Lattice Weak Gravity Conjecture

• Lattice Weak Gravity Conjecture (LWGC) is a refined version of the Weak Gravity Conjecture, requiring a superextremal state at every point in the charge lattice to ensure quantum-gravity consistency.
• It employs techniques from toroidal compactification, spectral flow, and perturbative string constructions to demonstrate that an infinite tower or finite-index sublattice of superextremal states exists.
• Variants like the Sublattice Weak Gravity Conjecture address counterexamples by demanding superextremal states on sublattices, linking gauge coupling cutoffs to implications for axion inflation and black-hole discharge.

The Lattice Weak Gravity Conjecture (LWGC) is a sharpened form of the Weak Gravity Conjecture that requires a superextremal state at every site in the full charge lattice of a quantum-gravity theory. In the formulation of Heidenreich–Reece–Rudelius, if $\Delta$ is the full charge lattice and $|q|=\sqrt{q_i g^{ij} q_j}$ is the gauge-invariant norm defined by the low-energy gauge kinetic matrix, then for every lattice vector $q\in\Delta$ there exists a particle state of charge $q$ and mass $m(q)$ such that $m(q)\le M_{\rm Pl}|q|$, equivalently $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$ (Heidenreich et al., 2016). The conjecture was proposed in part because the ordinary WGC is not robust under toroidal compactification, whereas a lattice statement is (Heidenreich et al., 2015). Subsequent work showed that the strongest “full lattice” version fails in certain orbifold and string vacua, motivating the Sublattice Weak Gravity Conjecture (sLWGC), which requires superextremal states only on a finite-index sublattice of the full charge lattice (Heidenreich et al., 2016).

1. Definition and variants

For an Abelian gauge group $G=U(1)^r$, the electric charges take values in a lattice, denoted $\Delta$, $\Lambda$, or $|q|=\sqrt{q_i g^{ij} q_j}$0 in different sources, with $|q|=\sqrt{q_i g^{ij} q_j}$1 or $|q|=\sqrt{q_i g^{ij} q_j}$2. In the notation of (Heidenreich et al., 2016), the low-energy gauge kinetic function defines a positive-definite metric $|q|=\sqrt{q_i g^{ij} q_j}$3 on charge space, and the corresponding norm is

$|q|=\sqrt{q_i g^{ij} q_j}$4

Semiclassical extremal black holes of charge $|q|=\sqrt{q_i g^{ij} q_j}$5 satisfy

$|q|=\sqrt{q_i g^{ij} q_j}$6

so the LWGC demands

$|q|=\sqrt{q_i g^{ij} q_j}$7

In a simple one-$|q|=\sqrt{q_i g^{ij} q_j}$8 theory with gauge coupling $|q|=\sqrt{q_i g^{ij} q_j}$9, one often writes $q\in\Delta$0, so $q\in\Delta$1 and the condition reduces to $q\in\Delta$2 (Heidenreich et al., 2016).

The sublattice refinement replaces the full lattice by a finite-index sublattice. In the formulation of (Heidenreich et al., 2016), there exists a sublattice $q\in\Delta$3 of finite index $q\in\Delta$4 such that for every charge vector $q\in\Delta$5 there is a single-particle state whose mass obeys

$q\in\Delta$6

in Planck units. The same parameter $q\in\Delta$7 is often called the coarseness. Equivalently, one may state that every charge $q\in\Delta$8 has a superextremal multiple $q\in\Delta$9 (Heidenreich et al., 2016).

A nonabelian extension is obtained by choosing a Cartan subalgebra $q$0 and taking the electric charge lattice to be the weight lattice $q$1. In that setting, the nonabelian LWGC requires that for each weight $q$2 there be a superextremal state transforming in the irreducible representation with highest weight $q$3, with extremality compared schematically to $q$4 (Reece et al., 4 Mar 2026).

