---
title: Crosscap Coefficients in Non-Oriented Field Theories
url: https://www.emergentmind.com/topics/crosscap-coefficients
type: topic
---

# Crosscap Coefficients in Non-Oriented Field Theories

Searching arXiv for recent and foundational papers on crosscap coefficients across CFT/RCFT/XCFT and related usages.
Crosscap coefficients are quantities associated with crosscap states in conformal field theory and related non-orientable constructions. In the most standard RCFT usage, a crosscap state is expanded in crosscap Ishibashi states as \[ |C_u\rangle = \sum_i \Gamma_{ui}\, |C_i\rangle\rangle , \] and the coefficients \(\Gamma_{ui}\) are the crosscap coefficients [1811.07238]. Closely related notions appear in several adjacent literatures. In XCFT on \(\mathbb{RP}^2\), the operational crosscap data are the vacuum overlap \(p=Z_{\mathbb{RP}^2}=\langle 0|C\rangle\) and the one-point coefficients \(a_{\mathcal O}\) in \(\mathbb{RP}^2\) correlators, rather than an explicit Ishibashi expansion [2405.03755]. In CFT on \(\mathbb{RP}^d\), scalar one-point functions are written \(\langle O_i(x)\rangle = a_i/(1+x^2)^{\Delta_i}\), so \(a_i\) are intrinsic crosscap CFT data [1703.08159]. Higher-dimensional generalizations replace ordinary crosscaps by crosscap defects and promote the basic crosscap data to bulk one-point coefficients \(A_i\) in the quotient background [2604.19868]. In massive integrable QFT and integrable spin chains, the analogue of a crosscap coefficient is the overlap \(\langle \mathcal C|\Psi_L\rangle\) between a crosscap state and an energy eigenstate, with the vacuum overlap \(p=\langle \mathcal C|\Omega_L\rangle\) defining the crosscap entropy \(s_{\mathcal C}=\log|p|\) [2111.09901].

## 1. Standard RCFT formulation

In non-oriented RCFTs, crosscap states are introduced as coherent states preserving a diagonal subalgebra of the chiral symmetry. For each parity symmetry \(P_u\) with \(P_u^2=1\), the crosscap state is defined by \[ |C_u\rangle = \sum_i \Gamma_{ui}\, |C_i\rangle\rangle . \] The coefficients \(\Gamma_{ui}\) are the crosscap coefficients, and \(|C_i\rangle\rangle\) are crosscap Ishibashi states [1811.07238].

Open/closed duality on the annulus, Möbius strip, and Klein bottle constrains these coefficients. The modular transformation relevant for the Möbius strip is \[ \hat\chi_j(-1/4\tau) = \sum_i \chi_i(\tau)\, P_{ij}, \qquad P = \sqrt{T}\, S\, T^2\, S\, \sqrt{T}. \] The corresponding duality constraints include \[ k^i = \sum_j \Gamma_{uj}\Gamma_{vj} S_{ij}, \qquad m_{au}^i = \sum_j B_{aj}\Gamma_{uj} P_{ij}. \] This is the basic reason crosscap coefficients depend on the modular \(P\)-matrix rather than only on \(S\) [1811.07238].

For charge-conjugation modular invariant theories with trivial internal symmetry, the standard Pradisi-Sagnotti-Stanev solution is \[ \Gamma_{0i} = \frac{P_{0i}}{\sqrt{S_{0i}}}. \] More generally, for a simple current \(J\) of order \(N\), \[ \Gamma_{ni} = \frac{P_{J^n i}}{\sqrt{S_{0i}}}, \qquad n=0,\ldots,N-1. \] Thus the dependence of crosscap coefficients on the parity choice is explicit [1811.07238].

The vacuum component of the crosscap state is \(\Gamma_{u0}\). In the Klein bottle channel one has \[ Z_K \sim \Gamma_{u0}\Gamma_{v0}\, e^{\frac{\pi L c}{24\beta}}, \] so the ground-state degeneracy associated with a given crosscap is \(g_u=\Gamma_{u0}\), and the crosscap entropy is \[ S_C=\ln g_u=\ln \Gamma_{u0}. \] Using the PSS formula, \[ S_C=\ln\!\left(\frac{P_{u0}}{\sqrt{S_{00}}}\right). \] This is the direct non-oriented analogue of the Affleck-Ludwig boundary entropy [1811.07238].

## 2. Crosscap one-point data in \(\mathbb{RP}^d\) CFT

In CFT on real projective space, the crosscap involution is implemented after stereographic projection by \[ x^\mu \sim -\frac{x^\mu}{x^2}. \] The quotient preserves an \(SO(d,1)\) subgroup of the full conformal group [1703.08159].

