Lagged Fibonacci generator

A Lagged Fibonacci generator (LFG) is an example of a pseudorandom number generator. This class of random number generator is aimed at being an improvement on the 'standard' linear congruential generator. These are based on a generalisation of the Fibonacci sequence.

The Fibonacci sequence may be described by the recurrence relation:

:$$S_n = S_{n-1} + S_{n-2}

Hence, the new term is the sum of the last two terms in the sequence. This can be generalised to the sequence:

:$$S_n equiv S_{n-j} star S_{n-k} pmod{m}, 0 < j < k

In which case, the new term is some combination of any two previous terms. m is usually a power of 2 (m = 2^M), often 2³² or 2⁶⁴. The $$star operator denotes a general binary operation. This may be either addition, subtraction, multiplication, or the bitwise arithmetic exclusive-or operator (XOR). The theory of this type of generator is rather complex, and it may not be sufficient simply to choose random values for j and k. These generators also tend to be very sensitive to initialisation.

Generators of this type employ k words of state (they 'remember' the last k values).

If the operation used is addition, then the generator is described as an "Additive Lagged Fibonacci Generator" or ALFG, if multiplication is used, it is a "Multiplicative Lagged Fibonacci Generator" or MLFG, and if the XOR operation is used, it is called a "Two-tap Generalised Feedback Shift Register" or GFSR. The Mersenne twister algorithm is a variation on a GFSR. The GFSR is also related to the "Linear Feedback Shift Register", or LFSR.

Properties of Lagged Fibonacci Generators

Lagged Fibonacci generators have a maximum period of (2^k - 1)*2^M-1 if addition or subtraction is used, and (2^k-1) if exclusive-or operations are used to combine the previous values. If, on the other hand, multiplication is used, the maximum period is (2^k - 1)*2^M-3, or 1/4 of period of the additive case.

For the generator to achieve this maximum period, the polynomial:

:y = x^k + x^j + 1

must be primitive over the integers mod 2. Values of j and k satisfying this constraint have been published in the literature. Popular pairs are:

:{j = 7, k = 10}, {j = 5, k = 17}, {j = 24, k = 55}, {j = 65, k = 71}, {j = 128, k = 159} [http://www.ccs.uky.edu/csep/RN/RN.html] , {j = 6, k = 31}, {j = 31, k = 63}, {j = 97, k = 127}, {j = 353, k = 521}, {j = 168, k = 521}, {j = 334, k = 607}, {j = 273, k = 607}, {j = 418, k = 1279} [http://www.nersc.gov/nusers/resources/software/libs/math/random/www2.0/DOCS/www/parameters.html]

Another list of possible values for j and k is on page 28 of volume 2 of "The Art of Computer Programming"; the values listed above are in bold face:

:(1,2), (1,3), (2,3), (1,4), (3,4), (2,5), (3,5), (1,6), (5,6), (1,7), (6,7), (3,7), (4,7), (4,9), (5,9), (3,10), (7,10), (2,11), (9,11), (1,15), (14,15), (4,15), (11,15), (7,15), (8,15), (3,17), (14,17), (5,17), (12,17), (6,17), (11,17), (7,18), (11,18), (3,20), (17,20), (2,21), (19,21), (1,22), (21,22), (5,23), (18,23), (9,23), (14,23), (3,25), (22,25), (7,25), (18,25), (3,28), (25,28), (9,28), (19,28), (13,28), (15,28), (2,29), (27,29), (3,31), (28,31), (6,31), (25,31), (7,31), (24,31), (13,31), (18,31), (13,33), (20,33), (2,35), (33,35), (11,36), (25,36), (4,39), (35,39), (8,39), (31,39), (14,39), (25,39), (3,41), (38,41), (20,41), (21,41), (5,47), (42,47), (14,47), (33,47), (20,47), (27,47), (21,47), (26,47), (9,49), (40,49), (12,49), (37,49), (15,49), (34,49), (22,49), (27,49), (3,52), (49,52), (19,52), (33,52), (21,52), (31,52), (24,55), (31,55), (7,57), (50,57), (22,57), (35,57), (19,58), (39,58), (1,60), (59,60), (11,60), (49,60), (1,63), (62,63), (5,63), (58,63), (31,63), (32,63), (18,65), (47,65), (32,65), (33,65), (9,68), (59,68), (33,68), (35,68), (6,71), (65,71), (9,71), (62,71), (18,71), (53,71), (20,71), (51,71), (35,71), (36,71), (25,73), (48,73), (28,73), (45,73), (31,73), (42,73), (9,79), (70,79), (19,79), (60,79), (4,81), (77,81), (16,81), (65,81), (35,81), (46,81), (13,84), (71,84), (13,87), (74,87), (38,89), (51,89), (2,93), (91,93), (21,94), (73,94), (11,95), (84,95), (17,95), (78,95), (6,97), (91,97), (12,97), (85,97), (33,97), (64,97), (34,97), (63,97), (11,98), (87,98), (27,98), (71,98)

Note that the smaller number have short periods (only a few "random" numbers are generated before the first "random" number is repeated and the sequence restarts).

It is required that at least one of the first k values chosen to initialise the generator be odd.

It has been suggested that good ratios between j and k are approximately the golden ratioFact|date=April 2007.

Problems with LFGs

The initialisation of LFGs is a very complex problem. The output of LFGs is very sensitive to initial conditions, and statistical defects may appear initially but also periodically in the output sequence unless extreme care is taken Fact|date=June 2008. Another potential problem with LFGs is that the mathematical theory behind them is incomplete, making it necessary to rely on statistical tests rather than theoretical performance. These reasons, combined with the existence of the free and very high-quality Mersenne twister algorithm tend to make 'home-brewed' implementations of LFGs less than desirable in the presence of superior alternatives.

Usage

* Freeciv uses a lagged Fibonacci generator with {j = 24, k = 55} for its random number generator.
* The Boost library includes an implementation of a lagged Fibonacci generator.
* MATLAB uses a {j=??, k = 32} generator for its rand() function ( [http://www.mathworks.com/support/solutions/data/1-169M3.html?solution=1-169M3 info here] ).
* The Oracle Database implements this generator in its DBMS_RANDOM package (available in Oracle 8 and newer versions).

ee also

* Linear congruential generator
* Mersenne twister
* FISH (cipher)
* Pike
* VIC cipher