Article: 16024 of sci.math.num-analysis Xref: taurus.cs.nps.navy.mil sci.stat.consult:7790 sci.math.num-analysis:16024 Path: taurus.cs.nps.navy.mil!lll-winken.llnl.gov!uwm.edu!news.alpha.net!news.mathworks.com!udel!ssnet.com!usenet From: Bob Wheeler Newsgroups: sci.stat.consult,sci.math.num-analysis Subject: Marsaglia's Mother of all RNG's (Long?) Date: Fri, 28 Oct 94 19:32:08 EDT Organization: SSNet -- Public Internet Access in Delaware! Lines: 285 Distribution: inet Message-ID: <38s2p1$qaf@marlin.ssnet.com> NNTP-Posting-Host: echip.com Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Newsreader: NEWTNews & Chameleon -- TCP/IP for MS Windows from NetManage Several people have asked me to post this: First the C program, then George Marsaliga's post with details about the RNG. He claims a period of about 2^250 for this and that it passes all of the usual tests. I've tried it enough to be sure his claim is reasonable. The program: #include static short mother1[10]; static short mother2[10]; static short mStart=1; #define m16Long 65536L /* 2^16 */ #define m16Mask 0xFFFF /* mask for lower 16 bits */ #define m15Mask 0x7FFF /* mask for lower 15 bits */ #define m31Mask 0x7FFFFFFF /* mask for 31 bits */ #define m32Double 4294967295.0 /* 2^32-1 */ /* Mother ************************************************************** | George Marsaglia's The mother of all random number generators | producing uniformly distributed pseudo random 32 bit values with | period about 2^250. | The text of Marsaglia's posting is appended at the end of the function. | | The arrays mother1 and mother2 store carry values in their | first element, and random 16 bit numbers in elements 1 to 8. | These random numbers are moved to elements 2 to 9 and a new | carry and number are generated and placed in elements 0 and 1. | The arrays mother1 and mother2 are filled with random 16 bit values | on first call of Mother by another generator. mStart is the switch. | | Returns: | A 32 bit random number is obtained by combining the output of the | two generators and returned in *pSeed. It is also scaled by | 2^32-1 and returned as a double between 0 and 1 | | SEED: | The inital value of *pSeed may be any long value | | Bob Wheeler 8/8/94 */ double Mother(unsigned long *pSeed) { unsigned long number, number1, number2; short n, *p; unsigned short sNumber; /* Initialize motheri with 9 random values the first time */ if (mStart) { sNumber= *pSeed&m16Mask; /* The low 16 bits */ number= *pSeed&m31Mask; /* Only want 31 bits */ p=mother1; for (n=18;n--;) { number=30903*sNumber+(number>>16); /* One line multiply-with-cary */ *p++=sNumber=number&m16Mask; if (n==9) p=mother2; } /* make cary 15 bits */ mother1[0]&=m15Mask; mother2[0]&=m15Mask; mStart=0; } /* Move elements 1 to 8 to 2 to 9 */ memmove(mother1+2,mother1+1,8*sizeof(short)); memmove(mother2+2,mother2+1,8*sizeof(short)); /* Put the carry values in numberi */ number1=mother1[0]; number2=mother2[0]; /* Form the linear combinations */ number1+=1941*mother1[2]+1860*mother1[3]+1812*mother1[4]+1776*mother1[5]+ 1492*mother1[6]+1215*mother1[7]+1066*mother1[8]+12013*mother1[9]; number2+=1111*mother2[2]+2222*mother2[3]+3333*mother2[4]+4444*mother2[5]+ 5555*mother2[6]+6666*mother2[7]+7777*mother2[8]+9272*mother2[9]; /* Save the high bits of numberi as the new carry */ mother1[0]=number1/m16Long; mother2[0]=number2/m16Long; /* Put the low bits of numberi into motheri[1] */ mother1[1]=m16Mask&number1; mother2[1]=m16Mask&number2; /* Combine the two 16 bit random numbers into one 32 bit */ *pSeed=(((long)mother1[1])<<16)+(long)mother2[1]; /* Return a double value between 0 and 1 */ return ((double)*pSeed)/m32Double; } /* ********************* Marsaglia's comments Yet another RNG Random number generators are frequently posted on the network; my colleagues and I posted ULTRA in 1992 and, from the number of requests for releases to use it in software packages, it seems to be widely used. I have long been interested in RNG's and several of my early ones are used as system generators or in statistical packages. So why another one? And why here? Because I want to describe a generator, or rather, a class of generators, so promising I am inclined to call it The Mother of All Random Number Generators and because the generator seems promising enough to justify shortcutting the many months, even years, before new developments are widely known through publication in a journal. This new class leads to simple, fast programs that produce sequences with very long periods. They use multiplication, which experience has shown does a better