Because of the nontrivial background, scalar one-point functions need not vanish. For a scalar primary \(O_i\) of dimension \(\Delta_i\), \[ \langle O_i(x)\rangle = \frac{a_i}{(1+x^2)^{\Delta_i}}. \] The coefficients \(a_i\) parametrize a set of crosscap CFT data and cannot be probed using flat-space correlation functions [1703.08159].

For scalar two-point functions, conformal symmetry allows the cross-ratio \[ \eta = \frac{(x-y)^2}{(1+x^2)(1+y^2)} \in [0,1], \] and the reduced correlator has the decomposition \[ G(\eta) = \sum_k C_{12k}\,a_k\, G^{\mathrm{proj}}_{\Delta_k}(\eta). \] Thus the quantities entering the \(\mathbb{RP}^d\) scalar two-point bootstrap are the products \(C_{12k}a_k\), where \(C_{12k}\) are the ordinary flat-space OPE coefficients and \(a_k\) are the crosscap one-point coefficients [1703.08159].

The crosscap bootstrap equation arises from the involution \(\eta\mapsto 1-\eta\): \[ G(\eta) = \left(\frac{\eta}{1-\eta}\right)^{\Delta_+} G(1-\eta), \] or with parity sign \(\epsilon=\pm1\), \[ G(\eta)=\epsilon\left(\frac{\eta}{1-\eta}\right)^{\Delta_+}G(1-\eta). \] Accordingly, the bootstrap constrains the spectrum \(\{\Delta_k\}\) and the coefficient combinations \(\{C_{12k}a_k\}\), but not \(a_k\) and \(C_{12k}\) separately [1703.08159].

The same paper reformulates this in alpha space. If \(\widehat G(\alpha)\) is the spectral density, then pole positions encode operator dimensions and residues encode the products \(C_{12k}a_k\). Concretely, if \(\Delta_k=h+2\alpha_k\), then \[ C_{12k} a_k = -\frac{2}{Q_{\mathrm{proj}}(-\alpha_k)} \operatorname*{Res}_{\alpha=\alpha_k}\widehat G(\alpha). \] This gives a direct spectral encoding of crosscap data [1703.08159].

## 3. XCFT on \(\mathbb{RP}^2\) and \(\mathbb{K}^2\)

In the holographic XCFT literature, the central crosscap datum is the state \(|C\rangle\) satisfying the Virasoro gluing condition \[ \left(L_n-(-1)^n \bar L_{-n}\right)|C\rangle=0, \qquad n\in\mathbb Z. \] The \(\mathbb{RP}^2\) partition function is the vacuum overlap \[ Z_{\mathbb{RP}^2}=\langle 0|C\rangle, \] also called the \(p\)-function [2405.03755].

This work does not explicitly write \[ |C\rangle=\sum_i \Gamma_i\, |C_i\rangle\!\rangle, \] nor does it derive Ishibashi crosscap coefficients in the traditional RCFT sense. Instead, the objects playing the operational role of crosscap coefficients are the vacuum overlap \(p=\langle 0|C\rangle\) and the one-point coefficients \(a_{\mathcal O}\) in \(\mathbb{RP}^2\) correlators [2405.03755].

For a scalar primary \(\mathcal O\) of scaling dimension \(\Delta\) on \(\mathbb{RP}^2\) of radius \(r\), the one-point function is \[ \langle \mathcal O(\theta,\phi)\rangle_{\mathbb{RP}^2} = \frac{a_{\mathcal O}}{(2r)^\Delta}. \] The coefficient \(a_{\mathcal O}\) is precisely the dynamical coefficient attached to the operator \(\mathcal O\) in the presence of a crosscap, and in standard XCFT terminology it is the crosscap one-point coefficient [2405.03755].

In the holographic dual, \[ p = \exp\!\left(-\frac{\eta_*}{4G_N}\right) = \exp\!\left(-\frac{c}{6}\operatorname{arccoth}(-T)\right), \] with \(\eta_*=\operatorname{arccoth}(-T)\) and \(c=\frac{3}{2G_N}\). For heavy-but-probe scalar primaries, the one-point coefficient scales as \[ a_{\mathcal O}=\lambda\, e^{\eta_*\Delta}. \] This gives a semiclassical formula for the vacuum crosscap overlap and a semiclassical scaling law for nontrivial one-point coefficients, while still stopping short of a full RCFT-type \(\Gamma_i\) decomposition [2405.03755].

A complementary recent study of lattice Ising crosscap overlap on the self-dual critical line likewise does not compute RCFT crosscap coefficients \(\Gamma_i\). It studies the lattice crosscap overlap \(\langle \psi_0|\mathcal C_{\mathrm{latt}}\rangle\), shows that \[ \lim_{N\to\infty}\langle \psi_0 |\mathcal{C}_{\mathrm{latt}}\rangle = \sqrt{\frac{2+\sqrt2}{2}}, \] and interprets this as a finite-size lattice realization of the continuum vacuum crosscap overlap rather than a full Ishibashi decomposition [2601.21502].

