[proofplan] This is a quoted theorem from the elliptic regularity theory for the Hodge Laplacian on a closed Riemannian manifold. We do not reproduce the full argument; instead we state the standard proof technique and cite the references where the complete proof is given. The result is a special case of the general principle that weak solutions of elliptic equations with smooth coefficients are smooth — applied to the Hodge Laplacian $\Delta = d\delta + \delta d$, whose principal symbol $|\xi|^2 \mathrm{Id}_{\Lambda^p}$ is uniformly elliptic on a compact manifold. [/proofplan] [step:State the result as a quoted theorem from elliptic PDE] Let $(M, g)$ be a compact oriented Riemannian manifold without boundary, and let $\Delta : \Omega^p(M) \to \Omega^p(M)$ be the Hodge Laplacian. Fix $\alpha \in \Omega^p(M)$. A **weak solution** of $\Delta\omega = \alpha$ is a continuous linear functional $\ell : \Omega^p(M) \to \mathbb{R}$ such that \begin{align*} \ell(\Delta\beta) = \langle\langle \alpha, \beta \rangle\rangle_g \qquad \text{for all } \beta \in \Omega^p(M), \end{align*} where $\langle\langle \cdot, \cdot \rangle\rangle_g$ denotes the global $L^2$ inner product on $\Omega^p(M)$. The theorem asserts that every such $\ell$ is represented by a smooth $p$-form: there exists $\omega \in \Omega^p(M)$ with \begin{align*} \ell(\beta) = \langle\langle \omega, \beta \rangle\rangle_g \qquad \text{for all } \beta \in \Omega^p(M). \end{align*} This is the elliptic regularity theorem for the Hodge Laplacian on a closed manifold. It is a standard black box from the elliptic PDE theory of self-adjoint elliptic operators on compact manifolds; we use it as a quoted result and do not reproduce the proof here. [/step] [step:Outline the standard proof technique] The proof proceeds in three quoted ingredients from elliptic PDE on a compact manifold. **(1) Riesz representation in $L^2$.** A continuity estimate of the form $|\ell(\beta)| \le C \|\beta\|_{L^2}$ for all $\beta \in \Omega^p(M)$ — which is built into the definition of "weak solution" used in this development — extends $\ell$ uniquely to a bounded linear functional on $L^2(\Lambda^p T^*M)$. The Riesz representation theorem on the Hilbert space $L^2(\Lambda^p T^*M)$ then produces a unique $\omega_0 \in L^2(\Lambda^p T^*M)$ with $\ell(\beta) = \langle\langle \omega_0, \beta \rangle\rangle_g$ for all smooth $\beta$. **(2) Distributional equation.** The defining identity $\ell(\Delta\beta) = \langle\langle \alpha, \beta \rangle\rangle_g$ rewrites as \begin{align*} \langle\langle \omega_0, \Delta\beta \rangle\rangle_g = \langle\langle \alpha, \beta \rangle\rangle_g \qquad \text{for all } \beta \in \Omega^p(M), \end{align*} which says $\Delta\omega_0 = \alpha$ in the distributional sense (using formal self-adjointness of $\Delta$, see [Laplacian is Formally Self-Adjoint](/theorems/2743)). **(3) Elliptic regularity bootstrap.** Since $\alpha$ is smooth and the Hodge Laplacian is a uniformly elliptic second-order operator with smooth coefficients on the compact manifold $M$, the standard interior elliptic regularity theorem on Euclidean charts — applied via a partition of unity over a finite atlas — promotes the distributional solution $\omega_0 \in L^2$ to a smooth solution $\omega \in \Omega^p(M)$. The mechanism is the iterative bootstrap: $\omega_0 \in L^2$ with $\Delta\omega_0 = \alpha \in L^2$ gives $\omega_0 \in H^2_{\mathrm{loc}}$; then $\omega_0 \in H^2$ with $\Delta\omega_0 \in H^k$ gives $\omega_0 \in H^{k+2}$; iterating and using the Sobolev embedding $H^k \hookrightarrow C^m$ for $k$ large produces $\omega_0 \in C^\infty$. The two representatives $\omega_0$ and $\omega$ agree as $L^2$-classes, so $\ell(\beta) = \langle\langle \omega, \beta \rangle\rangle_g$ holds with the smooth $\omega$. For the complete proof, see Aubin, *Some Nonlinear Problems in Riemannian Geometry*, Chapter 3 (regularity of elliptic operators on Riemannian manifolds); Warner, *Foundations of Differentiable Manifolds and Lie Groups*, Chapter 6 (Hodge theory and the regularity of $\Delta$); or Evans, *Partial Differential Equations*, Chapter 6 (interior $H^k$-regularity for second-order linear elliptic operators), the last of which treats the Euclidean version that is then transferred to $M$ via the partition-of-unity argument. [/step] [step:Discuss the role of compactness, ellipticity, and the smoothness of coefficients] Three hypotheses combine to make the regularity bootstrap work, and each plays a distinct role. **Compactness of $M$.** Compactness ensures that a finite atlas suffices to cover $M$ and that the partition of unity used to localize the elliptic estimates has finitely many terms with bounded supports. Without compactness, the constants in the elliptic estimates could blow up at infinity, and the bootstrap would not propagate from local to global regularity uniformly. Compactness also ensures that the global $L^2$ inner product is well-defined as an integral of a continuous integrand against a finite measure. **Uniform ellipticity of $\Delta$.** The principal symbol of $\Delta = d\delta + \delta d$ at $\xi \in T_x^*M$ is $\sigma_2(\Delta)(\xi) = |\xi|^2_g \, \mathrm{Id}_{\Lambda^p T_x^*M}$, which is positive-definite for $\xi \neq 0$. On the compact manifold $M$, the lower bound $|\xi|^2_g \geq c |\xi|^2_{\mathrm{Eucl}}$ in any chart (with $c > 0$ depending on the chart and the metric components, all bounded on the compact closure of the chart) gives uniform ellipticity. Without uniform ellipticity, the regularity gain would degenerate at points where the symbol is degenerate, and the bootstrap could fail. **Smoothness of coefficients.** The metric $g$ is smooth on $M$, so the Christoffel symbols $\Gamma_{ij}^k$, the metric components $g_{ij}$, and their derivatives all appear in the local-coordinate expression of $\Delta$ and are bounded on the support of each chart. Smoothness ensures that the bootstrap can iterate indefinitely: each application increases the regularity of $\omega_0$ by two derivatives, and the right-hand side $\alpha$ retains its full regularity at every step. **Why the result fails without these hypotheses.** Without compactness, harmonic forms on a non-compact $(M,g)$ may fail to be smooth at infinity even if they satisfy the equation weakly — the bootstrap propagates regularity locally but loses control of constants over the manifold. Without uniform ellipticity, regularity can degenerate (the operator's failure to be invertible at degenerate points blocks the iteration). Without smoothness of $g$, the regularity of $\omega_0$ is bounded by the regularity of $g$ via the Schauder estimates, and one cannot reach $C^\infty$. [/step]