Fibre Inflation in String Cosmology

• Fibre Inflation is a string inflation framework where the fibre modulus in a fibred Calabi–Yau geometry drives inflation via a Starobinsky-like plateau potential.
• The models stabilize the overall volume and blow-up moduli using LARGE Volume Scenario techniques and non-perturbative as well as α'^3 corrections, while the flat fibre direction is lifted by string-loop effects.
• Recent extensions include chiral global embeddings, perturbative LVS, and assisted multi-fibre dynamics, which refine predictions for spectral indices, tensor-to-scalar ratios, and reheating scenarios.

Fibre Inflation is a class of type IIB string inflation models in which the inflaton is a Kähler modulus measuring the size of a K3 or $T^4$ fibre inside a fibred Calabi–Yau threefold. In its standard LARGE Volume Scenario (LVS) realization, $\alpha'^3$ corrections and non-perturbative effects stabilize the overall volume and a blow-up divisor, while the fibre direction remains flat at leading order and is lifted by subleading string-loop effects, producing a plateau potential of Starobinsky/$\alpha$-attractor type (0808.0691). Subsequent developments have extended the framework to chiral global embeddings, perturbative-LVS constructions, assisted multi-fibre dynamics, reheating and dark-radiation analyses, and a reappraisal of the apparent isocurvature instability of ultra-light axions (Cicoli et al., 2017).

1. Geometric definition and core structure

The original construction is based on a K3-fibred Calabi–Yau with volume

$$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$

where $\tau_1$ is the K3 or $T^4$ fibre modulus, $\tau_2$ is a base modulus, and $\tau_3$ is a blow-up cycle (0808.0691). Later presentations use the equivalent schematic form

$$\mathcal{V}\sim \lambda_f\,\tau_b\sqrt{\tau_f}-\lambda_s\,\tau_s^{3/2},$$

or $$\mathcal{V}\sim \alpha\sqrt{\tau_f}\,\tau_b-\sum_i\gamma_i\tau_{s_i}^{3/2}$$, with $\alpha'^3$0 the fibre modulus, $\alpha'^3$1 the base modulus, and $\alpha'^3$2 one or more blow-up divisors (Bera et al., 2024).

This geometry is not incidental. The fibred volume structure leaves one Kähler direction flat after leading LVS stabilization, and that flat direction is the fibre modulus used as the inflaton. In standard realizations, the heavy sector consists of the overall volume and the blow-up modulus, while the fibre direction is parametrically lighter because its potential arises only at subleading order (0808.0691).

The broader geometric setting is now understood more systematically. A later classification study argues that Fibre Inflation belongs to a wider family of LVS inflationary constructions that can coexist on the same Calabi–Yau threefold, provided the geometry contains a K3- or $\alpha'^3$3-fibration, two diagonal del Pezzo divisors, and a Wilson divisor. In that framework, Fibre Inflation, Poly-instanton inflation, and (Loop) Blow-up inflation become different orientifold or brane realizations of the same underlying geometry (Shukla, 9 Jun 2026).

2. Effective supergravity description and the inflaton potential

In standard LVS Fibre Inflation, the 4D $\alpha'^3$4 data take the familiar form

$\alpha'^3$5

with fluxes stabilizing complex-structure moduli and the axio-dilaton, and non-perturbative effects on a rigid blow-up divisor stabilizing the LVS sector (Kallosh et al., 2017). After this leading stabilization, the fibre direction remains flat because the dominant potential depends only on $\alpha'^3$6 and the blow-up modulus.

The inflaton potential is then generated by string-loop corrections and, in later variants, higher-derivative $\alpha'^3$7 terms. In the original model the inflationary regime is well described by

$\alpha'^3$8

with microscopic parameters entering essentially through the overall scale $\alpha'^3$9 (0808.0691). A more complete form used in later analyses is

$\alpha$0

obtained when the steepening term proportional to $\alpha$1 is negligible on the observable plateau (Cicoli et al., 2021).

This potential admits a direct interpretation in the language of $\alpha$2-attractors. Fibre Inflation was shown to realize plateau models with discrete effective values

$\alpha$3

depending on which fibre/base direction dominates the effective single-field reduction and on the leading correction that lifts the flat direction (Kallosh et al., 2017). The canonical field is logarithmic in the fibre modulus, so power-law dependence in $\alpha$4 becomes exponential in $\alpha$5, explaining the ubiquity of plateau potentials.