## 4. Higher-dimensional generalization: crosscap defects

A recent extension replaces the ordinary real-projective-space quotient by a \(\mathbb Z_2\) quotient with a \(p\)-dimensional fixed locus, producing a crosscap defect. In embedding space the quotient is \[ X^M\sim\iota_p(X^M), \qquad \iota_p(X^M)=\begin{cases} X^M,&M\in \{0,1,\dots,p,d+1\},\\ -X^M,&M\in \{p+1,\dots,d\}. \end{cases} \] The preserved bosonic symmetry is \(O^+(p+1,1)\times \frac{O(q)\times K}{\iota_p}\), with \(q=d-p\) [2604.19868].

The direct analogue of the ordinary crosscap coefficient is the bulk one-point coefficient \(A_i\). For a scalar primary one has \[ \langle\mathcal{O}_i(X)\rangle=\frac{A_i}{(X_-\cdot X_-)^{\Delta_i/2}} \qquad\Rightarrow\qquad \langle\mathcal{O}_i(x)\rangle_{\text{flat}}=\frac{A_i}{|x_\perp|^{\Delta_i}}. \] These \(A_i\) reduce to the familiar \(\mathbb{RP}^d\) crosscap coefficients when \(p=-1\) [2604.19868].

This generalized setup contains additional data not present in ordinary XCFT. The paper identifies the XDCFT data as \(A_i\), \(B_{i\hat k}\), \(\hat C_{\hat i\hat j\hat k}\), together with bulk CFT data. Here \(B_{i\hat k}\) are bulk-to-defect two-point/OPE coefficients, not crosscap one-point coefficients. In scalar bulk two-point functions, the defect-channel identity contribution is \(A_iA_j\), while the bulk and image channels involve the combinations \(C_{ijk}A_k\) [2604.19868].

Selection rules are sharper than in standard \(\mathbb{RP}^d\) CFT. For scalar bulk one-point functions, only \(\iota_p\)-even scalars can generically have nonzero \(A_i\), with a \(q=1\) exception. For spinning bulk one-point functions, only parity-even, \(\iota_p\)-even, even-spin symmetric traceless tensors can have nonzero \(A_i\) [2604.19868]. This suggests that the higher-codimension generalization preserves the conceptual role of the crosscap coefficient while embedding it into a richer defect-data system.

## 5. Integrable QFT and spin-chain amplitudes

In massive integrable field theories and integrable spin chains, crosscap coefficients are naturally realized as overlaps between a crosscap state and finite-volume energy eigenstates. The paper on integrable crosscap states defines the key amplitudes as \[ \langle \Psi_L|\mathcal C\rangle \qquad\text{or}\qquad \langle \mathcal C|\Psi_L\rangle, \] with vacuum overlap \[ p\equiv \langle \mathcal C|\Omega_L\rangle \] and crosscap entropy \[ s_{\mathcal C}=\log|p|. \] These are the massive analogues of CFT crosscap amplitudes [2111.09901].

A selection rule holds: only parity-symmetric states contribute. For diagonal scattering, parity acts by reversing all momenta, and only states with rapidity sets invariant under sign flip have nonzero overlap. The exact finite-volume vacuum formula is \[ |p|=\left|\langle \mathcal{C}|\Omega_L\rangle\right| =\sqrt{\left(1+\sqrt{\frac{Y(0)}{1+Y(0)}}\right)\frac{\det \left[1-\hat{G}_{-}\right]}{\det \left[1-\hat{G}_{+}\right]}}. \] For parity-symmetric excited states, \[ |\langle \mathcal{C}|\Psi_{L}\rangle| =\sqrt{\left(1+\sqrt{\frac{Y(0)}{1+Y(0)}}\right) \frac{\det \left[1-\hat{G}_{-}^{\bullet}\right]}{\det \left[1-\hat{G}_{+}^{\bullet}\right]}}. \] In the asymptotic large-volume limit, this simplifies to \[ |\langle \mathcal{C}|\Psi_{L}\rangle|\overset{L\to\infty}{=}\sqrt{\frac{\det G_{+}}{\det G_{-}}}. \] These determinant-ratio formulas are the main exact crosscap coefficients in the integrable setting [2111.09901].

