Modular Density of Smooth Maps

• Modular density is the condition where every energy-class (Sobolev or Musielak–Orlicz) map can be approximated in the modular norm by smooth maps, ensuring no Lavrentiev gap.
• The topic details analytic criteria—like integral growth conditions—and topological requirements such as k-connectedness of the target manifold for successful approximation.
• Failure of modular density, known as the Lavrentiev phenomenon, arises when either the required analytic growth or the necessary manifold connectivity is lacking, leading to nonattainable minimum energies in the smooth class.

Modular density of smooth maps concerns the approximation of Sobolev and related energy-class maps between manifolds by smooth maps, using a topology induced by appropriate modular functionals. The question is fundamental in geometric analysis, calculus of variations, and the theory of maps with controlled singularities. Modular density encapsulates the circumstances under which every energy-class map can be approximated in the modular topology (or norm) by smooth manifold-valued maps. The presence or failure of modular density (the so-called Lavrentiev phenomenon) is determined by a blend of analytic, geometric, and topological obstructions—chiefly, the growth properties of the modular functional and the connectivity of the target manifold.

1. Modular Functionals and Density in Energy Spaces

The modular approach generalizes the standard $L^p$ or Sobolev norms by considering integrals of the form

$$\rho_M(u)= \int_M\varphi(x, |Du(x)|) \, d\mathcal{H}^m(x),$$

where $\varphi(x,t)$ is a generalized Young function that can encode non-standard, double-phase, or Musielak–Orlicz growth (Antonini et al., 20 Dec 2025). The associated Musielak–Orlicz–Sobolev spaces $W^{1,\varphi}(M,N)$ consist of maps in the Sobolev class (locally in $W^{1,1}$) whose weak derivatives have finite modular. Modular convergence is defined by convergence in measure with vanishing modular distance.

Strong (modular) density of smooth maps means that for every $u \in W^{1,\varphi}(M,N)$, there exists a sequence $u_k \in C^\infty(M,N)$ such that

$$\lim_{k\to \infty} \int_M \varphi(x, |Du_k - Du|) \, dx = 0 \quad \text{and} \quad \lim_{k\to\infty} \int_M \varphi(x, |u_k-u|) \, dx = 0,$$

analogous to density in the norm topology for standard Sobolev spaces.

2. Analytical and Topological Criteria for Modular Density

Modular density hinges on both the analytic growth of $\varphi$ and the topology of the target manifold $N$ (Antonini et al., 20 Dec 2025). There are two distinct regimes:

• Analytic regime: If $\varphi$ satisfies specific integral growth conditions at infinity (see below), smooth maps are modularly dense in $W^{1,\varphi}(M,N)$ for arbitrary compact manifold $N$. Typical conditions, for $m=2$,

$$\int^\infty \frac{\varphi_M^-(t)}{t^3}\,dt = \infty,$$

or for $m \geq 3$,

$$\int^\infty \left(\frac{t}{\varphi_M^{-}(t)}\right)^{2/m-2} \left(\int_t^\infty\left(\frac{s}{\varphi_M^{-}(s)}\right)^{1/m-2} ds\right)^{-m} dt = \infty,$$

where $\varphi_M^{-}(t) = \inf_{x\in M} \varphi(x, t)$.

• Topological regime: If the analytic condition fails, modular density holds provided $N$ is $k$-connected with $1 < \gamma \leq k+1$, where $\gamma$ is the (upper) modular growth exponent (Antonini et al., 20 Dec 2025). In the classical case $\varphi(x, t) = t^p$, $N$ must be $[\![p]\!]$-connected for $p < m$.

