perturbative solution of Gauss’s law in first-order path-integral formalism (Wilson-Gerow, 2020). The starting point is the SU($E_{\rm bind}\ge 0$6) path integral with independent chromoelectric field $E_{\rm bind}\ge 0$7: $E_{\rm bind}\ge 0$8 Because $E_{\rm bind}\ge 0$9 appears linearly, integrating it out imposes the exact constraint

$C$0

or, equivalently, the functional delta

$C$1

To analyze the constraint, the paper uses the generalized gauge-covariant Coulomb gauge

$C$2

and decomposes the chromoelectric field into gauge-covariant transverse and longitudinal parts,

$C$3

After gauge fixing and integrating out $C$4, the static potential is generated solely by the solution of

$C$5

and the effective Hamiltonian becomes

$C$6

The paper emphasizes that no semiclassical approximation is involved: the equation is an exact operator statement, and the binding energy is determined entirely by its solution.

Assuming that low-energy physics is dominated by a dimension-2 gluon condensate, with

$C$7

the covariant Laplacian acquires an effective mass scale $C$8. The potential is then expanded as

$C$9

with a two-step recursion. Averaging over condensate directions and color states produces an interquark potential of the form

$(D_iE^i)^a-g\rho^a=0$0

The resulting string tension is

$(D_iE^i)^a-g\rho^a=0$1

For SU(3),

$(D_iE^i)^a-g\rho^a=0$2

since $(D_iE^i)^a-g\rho^a=0$3 and $(D_iE^i)^a-g\rho^a=0$4.

Here the thesis is especially strict: the relevant constraint is not a bound on admissible couplings but the exact equation from which the entire binding energy is said to follow.

5. Intrinsic enforcement in optimization and rule-based constraint systems

A recurring methodological point in constraint-based theories is that the constraint must be incorporated inside the optimality or transition system. In box-constrained optimal control, Pontryagin’s maximum principle for the Bolza problem requires pointwise maximization over the compact admissible set: $(D_iE^i)^a-g\rho^a=0$5 When $(D_iE^i)^a-g\rho^a=0$6 is $(D_iE^i)^a-g\rho^a=0$7 in $(D_iE^i)^a-g\rho^a=0$8, this is equivalent to the sign-complementarity rules

$(D_iE^i)^a-g\rho^a=0$9

For a strictly concave quadratic Hamiltonian in a scalar control,

$u(t)\in [u_{\min},u_{\max}]$00

the unconstrained optimizer is $u(t)\in [u_{\min},u_{\max}]$01, and the exact constrained maximizer is

$u(t)\in [u_{\min},u_{\max}]$02

The note distinguishes intrinsic projection, where the clamp appears directly in the two-point boundary-value problem, from post hoc truncation, where an unconstrained solution is clipped afterwards. The latter generally violates adjoint-equation consistency, changes the state trajectory, and need not satisfy first-order optimality, whereas the former preserves feasibility at every iteration (Wang, 7 Apr 2026).

An analogous insistence on intrinsic constraint handling appears in “Constraint Handling Rules with Binders, Patterns and Generic Quantification,” where ordinary first-order CHRs are extended so that binding structure is explicit through $u(t)\in [u_{\min},u_{\max}]$03-tree syntax and fresh constants are introduced by the generic quantifier $u(t)\in [u_{\min},u_{\max}]$04 (Serrano et al., 2017). Terms are considered modulo $u(t)\in [u_{\min},u_{\max}]$05-equivalence, restricted $u(t)\in [u_{\min},u_{\max}]$06-reduction, and the $u(t)\in [u_{\min},u_{\max}]$07-law; rule heads are restricted to patterns to preserve decidable matching. In the $u(t)\in [u_{\min},u_{\max}]$08-style operational semantics, a state has the form

$u(t)\in [u_{\min},u_{\max}]$09

and the Apply rule combines higher-order pattern unification, freshening of $u(t)\in [u_{\min},u_{\max}]$10-variables to new nominal constants, and propagation-history control. The paper states that Miller’s $u(t)\in [u_{\min},u_{\max}]$11-unification succeeds in linear time and proves declarative-semantics soundness, local confluence via joinable critical pairs, and termination by a level-mapping argument.

This suggests a broader methodological interpretation of the thesis: constraint satisfaction is not merely an output filter; it is part of the generative mechanism that defines admissible solutions.

6. Constrained binding in confined polymers and open technical limits

A physically distinct but structurally related use of binding-by-constraint appears in the study of polymer chains confined to narrow tubes (Fraser et al., 2019). There, two identical Gaussian chains that would not bind in bulk can bind under severe configurational restriction because chain entropy is reduced in the narrow channel. For chains of $u(t)\in [u_{\min},u_{\max}]$12 segments in a tube of diameter $u(t)\in [u_{\min},u_{\max}]$13, the free-energy difference between a state with $u(t)\in [u_{\min},u_{\max}]$14 bonded end monomers and the unbound state is written as

$u(t)\in [u_{\min},u_{\max}]$15

In the purely equilibrium picture, because $u(t)\in [u_{\min},u_{\max}]$16 is strictly linear in $u(t)\in [u_{\min},u_{\max}]$17, the optimum is only

$u(t)\in [u_{\min},u_{\max}]$18

with no finite interior solution. The observed finite overlap instead arises from a kinetic competition between reptation and bond formation, giving

$u(t)\in [u_{\min},u_{\max}]$19

with $u(t)\in [u_{\min},u_{\max}]$20 defined in terms of $u(t)\in [u_{\min},u_{\max}]$21, $u(t)\in [u_{\min},u_{\max}]$22, the free-chain diffusion constant $u(t)\in [u_{\min},u_{\max}]$23, and the per-monomer bond-formation rate constant $u(t)\in [u_{\min},u_{\max}]$24. The onset diameter obeys

$u(t)\in [u_{\min},u_{\max}]$25

Brownian-dynamics simulations with two coarse-grained Kremer-Grest chains, reflecting cylindrical confinement, $u(t)\in [u_{\min},u_{\max}]$26, periodicity along the tube axis, and tube diameters $u(t)\in [u_{\min},u_{\max}]$27 show measured overlaps $u(t)\in [u_{\min},u_{\max}]$28, respectively, and binding-time distributions with heavy tail $u(t)\in [u_{\min},u_{\max}]$29 (Fraser et al., 2019). The stated biological implication is that narrow channels of order $u(t)\in [u_{\min},u_{\max}]$30–$u(t)\in [u_{\min},u_{\max}]$31 nm could support entropically stimulated head-to-tail locking of unfolded protein chains.

Across the literatures surveyed here, open limits remain domain-specific. In causal models, the finiteness assumption simplifies the axiom system, and extensions to real-valued variables, approximate constraints, learnability, and analogues of do-calculus remain open (2301.06845). In AdS binding, generic scalar masses require numerical treatment rather than closed-form expressions (Andriolo et al., 2022). In confined polymers, heteropolymer disorder, wall interactions, electrostatics, crowding, and non-Gaussian statistics are not included (Fraser et al., 2019). These limits do not negate the thesis; rather, they specify where intrinsic constraint enforcement is already formalized and where it remains incomplete.