Using Boost PRNG to make a huge lookup table of random numbers - c++

I'm trying to use Boost's normal distribution to generate random numbers given different seeds. In other words, I need the same random numbers produced for seed1, seed2, etc.; thousands of seeds will be passed to the function over the course of the simulation. The random number generator will never be used unseeded. [Edit: "Key" is a better word than "seed"--see final description block below.] I'm not sure whether it makes the most sense to generate a single RNG and reseed it (and if so, how) or if it's easier to generate a new one each time. Here's what I have so far, which involves the construction of a new, seeded rng at each request for a random normal number:
double rnorm( int thisSeed ) {
boost::mt19937 rng( thisSeed );
boost::normal_distribution<> nd( 0.0, 1.0 ); // (mean, sd)
boost::variate_generator > var_nor( rng, nd );
return var_nor();
}
Is this dumb? I'm new to PRNGs and especially Boost's implementation.
A more thorough description of why I'm doing this:
I am creating a huge random energy landscape to simulate protein interactions: each sequence has a particular energy that's calculated as the sum of quenched Gaussian random numbers that depend on the values of particular amino acids at particular positions (and a few other sequence attributes). I want to use the PRNG to calculate what these pseudorandom values are: these values must be consistent (the same sequence should yield the same values), but there are way too many to store. As a simple example, I might have a sequence ARNDAMR and compute its total energy based on two subenergies: one is a random normal number that depends on having A in position 1 and D at position 4, and the other subenergy is a random number that depends on the last three amino acids. I'm converting the configurations into keys for use as seeds (arguments) for my PRNG. Many thousands of sequences will be constructed and mutated, so I need a way to compute energies quickly--so I need to know how best to seed and call my RNG. I will not be using the Boost RNG for anything other than these energy value "lookups."
Further (tl;dr) explanation:
I am going to have "key" values that are integers between 1 and 10^6 or 10^7. I want each to map to a Gaussian random number. There should not be any cross-correlation between the key values and their numbers (e.g., keys 145-148 should not map to autocorrelated "random" numbers).
I need a given key to return the same random number each time it (the key) is called in the simulation. I do not want to store the key-random number pairs in a lookup table.

Your approach fundmentally misunderstands how PRNGs work. If you reseed on every use, then you won't get random numbers at all, you'll just get a bad hash function of the seed. In particular, your numbers won't be normally distributed even if you're calling the PRNG's normal distribution function, because the PRNG only guarantees that the random numbers generated from a particular seed will be normal.
If you need a large set of random numbers to be repeatable for a specific set of inputs, then generate a single number which is a function of those inputs, seed the PRNG with that, then get numbers from the PRNG in a predictable sequence; it will produce the same sequence for the same inputs, and the numbers will be properly distributed by the PRNG.
If the set of inputs you use to determine the random sequence is large (and in particular, larger that the size of the seed for your PRNG), then you won't have a unique sequence for every set of inputs. That might be OK for your application, or you might want to use a PRNG with larger seeds.
Take a look at my public domain ojrandlib. It uses big seeds, and generates normally distributed numbers with the fast Ziggurat algorithm.
Edit after seeing your clarification:
Ah, now I see. There's no such thing as "a" Gaussian random. Distribution only makes sense with regard to the whole sequence from one seed, so what you need to do is create and seed a single generator, then fetch the Nth random value from that generator for each of your keys N. If you're not doing this in order (that is, if you're fetching from keys totally at random and not as part of a sequence) this will be very slow, but still possible. You may want to see if you can force a sequence, say by sorting the keys before you fetch them.
ojrandlib has a function discard() for this too, so that if you need to find the 1,000,000th number in a sequence, you can seed the PRNG and discard 999,999 of them, which is faster than actually generating them, but will still be pretty slow.
Probably better: instead of using your key to seed a Gaussian generator, compute a good hash function of the key + fixed seed (which will result in uniformly distributed random bits), then interpret those hash bits as two uniform floats, then do the Box-Muller or Ziggurat with those to transform the distribution. That way, the numbers you get will be all from the same "seed" (which is the input to the hash), but normally distributed. You don't need a cryptographically secure hash, so something like MurMurHash might work well, though you would probably be better off rolling your own for such a special purpose.
Thought users of my library might have similar problems to yours, so I investigated some possibilities. Here's some code that might work for you:
/* Thomas Wang's 32-bit integer hash */
uint32_t nth_rand32(uint32_t a) {
a -= a << 6;
a ^= a >> 17;
a -= a << 9;
a ^= a << 4;
a -= a << 3;
a ^= a << 10;
a ^= a >> 15;
return a;
}
/* Marsaglia polar method */
double nth_normal(int index) {
double f, g, w;
int skip = 0;
uint64_t x, y;
do {
x = (uint64_t)nth_rand32((index & ~1) + skip);
y = (uint64_t)nth_rand32((index | 1) + skip);
skip += 0x40000001;
x = (x << 20) | 0x3ff0000000000000ull;
f = *(double *)(&x) * 2.0 - 3.0;
y = (y << 20) | 0x3ff0000000000000ull;
g = *(double *)(&y) * 2.0 - 3.0;
w = f * f + g * g;
} while (w >= 1.0 || w == 0.0);
w = sqrt((-2.0 * log(w)) / w);
if (index & 1) w *= f;
else w *= g;
return w;
}
The hash doesn't pass diehard, but it's pretty good. I generated 10,000,000 random normals, and got this distribution (if this image upload works):
Not perfect, but not too bad. It would be a lot better with a more expensive hash, but I'll let you decide where the speed/accuracy tradeoff is for you.