Table: Summary of Main Criteria | Analytic Condition on $\varphi$ | Topological Condition on $N$ | Modular Density Holds? | |-----------------------------------------|------------------------------------|-------------------------| | Integral growth (see above) | None (any compact $N$) | Yes | | Fails | $N$ $k$-connected, $1 < \gamma \leq k+1$ | Yes (strong if $\gamma < k+1$) | | Fails | $N$ insufficiently connected | No (Lavrentiev gap) |

3. Lavrentiev Gap and Failure of Modular Density

The Lavrentiev gap expresses situations where the infimum of the modular energy over smooth maps is strictly greater than over all energy-class maps, corresponding to the failure of modular density (Antonini et al., 20 Dec 2025). Such gaps arise in Musielak–Orlicz classes when the spatial regularity condition (local continuity in $x$) on $\varphi$ fails, or when the analytic and topological thresholds above are not met.

In the Sobolev case, the obstruction to density for $W^{1,p}(Q^m;N)$ with $p < m$ is the nontriviality of the homotopy groups $\pi_\ell(N)$ for $\ell \leq [p]$. In general double-phase functionals $\varphi(x, t) = t^p + a(x)t^q$, the failure can occur when $q > p + \alpha \max\{1, p/m - 1\}$ for some Hölder $a(\cdot)$, leading to minimizers which cannot be obtained as modular limits of smooth maps.

4. Methods of Approximation and Modular Closures

Approximation of Sobolev or Musielak–Orlicz-class maps by smooth ones is achieved via a combination of analytic smoothing (convolution, mollification), geometric retraction into $N$, and topological surgery to handle singularities (Detaille, 2023, Bousquet et al., 2012, Antonini et al., 20 Dec 2025).

In the analytic regime, partition of unity and mollification yields smooth approximating sequences. Where topology plays a role, explicit constructions of retractions or Lipschitz "collapse" maps $\eta_\varepsilon: \mathbb{R}^N \to N$ fixing the $k$-skeleton of a triangulation of $N$ are critical. The energy cost in the modular depends on the modular growth exponent and the connectivity of $N$.

When topological obstructions exist, even optimal analytic techniques cannot bridge the Lavrentiev gap.

5. Relation to Classical Sobolev Density and Homotopy Theory

Modular density generalizes the classical density results for smooth maps in Sobolev spaces, formulated as: for $W^{s,p}(Q^m;N)$, $C^\infty$-maps are strongly dense if and only if $\pi_{[\![sp]\!]}(N)=0$ for $sp < m$ (Detaille, 2023, Bousquet et al., 2015). In Musielak–Orlicz spaces, the "critical" topological obstruction is governed by the effective upper growth exponent $\gamma$ and the $k$-connectedness of $N$ (Antonini et al., 20 Dec 2025).

The modular framework thus unifies analytic and topological density problems, reducing to the Sobolev setting when $\varphi(x,t)=t^p$.

6. Consequences and Open Problems

When modular density holds, there is no Lavrentiev gap: $$\inf_{u \in W^{1, \varphi}_{u_0}(M, N)} \int_M \varphi(x, |Du|) = \inf_{u \in C^\infty_{u_0}(M, N)} \int_M \varphi(x, |Du|).$$ This is fundamental in the calculus of variations, ensuring that minimizers among smooth maps are not obstructed by topology or geometry.

If density fails, minimizers may be nonsmooth, and the true minimum energy may be unattainable within the smooth class—a central manifestation of the Lavrentiev phenomenon.

Ongoing open questions include extending modular density results from cubes to general domains (e.g., manifolds with boundary), attaining density for classes with more general, possibly discontinuous, modulars, and refining the relation between analytic growth, topological invariants, and the dimension of admissible singular sets (Detaille, 2024, Antonini et al., 20 Dec 2025).

7. Significance and Connections

Modular density is central to understanding the structure of admissible singularities and regularity theory for manifold-valued variational problems. It bridges geometric analysis, the calculus of variations, and topological methods. The modern modular approach, including double-phase and Musielak–Orlicz settings, subsumes classical Sobolev density and provides a unified framework for a broad spectrum of functional and geometric-analytic variational problems (Antonini et al., 20 Dec 2025, Detaille, 2023, Bousquet et al., 2012).