A distinct development replaces standard LVS by perturbative LVS (pLVS), where the volume is stabilized without a non-perturbative superpotential. In that setting one takes $\alpha$6 and

$\alpha$7

so that BBHL $\alpha$8 and logarithmic one-loop corrections stabilize $\alpha$9 perturbatively (Bera et al., 2024). This change is central in later global embeddings because it removes the need for rigid del Pezzo divisors in the inflationary geometry and thereby relaxes the strongest Kähler-cone field-range bounds.

3. Inflationary dynamics, observables, and reheating

The hallmark of the original model is that slow roll is governed primarily by the field value rather than by continuous tuning of shape parameters. In the plateau regime,

$$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$0

so $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$1 and $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$2 are strongly correlated once the number of e-folds is specified (0808.0691). For a simple reheating history with $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$3, the original construction gave

$$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$4

(0808.0691).

A complementary reheating analysis used an effective equation-of-state parameter $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$5 and the approximate plateau relation

$$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$6

showing that the reheating curves of Fibre Inflation are very close to those of $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$7-attractors with $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$8 (Cabella et al., 2017). That study found that if future data were to prefer $$\mathcal{V}=\alpha\left(\sqrt{\tau_1}\,\tau_2-\gamma\,\tau_3^{3/2}\right),$$9, then larger values $\tau_1$0 would be favored, whereas $\tau_1$1 would instead favor $\tau_1$2 (Cabella et al., 2017).

A more microscopic reheating analysis computed inflaton decays to ultra-light bulk axions and visible-sector fields on D7-branes wrapping the inflaton divisor. It found a reheating temperature of order $\tau_1$3, fixing

$\tau_1$4

In the generic branch with non-zero gauge flux on the visible D7 stack, ultra-light axions contribute almost negligible dark radiation and the model gives

$\tau_1$5

If instead the flux on the inflaton divisor is turned off, the analysis yields $\tau_1$6, and fitting this branch requires $\tau_1$7 with $\tau_1$8 (Cicoli et al., 2018).

Direct fits to cosmological data sharpened this picture. A likelihood analysis using Planck 2018 and related datasets identified the preferred branch as the one with small dark radiation, obtaining

$\tau_1$9

and derived values

$T^4$0

for Planck-only fits (Cicoli et al., 2020). Large-dark-radiation branches were found to alleviate the Hubble tension somewhat but to be statistically disfavored relative to $T^4$1CDM (Cicoli et al., 2020).

4. Global embeddings, Kähler-cone constraints, and field-range bounds

A persistent issue in Fibre Inflation has been the transition from an effective single-field potential to a fully global compactification with all tadpoles, branes, orientifolds, and cone constraints implemented. A chiral global embedding demonstrated that this is possible in a type IIB orientifold with $T^4$2, a del Pezzo divisor for non-perturbative stabilization, multiple K3 fibrations, a fluxed D7-brane sector generating chirality, and a D-term condition that reduces the effective volume to the standard fibre-inflation form (Cicoli et al., 2017). That construction also showed explicitly that Kähler-cone conditions impose strong constraints on the allowed inflaton field range (Cicoli et al., 2017).

This field-range problem became especially clear in later global studies. In standard LVS K3-fibred models with a rigid del Pezzo divisor, the Kähler cone typically induces an upper bound on the fibre modulus such that the canonically normalized range is much smaller than the $T^4$3 usually required for the plateau. An explicit example gives $T^4$4, far below the needed range (Bera et al., 2024). This is the main obstruction to global single-field Fibre Inflation in Swiss-cheese compactifications with rigid divisors.

Perturbative LVS offers a concrete way around this obstruction. An explicit pLVS embedding on a K3-fibred Calabi–Yau with $T^4$5 and toroidal-like volume

$T^4$6

has a simple Kähler cone

$T^4$7

with no upper bound on the fibre modulus analogous to the standard LVS case (Bera et al., 2024). This suggests that the sharp field-range obstruction is tied less to fibre inflation per se than to the particular global geometry used for standard LVS stabilization.

The geometric classification has also widened. A scan over roughly $T^4$8 Calabi–Yau geometries and around a million toric divisors found two candidate threefolds with the required unified divisor structure for $T^4$9, together with 14 candidates for $\tau_2$0 and 45 for $\tau_2$1 (Shukla, 9 Jun 2026). This indicates that globally viable fibre-inflation geometries are sparse but not isolated.

5. Multifield structure, perturbations, and the “fake instability”

Although Fibre Inflation is often treated as effectively single-field, the underlying light sector generally contains ultra-light axions kinetically coupled to the fibre modulus. In the two-field subsector $\tau_2$2, the effective metric is curved, and the field-space Ricci scalar can be a negative constant,

$\tau_2$3

which naively lowers the isocurvature mass and can make the standard entropy mode appear tachyonic (Cicoli et al., 2021).