A later paper extends this structure to the Lieb-Liniger model. There, the normalized overlap with a parity-paired Bethe state is again \[ \frac{\langle\mathcal C|\boldsymbol\lambda_N\rangle}{\sqrt{\langle\boldsymbol\lambda_N|\boldsymbol\lambda_N\rangle}} = \sqrt{\frac{\det G_{N/2}^+}{\det G_{N/2}^-}}. \] The same formula also holds for compact and non-compact spin chains, with the paper emphasizing that the prefactor is trivial for the crosscap state [2305.16046]. This suggests a robust integrable notion of crosscap coefficient as an exact finite-volume overlap amplitude.

## 6. Microscopic extraction in \((2+1)\)D CFT

A distinct recent direction concerns \((2+1)\)D CFT on \(\mathbb{RP}^3\). There the crosscap coefficients are defined as amplitudes of scalar one-point functions \[ \langle\mathcal{O}_{i}(\vec{r})\rangle_{\mathbb{RP}^{3}} = \frac{\mathcal{A}_{i}}{(1+|\vec{r}|^{2})^{\Delta_{i}}}, \] and equivalently as overlaps \[ \mathcal{A}_{i} = \langle \mathcal{C}| \mathcal{O}_{i} \rangle. \] The paper explicitly says that \(\mathcal{A}_i\) is the crosscap coefficient [2507.20005].

The main novelty is numerical extraction from microscopic models. The authors construct lattice and continuum crosscap states by entangling antipodal degrees of freedom in Bell-type states. On the icosahedron, the lattice crosscap state is \[ |\mathcal{C}_{\rm IC}\rangle = \prod_{i\in {\rm half}} \left( |\uparrow\rangle_i|\uparrow\rangle_{\bar i} + |\downarrow\rangle_i|\downarrow\rangle_{\bar i} \right). \] In a spherical Landau-level realization, the conjectured continuum crosscap state is \[ |\mathcal C_{\rm LL}\rangle = \exp\Bigg\{ \int_{\rm upper} d\boldsymbol\Omega\, \Big[ \Psi^\dagger_\uparrow(\theta,\phi)\Psi^\dagger_\uparrow(\pi-\theta,\pi+\phi) + \Psi^\dagger_\downarrow(\theta,\phi)\Psi^\dagger_\downarrow(\pi-\theta,\pi+\phi) \Big] \Bigg\} |\emptyset\rangle. \] These constructions allow direct many-body overlaps with low-energy eigenstates [2507.20005].

For the \((2+1)\)D Ising CFT, the extrapolated spherical results are \(\mathcal A_{\mathbb I}=1.129\), \(\mathcal A_\epsilon/\mathcal A_{\mathbb I}=0.674\), and \(\mathcal A_{\epsilon'}/\mathcal A_{\mathbb I}=0.898\), in good agreement with the bootstrap ratios \(0.667\) and \(0.896\) [2507.20005]. An important distinction drawn in this work is that microscopic overlaps provide absolute amplitudes such as \(\mathcal A_{\mathbb I}\), whereas the cited bootstrap results provide only ratios. A plausible implication is that non-orientable CFT data in dimensions above two may be experimentally or numerically accessible in a more direct way than the standard RCFT formalism suggests.

## 6. Conceptual synthesis and scope

Across these literatures, “crosscap coefficients” does not denote a single universal formalism. In RCFT, the term most naturally refers to the coefficients \(\Gamma_{ui}\) in the expansion of a crosscap state into crosscap Ishibashi states [1811.07238]. In CFT on \(\mathbb{RP}^d\), the directly measurable quantities are the one-point coefficients \(a_i\), which enter scalar two-point functions through \(C_{12k}a_k\) [1703.08159]. In XCFT on \(\mathbb{RP}^2\) and \(\mathbb{K}^2\), the practically relevant data are the vacuum overlap \(p=\langle0|C\rangle\) and the one-point coefficients \(a_{\mathcal O}\), because the Ishibashi decomposition is not explicitly written [2405.03755]. In higher-dimensional crosscap-defect CFT, the analogous data are the bulk one-point coefficients \(A_i\), now supplemented by bulk-to-defect coefficients \(B_{i\hat k}\) and defect OPE data [2604.19868]. In integrable QFT and spin chains, the massive continuation of crosscap data is the family of overlaps \(\langle \mathcal C|\Psi_L\rangle\), governed by determinant ratios [2111.09901], [2305.16046]. In \((2+1)\)D microscopic models, crosscap coefficients are directly extracted as overlaps with Bell-paired antipodal crosscap states [2507.20005].

A common structural pattern is nevertheless visible. Crosscap coefficients always encode how non-orientable geometry couples to operator content: either through one-point functions, through vacuum overlaps and crosscap entropies, or through overlap amplitudes with parity-constrained eigenstates. This suggests that the essential meaning of a crosscap coefficient is not tied to a particular basis choice, but to the response of the theory to the insertion of a crosscap background or crosscap state.

Source: https://www.emergentmind.com/topics/crosscap-coefficients