In mathematics, a Golomb ruler is a set of marks at integer positions along a ruler such that no two pairs of marks are the same distance apart. The number of marks on the ruler is its order, and the largest distance between two of its marks is its length. Translation and reflection of a Golomb ruler are considered trivial, so the smallest mark is customarily put at 0 and the next mark at the smaller of its two possible values. Golomb rulers can be viewed as a one-dimensional special case of Costas arrays.
The Golomb ruler was named for Solomon W. Golomb and discovered independently by [1] and . Sophie Piccard also published early research on these sets, in 1939, stating as a theorem the claim that two Golomb rulers with the same distance set must be congruent. This turned out to be false for six-point rulers, but true otherwise.[2]
There is no requirement that a Golomb ruler be able to measure all distances up to its length, but if it does, it is called a perfect Golomb ruler. It has been proved that no perfect Golomb ruler exists for five or more marks. A Golomb ruler is optimal if no shorter Golomb ruler of the same order exists. Creating Golomb rulers is easy, but proving the optimal Golomb ruler (or rulers) for a specified order is computationally very challenging.
Distributed.net has completed distributed massively parallel searches for optimal order-24 through order-28 Golomb rulers, each time confirming the suspected candidate ruler.[3] [4] [5] [6] [7]
Currently, the complexity of finding optimal Golomb rulers (OGRs) of arbitrary order n (where n is given in unary) is unknown. In the past there was some speculation that it is an NP-hard problem.[8] Problems related to the construction of Golomb rulers are provably shown to be NP-hard, where it is also noted that no known NP-complete problem has similar flavor to finding Golomb rulers.[9]
A set of integers
A=\{a1,a2,...,am\}
a1<a2<...<am
foralli,j,k,l\in\left\{1,2,...,m\right\}suchthati ≠ jandk ≠ l, ai-aj=ak-al\iffi=kandj=l.
The order of such a Golomb ruler is
m
am-a1
a1=0
m>2
a2-a1<am-am-1
f:\left\{1,2,...,m\right\}\to\left\{0,1,...,n\right\}
f(1)=0
f(m)=n
foralli,j,k,l\in\left\{1,2,...,m\right\}suchthati ≠ jandk ≠ l,f(i)-f(j)=f(k)-f(l)\iffi=kandj=l.
The order of such a Golomb ruler is
m
n
f(2)<f(m)-f(m-1)
m>2
A Golomb ruler of order m with length n may be optimal in either of two respects:[11]
The general term optimal Golomb ruler is used to refer to the second type of optimality.
Golomb rulers are used within information theory related to error correcting codes.[12]
Golomb rulers are used in the selection of radio frequencies to reduce the effects of intermodulation interference with both terrestrial[13] and extraterrestrial[14] applications.
Golomb rulers are used in the design of phased arrays of radio antennas. In radio astronomy one-dimensional synthesis arrays can have the antennas in a Golomb ruler configuration in order to obtain minimum redundancy of the Fourier component sampling.[15] [16]
Multi-ratio current transformers use Golomb rulers to place transformer tap points.
A number of construction methods produce asymptotically optimal Golomb rulers.
The following construction, due to Paul Erdős and Pál Turán, produces a Golomb ruler for every odd prime p.[17]
2pk+(k2\bmodp),k\in[0,p-1]
The following table contains all known optimal Golomb rulers, excluding those with marks in the reverse order. The first four are perfect.
