Rastrigin function

Rastrigin function of two variables

In 3D

Contour

In mathematical optimization, the Rastrigin function is a non-convex function used as a performance test problem for optimization algorithms. It is a typical example of non-linear multimodal function. It was first proposed in 1974 by Rastrigin ^[1] as a 2-dimensional function and has been generalized by Rudolph. ^[2] The generalized version was popularized by Hoffmeister & Bäck ^[3] and Mühlenbein et al. ^[4] Finding the minimum of this function is a fairly difficult problem due to its large search space and its large number of local minima.

On an ${\displaystyle n}$-dimensional domain it is defined by:

${\displaystyle f(\mathbf {x} )=An+\sum _{i=1}^{n}\left[x_{i}^{2}-A\cos(2\pi x_{i})\right]}$

where ${\displaystyle A=10}$ and ${\displaystyle x_{i}\in [-5.12,5.12]}$. There are many extrema:

• The global minimum is at ${\displaystyle \mathbf {x} =\mathbf {0} }$ where ${\displaystyle f(\mathbf {x} )=0}$.
• The maximum function value for ${\displaystyle x_{i}\in [-5.12,5.12]}$ is located around ${\displaystyle x_{i}\in [\pm 4.52299366...,...,\pm 4.52299366...]}$:

Number of dimensions	Maximum value at ${\displaystyle \pm 4.52299366}$
1	40.35329019
2	80.70658039
3	121.0598706
4	161.4131608
5	201.7664509
6	242.1197412
7	282.4730314
8	322.8263216
9	363.1796117

Here are all the values at 0.5 interval listed for the 2D Rastrigin function with ${\displaystyle x_{i}\in [-5.12,5.12]}$:

${\displaystyle f(x)}$		${\displaystyle x1}$
		${\displaystyle 0}$	${\displaystyle \pm 0.5}$	${\displaystyle \pm 1}$	${\displaystyle \pm 1.5}$	${\displaystyle \pm 2}$	${\displaystyle \pm 2.5}$	${\displaystyle \pm 3}$	${\displaystyle \pm 3.5}$	${\displaystyle \pm 4}$	${\displaystyle \pm 4.5}$	${\displaystyle \pm 5}$	${\displaystyle \pm 5.12}$
${\displaystyle x2}$	${\displaystyle 0}$	0	20.25	1	22.25	4	26.25	9	32.25	16	40.25	25	28.92
	${\displaystyle \pm 0.5}$	20.25	40.5	21.25	42.5	24.25	46.5	29.25	52.5	36.25	60.5	45.25	49.17
	${\displaystyle \pm 1}$	1	21.25	2	23.25	5	27.25	10	33.25	17	41.25	26	29.92
	${\displaystyle \pm 1.5}$	22.25	42.5	23.25	44.5	26.25	48.5	31.25	54.5	38.25	62.5	47.25	51.17
	${\displaystyle \pm 2}$	4	24.25	5	26.25	8	30.25	13	36.25	20	44.25	29	32.92
	${\displaystyle \pm 2.5}$	26.25	46.5	27.25	48.5	30.25	52.5	35.25	58.5	42.25	66.5	51.25	55.17
	${\displaystyle \pm 3}$	9	29.25	10	31.25	13	35.25	18	41.25	25	49.25	34	37.92
	${\displaystyle \pm 3.5}$	32.25	52.5	33.25	54.5	36.25	58.5	41.25	64.5	48.25	72.5	57.25	61.17
	${\displaystyle \pm 4}$	16	36.25	17	38.25	20	42.25	25	48.25	32	56.25	41	44.92
	${\displaystyle \pm 4.5}$	40.25	60.5	41.25	62.5	44.25	66.5	49.25	72.5	56.25	80.5	65.25	69.17
	${\displaystyle \pm 5}$	25	45.25	26	47.25	29	51.25	34	57.25	41	65.25	50	53.92
	${\displaystyle \pm 5.12}$	28.92	49.17	29.92	51.17	32.92	55.17	37.92	61.17	44.92	69.17	53.92	57.85

The abundance of local minima underlines the necessity of a global optimization algorithm when needing to find the global minimum. Local optimization algorithms are likely to get stuck in a local minimum.

See also

• Test functions for optimization

Notes

1. Rastrigin, L. A. "Systems of extremal control." Mir, Moscow (1974).
2. G. Rudolph. "Globale Optimierung mit parallelen Evolutionsstrategien". Diplomarbeit. Department of Computer Science, University of Dortmund, July 1990.
3. F. Hoffmeister and T. Bäck. "Genetic Algorithms and Evolution Strategies: Similarities and Differences", pages 455–469 in: H.-P. Schwefel and R. Männer (eds.): Parallel Problem Solving from Nature, PPSN I, Proceedings, Springer, 1991.
4. H. Mühlenbein, D. Schomisch and J. Born. "The Parallel Genetic Algorithm as Function Optimizer ". Parallel Computing, 17, pages 619–632, 1991.