2. Motivation from compactification, black holes, and global symmetries

A principal motivation for the LWGC is the failure of the ordinary WGC under circle compactification. In the analysis of (Heidenreich et al., 2015), compactifying a $q$5-dimensional theory on $q$6 produces both the original photon and a Kaluza–Klein (KK) photon. The KK graviton modes themselves marginally satisfy $q$7, but their charge-to-mass vectors are too sparse for the convex hull of a finite set of states to contain the full extremality ellipse in the reduced theory. The result is that the ordinary WGC in $q$8 dimensions does not guarantee the WGC in $q$9 dimensions unless either there is a lower cutoff on the compactification radius or the higher-dimensional theory already contains an infinite tower of superextremal states of all charges. This observation led directly to the lattice formulation (Heidenreich et al., 2015).

The same line of reasoning is tied to black-hole discharge. Extremal or slightly superextremal black holes of charge $m(q)$0 and mass $m(q)$1 should be able to spontaneously discharge by emitting a particle of the same charge. If no such particle satisfies $m(q)$2, then large-$m(q)$3 black holes would be absolutely stable and yield remnants (Heidenreich et al., 2016).

The absence of continuous global symmetries provides a complementary motivation. As $m(q)$4, a $m(q)$5 becomes indistinguishable from a global symmetry. The LWGC makes the expected obstruction precise by asserting that an infinite tower of charged states becomes light, with masses $m(q)$6 for integer $m(q)$7, so the effective theory collapses in the limit (Heidenreich et al., 2016).

A particularly simple realization occurs in pure gravity compactified on a circle. In $m(q)$8 compactification, KK modes carry integer momentum $m(q)$9, viewed in $m(q)\le M_{\rm Pl}|q|$0 dimensions as electric charge $m(q)\le M_{\rm Pl}|q|$1 under the graviphoton, with masses

$m(q)\le M_{\rm Pl}|q|$2

The classical extremality condition for the corresponding charged black holes is

$m(q)\le M_{\rm Pl}|q|$3

so the elementary KK mode at charge $m(q)\le M_{\rm Pl}|q|$4 exactly saturates the extremality bound. In this simplest toroidal compactification, every lattice site is populated and one has $m(q)\le M_{\rm Pl}|q|$5 (Heidenreich et al., 2016).

3. Realizations in perturbative string theory and M-theory

Perturbative string theory supplied some of the earliest evidence for lattice population. In ten-dimensional $m(q)\le M_{\rm Pl}|q|$6 heterotic theory, the Cartan subalgebra is $m(q)\le M_{\rm Pl}|q|$7 and the charge lattice is the even spin-weight lattice

$m(q)\le M_{\rm Pl}|q|$8

with $m(q)\le M_{\rm Pl}|q|$9. Level matching forces the lightest state of charge $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$0 to obey

$Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$1

and therefore

$Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$2

for every nonzero $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$3. In this sense the perturbative heterotic string saturates the LWGC at each lattice point (Heidenreich et al., 2015).

A more general perturbative argument uses modular invariance. In a closed-string vacuum, each NS–NS $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$4 corresponds to a conserved worldsheet current, and the flavored torus partition function with chemical potentials is quasi-periodic under shifts by the dual charge lattice. Modular invariance forces a spectral-flow symmetry

$Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$5

so acting on the vacuum generates states of arbitrarily large charge with conformal weight saturating $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$6. Level matching then gives spacetime masses

$Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$7

which exactly meets the naive extremality bound. In this way one obtains an infinite family of superextremal string states at each site of the dual lattice $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$8 (Heidenreich et al., 2016).

This perturbative picture was strengthened by a proof of a strict sublattice form in bosonic string theory. In any perturbative bosonic-string compactification of the NS–NS sector to spacetime dimension $Z(q)\equiv |q|M_{\rm Pl}/m(q)\ge1$9, there is a finite-index sublattice $G=U(1)^r$0 such that for every $G=U(1)^r$1 there exists a physical state with charge $G=U(1)^r$2 whose charge-to-mass ratio is strictly larger than that of a large extremal black hole with parallel charge. The proof combines spectral flow of the flavored partition function with a worldsheet computation of long-range self-forces, and applies at tree level in $G=U(1)^r$3 and in the two-derivative effective action (Heidenreich et al., 2024).