| Order | Length | Marks | Proved | Proof discovered by |
|---|---|---|---|---|
| 1 | 0 | 0 | 1952[18] | Wallace Babcock |
| 2 | 1 | 0 1 | 1952 | Wallace Babcock |
| 3 | 3 | 0 1 3 | 1952 | Wallace Babcock |
| 4 | 6 | 0 1 4 6 | 1952 | Wallace Babcock |
| 5 | 11 | 0 1 4 9 11 0 2 7 8 11 | c. 1967[19] | John P. Robinson and Arthur J. Bernstein |
| 6 | 17 | 0 1 4 10 12 17 0 1 4 10 15 17 0 1 8 11 13 17 0 1 8 12 14 17 | c. 1967 | John P. Robinson and Arthur J. Bernstein |
| 7 | 25 | 0 1 4 10 18 23 25 0 1 7 11 20 23 25 0 1 11 16 19 23 25 0 2 3 10 16 21 25 0 2 7 13 21 22 25 | c. 1967 | John P. Robinson and Arthur J. Bernstein |
| 8 | 34 | 0 1 4 9 15 22 32 34 | 1972 | William Mixon |
| 9 | 44 | 0 1 5 12 25 27 35 41 44 | 1972 | William Mixon |
| 10 | 55 | 0 1 6 10 23 26 34 41 53 55 | 1972 | William Mixon |
| 11 | 72 | 0 1 4 13 28 33 47 54 64 70 72 0 1 9 19 24 31 52 56 58 69 72 | 1972 | William Mixon |
| 12 | 85 | 0 2 6 24 29 40 43 55 68 75 76 85 | 1979 | John P. Robinson |
| 13 | 106 | 0 2 5 25 37 43 59 70 85 89 98 99 106 | 1981 | John P. Robinson |
| 14 | 127 | 0 4 6 20 35 52 59 77 78 86 89 99 122 127 | 1985 | James B. Shearer |
| 15 | 151 | 0 4 20 30 57 59 62 76 100 111 123 136 144 145 151 | 1985 | James B. Shearer |
| 16 | 177 | 0 1 4 11 26 32 56 68 76 115 117 134 150 163 168 177 | 1986 | James B. Shearer |
| 17 | 199 | 0 5 7 17 52 56 67 80 81 100 122 138 159 165 168 191 199 | 1993 | W. Olin Sibert |
| 18 | 216 | 0 2 10 22 53 56 82 83 89 98 130 148 153 167 188 192 205 216 | 1993 | W. Olin Sibert |
| 19 | 246 | 0 1 6 25 32 72 100 108 120 130 153 169 187 190 204 231 233 242 246 | 1994 | Apostolos Dollas, William T. Rankin and David McCracken |
| 20 | 283 | 0 1 8 11 68 77 94 116 121 156 158 179 194 208 212 228 240 253 259 283 | 1997? | Mark Garry, David Vanderschel et al. (web project) |
| 21 | 333 | 0 2 24 56 77 82 83 95 129 144 179 186 195 255 265 285 293 296 310 329 333 | 8 May 1998[20] | Mark Garry, David Vanderschel et al. (web project) |
| 22 | 356 | 0 1 9 14 43 70 106 122 124 128 159 179 204 223 253 263 270 291 330 341 353 356 | 1999 | Mark Garry, David Vanderschel et al. (web project) |
| 23 | 372 | 0 3 7 17 61 66 91 99 114 159 171 199 200 226 235 246 277 316 329 348 350 366 372 | 1999 | Mark Garry, David Vanderschel et al. (web project) |
| 24 | 425 | 0 9 33 37 38 97 122 129 140 142 152 191 205 208 252 278 286 326 332 353 368 384 403 425 | distributed.net | |
| 25 | 480 | 0 12 29 39 72 91 146 157 160 161 166 191 207 214 258 290 316 354 372 394 396 431 459 467 480 | distributed.net | |
| 26 | 492 | 0 1 33 83 104 110 124 163 185 200 203 249 251 258 314 318 343 356 386 430 440 456 464 475 487 492 | distributed.net | |
| 27 | 553 | 0 3 15 41 66 95 97 106 142 152 220 221 225 242 295 330 338 354 382 388 402 415 486 504 523 546 553 | distributed.net | |
| 28 | 585 | 0 3 15 41 66 95 97 106 142 152 220 221 225 242 295 330 338 354 382 388 402 415 486 504 523 546 553 585 | distributed.net | |
The optimal ruler would have been known before this date; this date represents that date when it was discovered to be optimal (because all other rulers were proved to not be smaller). For example, the ruler that turned out to be optimal for order 26 was recorded on, but it was not known to be optimal until all other possibilities were exhausted on .