From Baby Universes to Narain Moduli: Solving Holographic Non-Factorization via SymTFT

    ## TL;DR
    Why does gravity look like an average of many theories? This paper provides a sophisticated answer: **Ensemble averaging in holography is actually a sum over topological completions of a single "relative" theory.** By using the Symmetry TFT (SymTFT) framework, the author shows that the weird non-factorization of wormholes is just the result of not knowing which topological boundary condition to pick.

    ## The Core Intuition: Relative vs. Absolute
    In modern physics, we've learned that many QFTs are "relative"—they aren't fully defined unless they are attached to a (d+1)-dimensional bulk (the SymTFT). 
    - The **Physical Boundary** handles the "what": the Hamiltonian, the local operators, and the dynamics.
    - The **Topological Boundary** (the Cap) handles the "how": the global form of the symmetry group, the charge lattice, and the specific Hilbert space.

    The author’s key insight is that **Euclidean wormholes** in gravity are effectively performing a sum over all possible topological caps.

    ## Methodology: The SymTFT Slab
    Imagine a sandwich (a slab). On the left, you have your physical world. On the right, you have a topological "mirror." If you average over what happens on the right side while keeping the left side fixed, you get an ensemble.

    ### 1. The Marolf-Maxfield Model (Discrete Case)
    In this 2D topological model, the "caps" are labeled by finite sets $S$. The author shows that a groupoid sum over these sets—weighting them by $1/|Aut(S)|$—perfectly reproduces the Poisson distribution and Bell-polynomial moments found in baby universe models.
    
    ### 2. Narain Moduli (Continuous Case)
    For $c$ compact bosons, the SymTFT is an $\mathbb{R}$-valued BF theory. Here, the "caps" are **maximal isotropic subgroups** (Lagrangian sublattices). The author proves that the space of these caps is exactly the **Narain Moduli Space**:
    $$ \mathcal{L}_{\mathrm{Narain}}^{(c)} = O(c, c; \mathbb{Z}) \backslash O(c, c; \mathbb{R}) / (O(c) 	imes O(c)) $$
    By integrating over this space using the natural Haar measure, we recover the standard Narain average used in 3D gravity.

    ![The SymTFT Slab Setup](Image_Placeholder)
    *(Figure 1: Conceptual diagram of the SymTFT slab with Physical Boundary $B$ and Topological Boundary $L$)*

    ## Experimental Evidence: The Siegel-Weil Formula
    The most striking result comes from the Narain case. When we average the partition function over the space of topological boundary conditions, the SymTFT naturally leads us to the **Siegel-Weil formula**. For $c > 2$, the result is:
    $$ \langle Z(	au) \rangle = \frac{E_{c / 2}(	au)}{	au_{2}^{c / 2} | \eta (	au) | ^{2 c}} $$
    This isn't just a coincidence; it's a rigorous manifestation of the fact that the bulk gravity theory doesn't see a specific lattice, but rather the average of all "bosonic" completions.

    ![Narain Average Results](Image_Placeholder)
    *(Figure 2: Comparison of the averaged partition function vs. individual theory components)*

    ## Academic Perspective: Why This Matters
    This work bridges two massive islands: **Generalized Symmetries** and **Ensemble Holography**. 
    
    Historically, people argued whether gravity is "one theory" (factorization) or "an average of many theories" (non-factorization). This paper suggests a middle ground: **Gravity is one relative theory, but we are forced to average over its possible absolute completions.**

    ## Limitations & Future Work
    - **JT Gravity**: The author proposes a link to the Schwarzian mode but notes the classification of SL(2,R) boundary conditions is still "sloppy."
    - **3D Gravity**: The "Virasoro TQFT" is the next frontier. It would involve a continuous average over Virasoro primary spectra and OPE data.

    ## Conclusion
    By reframing ensemble averaging as "Topological Boundary Averaging," we move away from the idea that gravity is "messy" or "random." Instead, the randomness is restricted to the topological sector, while the local dynamics remain perfectly fixed and well-behaved.