In M-theory on a Calabi–Yau threefold $G=U(1)^r$4, the relevant charge lattice is $G=U(1)^r$5, with electric charges carried by M2-branes wrapping holomorphic curves. The mass of a wrapped M2 is $G=U(1)^r$6, so supersymmetry gives BPS states saturating the extremality bound whenever the curve class admits a holomorphic representative (Heidenreich et al., 2016). Explicit tests based on genus-zero Gopakumar–Vafa invariants in favorable Calabi–Yau hypersurfaces with $G=U(1)^r$7 found that every integral charge in an explicitly determined cone $G=U(1)^r$8 has $G=U(1)^r$9, hence at least one BPS state with $\Delta$0, and in all examples the stronger lattice WGC held in the BPS sector (Gendler et al., 2022).

4. Counterexamples and the sublattice refinement

The strongest full-lattice version is not universal in known string vacua. A prototypical counterexample is Type II on $\Delta$1 with freely acting shifts. For one graviphoton, the KK spectrum is projected out for odd momentum $\Delta$2, so the state with minimal nonzero charge has

$\Delta$3

which can exceed the extremality bound. No lighter state of that same charge exists. Thus some lattice sites are empty of superextremal states, and in this example the lightest charged particle can be subextremal. Nevertheless, a finite-index sublattice, for example the charges with $\Delta$4 even, remains fully populated, in accord with the sLWGC. Similar phenomena occur in heterotic orbifolds with tuned radii and Wilson lines (Heidenreich et al., 2016).

These counterexamples shifted attention to the structure of LWGC failure itself. A systematic survey of effective-field-theory, string-theory, and M-theory examples found that when the LWGC fails but the sLWGC holds, one encounters a proper sublattice $\Delta$5 of superextremal electric charges together with a proper superlattice $\Delta$6 of fractionally charged monopoles. The relation

$\Delta$7

links the two. The fractionally charged monopoles cannot exist as isolated objects: they are confined by finite-tension flux tubes and deconfine only when the flux-tube tension tends to zero (Etheredge et al., 20 Feb 2025).

In nonabelian examples from heterotic toroidal orbifolds, the same pattern is organized by the global form of the gauge group. For a discrete subgroup $\Delta$8, the sublattice of superextremal electric charges is the weight lattice of the quotient group $\Delta$9, and passing to $\Lambda$0 restores the LWGC on the reduced electric lattice. In all examples considered, confined monopoles populate the magnetic lattice of $\Lambda$1, and the coarseness of LWGC violation is bounded by the maximal order of the center. The paper states that this suggests LWGC violation cannot occur for gauge groups with trivial centers (Reece et al., 4 Mar 2026).

5. Phenomenological, axionic, and mathematical implications

One of the most immediate implications of any lattice or sublattice tower is a gauge-coupling-dependent ultraviolet cutoff. Since a one-$\Lambda$2 LWGC tower has masses bounded by $\Lambda$3, an infinite number of charged states become light as $\Lambda$4, so the effective field theory must break down at or below

$\Lambda$5

In four dimensions, combining the electric sLWGC with the species bound yields a stronger gravitational cutoff estimate,

$\Lambda$6

through the scaling $\Lambda$7 together with $\Lambda$8 (Heidenreich et al., 2016, Heidenreich et al., 2016).

Axions provide a parallel “0-form” version. The axionic WGC demands an instanton of charge $\Lambda$9 and action $|q|=\sqrt{q_i g^{ij} q_j}$00 satisfying

$|q|=\sqrt{q_i g^{ij} q_j}$01

and the lattice version sets this bound on every site of the instanton charge lattice. In the original discussion, this led to the conclusion that successful single-axion large-field inflation with $|q|=\sqrt{q_i g^{ij} q_j}$02 conflicts with $|q|=\sqrt{q_i g^{ij} q_j}$03, and that multifield remedies such as $|q|=\sqrt{q_i g^{ij} q_j}$04-flation and alignment/KNP are similarly bounded once convex-hull or lattice versions are imposed, with effective field range $|q|=\sqrt{q_i g^{ij} q_j}$05 (Heidenreich et al., 2016). Later work on instanton resummation sharpened this by deriving a volume bound on the axion fundamental domain from the sLWGC, while also identifying loopholes: coherent single-axion instanton sums, KNP alignment with two species, clockwork with a dominant aligned pair, and a stretched $|q|=\sqrt{q_i g^{ij} q_j}$06-flation construction can evade the bound under the stated conditions (Heidenreich et al., 2019).

