---
title: 'DR-Submodularity: Theory and Algorithms'
url: https://www.emergentmind.com/topics/dr-submodularity
type: topic
---

# DR-Submodularity: Theory and Algorithms

DR-submodularity is a generalization of the classical diminishing returns property of discrete submodular set functions to more general domains, such as the integer lattice, distributive lattices, and continuous domains. A function is DR-submodular if its marginal gain from increasing a coordinate (or “adding” an element) decreases as the current input increases, formalizing the concept that early investments or selections provide larger incremental benefits than later ones. DR-submodularity underpins the structure of many practical optimization problems in machine learning, economics, network theory, and combinatorial optimization, enabling the design of polynomial-time algorithms with provable approximation guarantees in otherwise intractable non-convex or combinatorial settings.

## 1. Formal Definitions and Fundamental Properties

Traditional submodularity for set functions $f: 2^N \to \mathbb{R}$ is characterized by:
\[
f(S \cup \{e\}) - f(S) \geq f(T \cup \{e\}) - f(T)
\]
for all $S \subseteq T \subseteq N$ and $e \notin T$. This is the diminishing returns (DR) property: marginal gains decrease as the set grows.

**DR-submodularity** extends this to multivariate and continuous domains. For functions $f: \mathcal{D} \to \mathbb{R}$ where $\mathcal{D}$ is a product domain (often the integer lattice $\{0, \ldots, C\}^n$, a distributive lattice, or $[0,1]^n$), the DR property requires that for all $x \leq y$ (coordinate-wise) and every feasible coordinate $i$ and increment $k \geq 0$:
\[
f(x + k\chi_i) - f(x) \ge f(y + k\chi_i) - f(y)
\]
where $\chi_i$ is the $i$th standard basis vector.

Key properties:
- For set functions, submodularity and the DR property are equivalent.
- For integer or continuous domains, submodularity (via the lattice inequality $f(x) + f(y) \ge f(x \wedge y) + f(x \vee y)$) does **not** imply DR-submodularity; additional coordinate-wise concavity is often required [1505.05423], [2006.13474].
- In the continuous setting, if $f$ is twice differentiable, DR-submodularity is equivalent to all cross-partial derivatives off the diagonal being non-positive and diagonal terms non-positive (i.e., $f$ is coordinate-wise concave) [2006.13474].

## 2. Algorithmic Frameworks for DR-Submodular Maximization

Efficient algorithms for DR-submodular maximization typically build on generalizations of greedy or double greedy paradigms and exploit concavity along nonnegative directions.

- **Unconstrained Maximization**:  
  - For monotone DR-submodular functions over distributive lattices, a greedy approach yields a $1/2$-approximation [1505.05423].
  - For set functions and integer lattices, double greedy frameworks achieve a $1/2$-approximation for DR-submodular objectives (randomized), and $1/3$ for general submodular (non-DR) functions [1505.05423], [1612.00960].
  - For continuous domains with box constraints, the DR-DoubleGreedy algorithm achieves a tight $1/2$-approximation in linear time [1805.07482].

- **Constrained Maximization**:
  - For matroid or poset matroid constraints and monotone DR-submodular functions, greedy selection (augmenting with feasible elements maximizing marginal gain) attains $1/2$-approximation; for cardinality (uniform matroid), it reaches $(1 - 1/e)$ [1505.05423].
  - Continuous greedy algorithms and Frank-Wolfe–style projection-free optimization provide $(1 - 1/e)$-approximation for monotone DR-submodular functions under down-closed convex constraints [1606.05615], [1505.05423], [1711.02515].

- **Non-monotone Settings**:  
  - For non-monotone DR-submodular maximization, double greedy–type algorithms attain $1/(2+\varepsilon)$-approximation in strongly polynomial time [1612.00960], $1/4$ via a two-phase or discretized approach with convergence guarantees [1711.02515], [2203.14740].

## 3. Hardness and Complexity Landscape

While many tractable cases exist, particular constraints can render DR-submodular maximization intractable:

- **Knapsack Constraints**: Knapsack constraints in general distributive lattice settings cause a dramatic hardness increase—no constant-factor approximation is achievable unless $3$-SAT can be solved in sub-exponential time [1505.05423]. The inapproximability bound flows from a reduction to dense subhypergraph problems.
- **General Continuous Nonconvexity**: DR-submodular maximization is NP-hard in general, and the best possible polynomial-time approximation ratios under value-oracle access are $1 - 1/e$ for monotone and $1/2$ for non-monotone objectives unless RP = NP [2006.13474].
- **Recent Advances**: The best-known solver for multilinear extension maximization subject to down-closed constraints achieves a $0.401$ approximation, nearly matching the inapproximability bound $0.478$ [2311.01129].