This led to claims of a geometrical instability. The later resolution is that the divergence occurs in an ill-defined entropy variable. On the slow-roll attractor the axion velocity decays as

$\tau_2$4

so the background trajectory rapidly aligns with the inflaton direction and the normal vector used in the standard entropy decomposition becomes degenerate (Cicoli et al., 2021). When the physically relevant relative entropy perturbation $\tau_2$5, defined via the non-adiabatic pressure, is used instead, one finds

$\tau_2$6

on the attractor, and the axionic isocurvature perturbations decay during inflation (Cicoli et al., 2021). The conclusion is that the apparent tachyonic instability is a coordinate artifact, not a physical destabilization.

A different multifield issue concerns the end of the single-field regime. Fibre Inflation was related to $\tau_2$7-attractors in a way that also clarifies its limitations: above a certain field range, higher-order corrections steepen the potential, slow roll breaks down, and the overall volume begins to participate in the dynamics, so the effective description becomes genuinely two-field (Kallosh et al., 2017). Observable inflation is assumed to occur well before this turnover.

The model has also been invoked in attempts to explain low-$\tau_2$8 CMB power loss by a fast-roll onset. That analysis concluded that the original loop-generated fibre-inflation potential is not steep enough outside the plateau to suppress large-scale power in the observable window. A steeper exponential contribution,

$\tau_2$9

can produce the effect if appropriately tuned, but this contribution was not derived from the original string construction (Pedro et al., 2013). Thus low-$\tau_3$0 suppression is not a generic prediction of Fibre Inflation.

6. Perturbative-LVS, assisted multi-fibre inflation, and recent extensions

The most substantial recent shift is from standard LVS to perturbative LVS and from single-field to assisted multi-fibre dynamics. In a concrete pLVS model with three K3 divisors and volume

$\tau_3$1

all three Kähler moduli are fibre-like, and after volume stabilization two light shape directions remain. In the canonical basis the total displacement satisfies

$\tau_3$2

while the assisted relation is

$\tau_3$3

For a benchmark two-fibre trajectory, one finds

$\tau_3$4

together with

$\tau_3$5

(Leontaris et al., 27 Jun 2025). The key point is not a smaller effective displacement, but that each individual modulus moves less than in the single-field case.

A closely related global-embedding study summarizes the motivation by noting that standard single-field fibre inflation typically requires a trans-Planckian excursion of order $\tau_3$6, while explicit global embeddings often run into Kähler-cone bounds before the full plateau can be traversed. Assisted fibre inflation shares this burden among several fibre moduli so that successful inflation can occur before any individual modulus approaches its cone boundary (Leontaris et al., 21 Apr 2026). This also weakens the tension with Swampland Distance Conjecture reasoning, since the effective large displacement is distributed across several mid-range motions rather than along a single modulus direction (Leontaris et al., 21 Apr 2026).

A different pLVS extension introduces a one-loop redefinition of the base modulus rather than the fibre modulus. In that construction, the leading flat direction is still the fibre modulus, but the potential is modified by the base redefinition together with string loops, BBHL $\tau_3$7, and $\tau_3$8 corrections. The resulting inflationary scenarios produce spectral indices aligned with ACT data and tensor-to-scalar ratios in the range

$\tau_3$9

(Chakraborty et al., 24 Nov 2025). The same framework then extends the axionic sector to late-time quintessence through poly-instanton corrections, linking early- and late-time cosmic acceleration in a common pLVS setting (Chakraborty et al., 24 Nov 2025).

Taken together, these developments support a more differentiated view of Fibre Inflation. The original single-fibre LVS model remains the canonical large-field construction with $$\mathcal{V}\sim \lambda_f\,\tau_b\sqrt{\tau_f}-\lambda_s\,\tau_s^{3/2},$$0 and $$\mathcal{V}\sim \lambda_f\,\tau_b\sqrt{\tau_f}-\lambda_s\,\tau_s^{3/2},$$1, but the modern literature treats it less as a unique model than as a family of closely related mechanisms: standard LVS, perturbative LVS, chiral global embeddings, assisted multi-fibre dynamics, and hybrid early-/late-universe extensions. The common structural element across these variants is unchanged: inflation is driven by a fibre Kähler modulus whose potential is protected at leading order by the fibred geometry and generated only by subleading string corrections.