---
title: 'Dinkelbach’s Algorithm: Fractional Programming'
url: https://www.emergentmind.com/topics/dinkelbach-s-algorithm
type: topic
---

# Dinkelbach’s Algorithm: Fractional Programming

Dinkelbach’s algorithm is a foundational procedure for solving fractional programming problems—those in which the objective function is a ratio of two (often convex) functions—by reducing them to a sequence of parametric subproblems. Since its inception in 1967, the algorithm and its variants have played a major role in optimization, with applications in communications, scheduling, signal processing, and combinatorial optimization. Ongoing developments—such as interval variants, minimal-correction accelerations, and look-ahead modifications—have addressed both the theoretical convergence rate and practical applicability of the method, notably maintaining computational efficiency by solving only one subproblem per iteration [2510.26257, 2004.08634].

## 1. Fractional Programming Formulation

Fractional programming aims to optimize a ratio-valued objective:
\[
\min_{x \in \mathcal F} \frac{f_1(x)}{f_2(x)}, \qquad \text{with } f_2(x) > 0.
\]
This can be equivalently translated into the parametric root-finding problem for
\[
g(\alpha) = \max_{x \in \mathcal F} \left\{ -f_1(x) + \alpha f_2(x) \right\}, \qquad \text{with root } \alpha^* \text{ where } g(\alpha^*) = 0.
\]
Here, $g$ is strictly increasing and convex in $\alpha$. In the combinatorial context, the generic problem is
\[
\inf_{x \in \mathcal D} \frac{c^\top x}{d^\top x},
\]
leading to
\[
f(\delta) = \inf_{x \in \mathcal D} \left\{ c^\top x - \delta d^\top x \right\},
\]
where $f$ is concave, piecewise-linear, and nonincreasing in $\delta$ [2004.08634]. The fractional problem is reduced to finding the root (largest $\delta$ so that $f(\delta)=0$).

## 2. Classical Dinkelbach and Newton–Dinkelbach Methods

### Classical Update

The core Dinkelbach update is equivalent to a Newton method for root-finding:
\[
\alpha_{k+1} = \alpha_k - \frac{g(\alpha_k)}{f_2(x_k)}
\]
where $x_k \in \arg\max_{x \in \mathcal{F}}\{-f_1(x)+\alpha_k f_2(x)\}$. Given $g(\alpha_k) = -f_1(x_k) + \alpha_k f_2(x_k)$, the update simplifies to
\[
\alpha_{k+1} = \frac{f_1(x_k)}{f_2(x_k)}.
\]
The sequence $\alpha_k$ is monotonic decreasing and converges to $\alpha^*$. Local convergence analysis shows quadratic rate under $g \in C^2$ [2510.26257]:
\[
|e_{k+1}| \le C |e_k|^2, \qquad e_k = \alpha_k - \alpha^*.
\]
The Newton–Dinkelbach update in discrete/combinatorial settings adapts by using negative supergradients of $f$ at each iterate, yielding similar monotonicity and convergence [2004.08634].

## 3. Interval and Accelerated Dinkelbach Variants

### Interval Dinkelbach Method

The interval Dinkelbach method (Pardalos–Phillips ‘91) maintains a bracketing sequence $[\gamma_k,\,\alpha_k]$ with updates:
- Lower-bound (secant): 
  \[
  \gamma_{k+1} = \gamma_k - g(\gamma_k) \frac{\alpha_k - \gamma_k}{g(\alpha_k) - g(\gamma_k)}
  \]
- Upper-bound (Dinkelbach): 
  \[
  \alpha_{k+1} = \frac{f_1(x^\alpha_k)}{f_2(x^\alpha_k)}.
  \]
Under $g \in C^2$, the convergence rates are quadratic for the upper bound and superlinear for the lower [2510.26257].

### Minimal Correction: Accelerated Interval Dinkelbach

An acceleration is achieved by replacing the upper-bound update with the minimum of two tangent-zeroes at $\alpha_k$ and $\gamma_{k+1}$:
\[
\alpha_{k+1} = \min \left\{
  \frac{f_1(x^\alpha_k)}{f_2(x^\alpha_k)},\,
  \frac{f_1(x^{\gamma}_{k+1})}{f_2(x^{\gamma}_{k+1})}
\right\}
\]
where $x^{\gamma}_{k+1}$ and $x^\alpha_k$ are optimizers at the respective points. This requires no additional subproblem solves—only an $O(1)$ overhead. Under $g \in C^2$, the lower bound converges cubically, while the upper bound converges superquadratically [2510.26257].

### Look-Ahead and Non-Monotone Accelerated Dinkelbach

Recent advances combine a two-step “accelerated secant–Newton” method (with order $1+\sqrt{2}$) and a safeguard based on Dinkelbach’s tangent update:
- If an extrapolated (look-ahead) step is promising, it is accepted; if not, the classical update is used.
- When the method would drop below the solution, a tangent-min correction maintains global convergence and superquadratic acceleration [2510.26257, 2004.08634].