## 4. Mathematical Formulations and Approximation Guarantees

Foundational inequalities:

| Problem/domain                   | Approximation Ratio      | Reference            |
|----------------------------------|-------------------------|----------------------|
| Unconstrained integer lattice    | $1/3$ (general submod.) | [1505.05423]         |
| Unconstrained DR-submodular      | $1/2$                   | [1505.05423], [1805.07482] |
| Monotone, card. constraint       | $1 - 1/e$               | [1505.05423], [2006.13474] |
| Monotone, poset matroid          | $1/2$                   | [1505.05423]         |
| Non-monotone continuous          | $1/3$ (FW/DoubleGreedy) | [1606.05615]         |
| Non-monotone contin. (box)       | $1/2$ (DR-DoubleGreedy) | [1805.07482]         |
| Down-closed constraint (ML ext.) | $0.401$                 | [2311.01129]         |
| Knapsack/distributive lattice    | no const. approx.       | [1505.05423]         |

Representative formulas:

- **Lattice DR-submodularity**:  
  $f(x) + f(y) \geq f(x \wedge y) + f(x \vee y)$  
  $f(x + \chi_i) - f(x) \geq f(y + \chi_i) - f(y)$ for $x \leq y$ (integer lattice).
- **Continuous DR-submodularity (gradient characterization)**:  
  $\nabla_i f(x) \geq \nabla_i f(y)$ for $x \leq y$.
- **Multilinear extension (randomization for sets)**:  
  $F(x) = \mathbb{E}_{S \sim x}[f(S)]$  
  (where $S$ is a random set, including each element independently with probability $x_i$).

- **History-aware lower bound [2311.01129]**:  
  $F\left(1-\mathbf{a}\odot e^{-\int_0^t\mathbf{x}(\tau)d\tau}\right) \geq e^{-t}\left[F(1-\mathbf{a}) + \sum_{i=1}^\infty \frac{1}{i!}\int_{[0,t]^i}F\left((1-\mathbf{a})\oplus_{j=1}^i \mathbf{x}(\tau_j)\right) d\tau\right]$

## 5. Applications in Optimization and Machine Learning

DR-submodular maximization arises in a wide variety of resource allocation, combinatorial, and statistical problems:

- **Budget allocation and marketing**, where additional budget yields diminishing growth in influence or revenue [1606.05615], [2212.06646].
- **Sensor placement**, for coverage with redundancy and reliability [1903.00950], [1907.04279].
- **Facility location**, where cost-benefit of new facilities is decreasing with existing coverage [1606.05615].
- **Mean field inference and MAP for DPPs**: maximizing expected value or log-likelihood functions with structured repulsion [1711.02515], [1805.07482], [2006.13474].
- **Profit maximization in social networks**: allocating repeated trials or investments with diminishing conversion rates [2212.06646].
- **Online and game-theoretical frameworks** where distributed agents maximize a monotone DR-submodular social function with proven bounds on equilibria inefficiency, related to price of anarchy [1903.00950], [1907.00316], [1907.00312].

## 6. Advanced Algorithmic Developments and Open Problems

Innovations in DR-submodular optimization span multiple algorithmic fronts:

- **Continuous relaxation and rounding**: Multilinear extension maximization with randomized rounding is central to approaching combinatorial constraints [2311.01129].
- **Derivative-free and noisy optimization**: Black-box methods such as LDGM yield robustness to non-differentiability and noise, matching gradient-based methods in approximation quality [1810.06833].
- **Projection-free and bandit algorithms**: Recent frameworks achieve first regret guarantees for stochastic DR-submodular maximization under bandit feedback, exploiting smoothing and momentum techniques [2305.16671], [2405.00065].
- **Strong/curved DR-submodularity**: When the objective enjoys strong concavity along nonnegative directions, fast algorithms with improved approximation and linear convergence can be realized [2111.07990].
- **Oracle complexity**: For general convex constraints, stochastic value oracle models require $O(1/\varepsilon^5)$ calls for $O(\varepsilon)$-approximation in the worst case [2305.16671].

Key open questions remain:
- Can the $0.401$ approximation for multilinear extension maximization under down-closed constraints be further improved, as the inapproximability barrier is $0.478$ [2311.01129]?
- Do adaptive or history-dependent strategies exploiting the new “history-aware” bounds enable further progress [2311.01129]?
- Under which settings (e.g., additional structure, dynamic/adversarial, composite constraints) can the known performance gaps be narrowed?

## 7. Impact and Broader Significance

DR-submodularity has fundamentally reshaped understanding of non-convex optimization in both discrete and continuous settings. The extension of the diminishing returns paradigm to richer domains has allowed for algorithmic advances in areas previously considered intractable:

- DR-submodularity underlies algorithms that efficiently bridge the gap between combinatorial and convex optimization, leveraging multilinear relaxations and randomized rounding.
- The clear separation between submodularity and DR-submodularity on lattices has elucidated sources of algorithmic hardness, indicating the necessity of the diminishing returns property for tractability especially outside the Boolean cube [1505.05423].
- Theoretical results on inapproximability, tight lower bounds, and oracle complexity expose intrinsic barriers and guide algorithm development.

Recent frameworks unify diverse settings—monotone/non-monotone, continuous/lattice, deterministic/stochastic, full-information/bandit/zero-order feedback—offering a comprehensive, modular toolbox for non-convex and non-monotone optimization with rigorous guarantees [2305.16671], [2405.00065].

This synthesis reflects the depth and diversity of DR-submodular optimization, encompassing rigorous theoretical analysis, algorithmic innovation, and practical applications across machine learning, data science, and operations research.

Source: https://www.emergentmind.com/topics/dr-submodularity