job of mixing bits than do +,- or exclusive-or, and they do it with easily- implemented arithmetic modulo a power of 2, unlike arithmetic modulo a prime. The latter, while satisfactory, is difficult to implement. But the arithmetic here modulo 2^16 or 2^32 does not suffer the flaws of ordinary congruential generators for those moduli: trailing bits too regular. On the contrary, all bits of the integers produced by this new method, whether leading or trailing, have passed extensive tests of randomness. Here is an idea of how it works, using, say, integers of six decimal digits from which we return random 3- digit integers. Start with n=123456, the seed. Then form a new n=672*456+123=306555 and return 555. Then form a new n=672*555+306=373266 and return 266. Then form a new n=672*266+373=179125 and return 125, and so on. Got it? This is a multiply-with-carry sequence x(n)=672*x(n-1)+ carry mod b=1000, where the carry is the number of b's dropped in the modular reduction. The resulting sequence of 3- digit x's has period 335,999. Try it. No big deal, but that's just an example to give the idea. Now consider the sequence of 16-bit integers produced by the two C statements: k=30903*(k&65535)+(k>>16); return(k&65535); Notice that it is doing just what we did in the example: multiply the bottom half (by 30903, carefully chosen), add the top half and return the new bottom. That will produce a sequence of 16-bit integers with period > 2^29, and if we concatenate two such: k=30903*(k&65535)+(k>>16); j=18000*(j&65535)+(j>>16); return((k<<16)+j); we get a sequence of more than 2^59 32-bit integers before cycling. The following segment in a (properly initialized) C procedure will generate more than 2^118 32-bit random integers from six random seed values i,j,k,l,m,n: k=30903*(k&65535)+(k>>16); j=18000*(j&65535)+(j>>16); i=29013*(i&65535)+(i>>16); l=30345*(l&65535)+(l>>16); m=30903*(m&65535)+(m>>16); n=31083*(n&65535)+(n>>16); return((k+i+m)>>16)+j+l+n); And it will do it much faster than any of several widely used generators designed to use 16-bit integer arithmetic, such as that of Wichman-Hill that combines congruential sequences for three 15-bit primes (Applied Statistics, v31, p188-190, 1982), period about 2^42. I call these multiply-with-carry generators. Here is an extravagant 16-bit example that is easily implemented in C or Fortran. It does such a thorough job of mixing the bits of the previous eight values that it is difficult to imagine a test of randomness it could not pass: x[n]=12013x[n-8]+1066x[n-7]+1215x[n-6]+1492x[n-5]+1776x[n-4] +1812x[n-3]+1860x[n-2]+1941x[n-1]+carry mod 2^16. The linear combination occupies at most 31 bits of a 32-bit integer. The bottom 16 is the output, the top 15 the next carry. It is probably best to implement with 8 case segments. It takes 8 microseconds on my PC. Of course it just provides 16-bit random integers, but awfully good ones. For 32 bits you would have to combine it with another, such as x[n]=9272x[n-8]+7777x[n-7]+6666x[n-6]+5555x[n-5]+4444x[n-4] +3333x[n-3]+2222x[n-2]+1111x[n-1]+carry mod 2^16. Concatenating those two gives a sequence of 32-bit random integers (from 16 random 16-bit seeds), period about 2^250. It is so awesome it may merit the Mother of All RNG's title. The coefficients in those two linear combinations suggest that it is easy to get long-period sequences, and that is true. The result is due to Cemal Kac, who extended the theory we gave for add-with-carry sequences: Choose a base b and give r seed values x[1],...,x[r] and an initial 'carry' c. Then the multiply-with-carry sequence x[n]=a1*x[n-1]+a2*x[n-2]+...+ar*x[n-r]+carry mod b, where the new carry is the number of b's dropped in the modular reduction, will have period the order of b in the group of residues relatively prime to m=ar*b^r+...+a1b^1-1. Furthermore, the x's are, in reverse order, the digits in the expansion of k/m to the base b, for some 0 2^92, for 32-bit arithmetic: x[n]=1111111464*(x[n-1]+x[n-2]) + carry mod 2^32. Suppose you have functions, say top() and bot(), that give the top and bottom halves of a 64-bit result. Then, with initial 32-bit x, y and carry c, simple statements such as y=bot(1111111464*(x+y)+c) x=y c=top(y) will, repeated, give over 2^92 random 32-bit y's. Not many machines have 64 bit integers yet. But most assemblers for modern CPU's permit access to the top and bottom halves of a 64-bit product. I don't know how to readily access the top half of a 64-bit product in C. Can anyone suggest how it might be done? (in integer arithmetic) George Marsaglia geo@stat.fsu.edu