## 4. Convergence Analysis and Complexity

The convergence of Dinkelbach variants depends crucially on regularity conditions for the parametric function ($g$ or $f$):
- **Classical Dinkelbach:** Quadratic convergence under $C^2$ regularity near the solution. 
- **Interval Dinkelbach:** Quadratic (upper), superlinear (lower bound).
- **Accelerated Minimal Correction:** Lower bound converges cubically, upper bound converges superquadratically.
- **Non-Monotone Accelerated Dinkelbach:** Incorporates two-point acceleration and periodic behavior determined by the sign of
  \[
  \chi = 3g''(\alpha^*)^2 - 2g'(\alpha^*)g'''(\alpha^*).
  \]
  The average asymptotic order per iteration exceeds 2 in all regimes, with $\sqrt{5}$ for period 2 and $\sqrt[3]{12}\approx 2.289$ for period 3 [2510.26257].

For the accelerated Newton–Dinkelbach scheme in discrete optimization, the Bregman divergence between iterates and the optimum halves every two steps, yielding $O(\log(1/\varepsilon))$ complexity for $\varepsilon$-approximation and strongly polynomial bounds in several combinatorial settings [2004.08634].

### Application-Specific Complexity

| Problem Type                      | Iteration Bound            | Per-Iteration Cost            |
|------------------------------------|---------------------------|-------------------------------|
| Linear Fractional Combinatorial    | $O(m\log m)$              | 1 linear optimization         |
| 2-Variable-Per-Inequality Systems  | $O(mn)$                   | $O(mn)$ (general), $O(m+n\log n)$ for DMDPs |
| Parametric Submodular Minimization | $O(n^2)$                  | 1 submodular fn minimization  |

[2004.08634]

## 5. Implementation and Practical Considerations

Each Dinkelbach-type iteration solves a single parametric subproblem of maximizing $-f_1(x)+\alpha f_2(x)$ or minimizing $c^\top x - \delta d^\top x$ subject to the original constraints. For accelerated variants, additional computations (extra tangent evaluations and screening conditions) require only $O(1)$ overhead per iteration if the subproblem is computationally expensive [2510.26257].

The only algorithmic tuning typically required is a screening parameter $\rho>1$, with practical performance not unduly sensitive to its value. There are no requirements for line searches or trust-region mechanisms in accelerated frameworks.

Global convergence relies on $g$ being strictly increasing/convex and at least $C^2$; local superquadratic convergence needs $C^3$. In the combinatorial context, the accelerated method leverages combinatorial monotonicity conditions, e.g., the Goemans–Radzik and subpath-monotonicity lemmas [2004.08634].

## 6. Applications and Research Directions

Dinkelbach’s method and its accelerations are deployed in a broad spectrum of optimization problems, including:
- **Communications:** Energy-efficiency maximization in wireless networks, device-to-device (D2D), non-orthogonal multiple access (NOMA), simultaneous wireless information and power transfer (SWIPT).
- **Signal Processing:** Ratio-of-quadratics, total-least-squares.
- **Scheduling:** Cyclic scheduling.
- **Combinatorial and Discrete Optimization:** Fractional combinatorial auctions, label-correcting algorithms for systems with two variables per inequality, deterministic Markov Decision Processes [2510.26257, 2004.08634].
- **Submodular Minimization:** Parametric versions with strongly polynomial iteration bounds.

Further development focuses on pushing convergence guarantees towards lower iteration complexity, maintaining the essential property of single subproblem solves per iteration, and deriving problem-specific acceleration mechanisms.

## 7. Auxiliary Results and Structural Lemmas

The theoretical basis for the fast convergence and iteration complexity in discrete and continuous fractional programming with Dinkelbach-type methods relies on several auxiliary lemmas:
- **Monotonicity:** Decreasing sequence of iterates and increasing objective values.
- **Bregman Divergence Halving:** Each two accelerated steps halve the divergence to the optimum [2004.08634].
- **Goemans–Radzik Lemma:** Halving sequences of $\{0,1\}$-vectors have $O(m\log m)$ length.
- **Subpath Monotonicity:** Governs iteration bounds in certain network flow and label-correcting settings.
- **Growth of Ring Families:** Limits the number of discrete changes in parametric submodular minimization.

These structural insights enable the design and analysis of strongly polynomial-time algorithms for large classes of fractional and parametric optimization problems.

---

For in-depth algorithmic descriptions and proofs, see [2510.26257] for accelerated continuous Dinkelbach variants and [2004.08634] for combinatorial and submodular applications of the Newton–Dinkelbach and its accelerations.

Source: https://www.emergentmind.com/topics/dinkelbach-s-algorithm