The conjecture also has direct consequences for the QCD axion and gravitational-wave phenomenology. Using the QCD instanton action $|q|=\sqrt{q_i g^{ij} q_j}$07 for $|q|=\sqrt{q_i g^{ij} q_j}$08, one obtains

$|q|=\sqrt{q_i g^{ij} q_j}$09

Observation of an axion with $|q|=\sqrt{q_i g^{ij} q_j}$10 would falsify the axionic WGC, while light bosons with such large decay constants would trigger black-hole superradiance and produce monochromatic gravitational-wave signals potentially observable by LIGO/Virgo/KAGRA (Heidenreich et al., 2016).

On the mathematical side, the LWGC has been connected to AdS/CFT and algebraic geometry. In AdS$|q|=\sqrt{q_i g^{ij} q_j}$11/CFT$|q|=\sqrt{q_i g^{ij} q_j}$12, the analogue of the particle mass is the scaling dimension of a charged operator, and the proposed bound becomes

$|q|=\sqrt{q_i g^{ij} q_j}$13

In M-theory on a Calabi–Yau threefold, the geometric LWGC statement is that every effective homology class $|q|=\sqrt{q_i g^{ij} q_j}$14 admits a holomorphic curve representative, equivalently that the full Mori cone is generated by actual holomorphic curves (Heidenreich et al., 2016).

6. Later formulations, asymptotic results, and current status

The broader literature distinguishes the full LWGC from tower and sublattice formulations. Infrared consistency arguments based on causality of photon propagation and analyticity of the S-matrix imply the convex-hull condition and, after KK compactification, an infinite tower of superextremal states that must include bifundamentals. However, the resulting “tower WGC” does not require the spectrum to occupy a full charge lattice or even a full sublattice: large charge gaps are allowed provided that sufficiently many superextremal states exist in each relevant direction of charge space (Andriolo et al., 2018).

A different refinement is asymptotic. In five-dimensional compactifications of M-theory on Calabi–Yau threefolds, weakly coupled gauge groups arise only in specific infinite-distance limits, classified as Type $|q|=\sqrt{q_i g^{ij} q_j}$15-, K3-, or $|q|=\sqrt{q_i g^{ij} q_j}$16-limits. For the corresponding weakly coupled $|q|=\sqrt{q_i g^{ij} q_j}$17 factors, every ray $|q|=\sqrt{q_i g^{ij} q_j}$18 is populated either by a tower of BPS states when $|q|=\sqrt{q_i g^{ij} q_j}$19 or by a tower of non-BPS states when $|q|=\sqrt{q_i g^{ij} q_j}$20, and in both cases the superextremality condition is satisfied in the weak-coupling regime. This establishes an Asymptotic Lattice WGC in that setting and ties the result to the Emergent String Conjecture (Cota et al., 2022).

The current status is therefore differentiated rather than uniform. Simple toroidal compactifications and several large classes of perturbative and geometric compactifications realize the full lattice statement, sometimes even with BPS saturation (Heidenreich et al., 2015, Gendler et al., 2022). Orbifold and Wilson-line examples show that the strongest form can fail (Heidenreich et al., 2016). The sLWGC has substantially broader support, including proofs in perturbative bosonic string theory and systematic evidence in KK and perturbative string constructions (Heidenreich et al., 2016, Heidenreich et al., 2024). A plausible implication is that the most robust content of the original proposal is not merely the existence of one superextremal particle, but the requirement that quantum gravity populate charge space by an infinite tower whose precise arithmetic structure depends on global properties of the gauge group, compactification data, and the ultraviolet completion.