Related

Why is rand()%6 biased?

When reading how to use std::rand, I found this code on cppreference.com
int x = 7;
while(x > 6)
x = 1 + std::rand()/((RAND_MAX + 1u)/6); // Note: 1+rand()%6 is biased
What is wrong with the expression on the right? Tried it and it works perfectly.
There are two issues with rand() % 6 (the 1+ doesn't affect either problem).
First, as several answers have pointed out, if the low bits of rand() aren't appropriately uniform, the result of the remainder operator is also not uniform.
Second, if the number of distinct values produced by rand() is not a multiple of 6, then the remainder will produce more low values than high values. That's true even if rand() returns perfectly distributed values.
As an extreme example, pretend that rand() produces uniformly distributed values in the range [0..6]. If you look at the remainders for those values, when rand() returns a value in the range [0..5], the remainder produces uniformly distributed results in the range [0..5]. When rand() returns 6, rand() % 6 returns 0, just as if rand() had returned 0. So you get a distribution with twice as many 0's as any other value.
The second is the real problem with rand() % 6.
The way to avoid that problem is to discard values that would produce non-uniform duplicates. You calculate the largest multiple of 6 that's less than or equal to RAND_MAX, and whenever rand() returns a value that's greater than or equal to that multiple you reject it and call `rand() again, as many times a needed.
So:
int max = 6 * ((RAND_MAX + 1u) / 6)
int value = rand();
while (value >= max)
value = rand();
That's a different implementation of the code in question, intended to more clearly show what's going on.
There are hidden depths here:
The use of the small u in RAND_MAX + 1u. RAND_MAX is defined to be an int type, and is often the largest possible int. The behaviour of RAND_MAX + 1 would be undefined in such instances as you'd be overflowing a signed type. Writing 1u forces type conversion of RAND_MAX to unsigned, so obviating the overflow.
The use of % 6 can (but on every implementation of std::rand I've seen doesn't) introduce any additional statistical bias above and beyond the alternative presented. Such instances where % 6 is hazardous are cases where the number generator has correlation plains in the low order bits, such as a rather famous IBM implementation (in C) of rand in, I think, the 1970s which flipped the high and low bits as "a final flourish". A further consideration is that 6 is very small cf. RAND_MAX, so there will be a minimal effect if RAND_MAX is not a multiple of 6, which it probably isn't.
In conclusion, these days, due to its tractability, I'd use % 6. It's not likely to introduce any statistical anomalies beyond those introduced by the generator itself. If you are still in doubt, test your generator to see if it has the appropriate statistical properties for your use case.
This example code illustrates that std::rand is a case of legacy cargo cult balderdash that should make your eyebrows raise every time you see it.
There are several issues here:
The contract people usually assume—even the poor hapless souls who don't know any better and won't think of it in precisely these terms—is that rand samples from the uniform distribution on the integers in 0, 1, 2, …, RAND_MAX, and each call yields an independent sample.
The first problem is that the assumed contract, independent uniform random samples in each call, is not actually what the documentation says—and in practice, implementations historically failed to provide even the barest simulacrum of independence. For example, C99 §7.20.2.1 ‘The rand function’ says, without elaboration:
The rand function computes a sequence of pseudo-random integers in the range 0 to RAND_MAX.
This is a meaningless sentence, because pseudorandomness is a property of a function (or family of functions), not of an integer, but that doesn't stop even ISO bureaucrats from abusing the language. After all, the only readers who would be upset by it know better than to read the documentation for rand for fear of their brain cells decaying.
A typical historical implementation in C works like this:
static unsigned int seed = 1;
static void
srand(unsigned int s)
{
seed = s;
}
static unsigned int
rand(void)
{
seed = (seed*1103515245 + 12345) % ((unsigned long)RAND_MAX + 1);
return (int)seed;
}
This has the unfortunate property that even though a single sample may be uniformly distributed under a uniform random seed (which depends on the specific value of RAND_MAX), it alternates between even and odd integers in consecutive calls—after
int a = rand();
int b = rand();
the expression (a & 1) ^ (b & 1) yields 1 with 100% probability, which is not the case for independent random samples on any distribution supported on even and odd integers. Thus, a cargo cult emerged that one should discard the low-order bits to chase the elusive beast of ‘better randomness’. (Spoiler alert: This is not a technical term. This is a sign that whosever prose you are reading either doesn't know what they're talking about, or thinks you are clueless and must be condescended to.)
The second problem is that even if each call did sample independently from a uniform random distribution on 0, 1, 2, …, RAND_MAX, the outcome of rand() % 6 would not be uniformly distributed in 0, 1, 2, 3, 4, 5 like a die roll, unless RAND_MAX is congruent to -1 modulo 6. Simple counterexample: If RAND_MAX = 6, then from rand(), all outcomes have equal probability 1/7, but from rand() % 6, the outcome 0 has probability 2/7 while all other outcomes have probability 1/7.
The right way to do this is with rejection sampling: repeatedly draw an independent uniform random sample s from 0, 1, 2, …, RAND_MAX, and reject (for example) the outcomes 0, 1, 2, …, ((RAND_MAX + 1) % 6) - 1—if you get one of those, start over; otherwise, yield s % 6.
unsigned int s;
while ((s = rand()) < ((unsigned long)RAND_MAX + 1) % 6)
continue;
return s % 6;
This way, the set of outcomes from rand() that we accept is evenly divisible by 6, and each possible outcome from s % 6 is obtained by the same number of accepted outcomes from rand(), so if rand() is uniformly distributed then so is s. There is no bound on the number of trials, but the expected number is less than 2, and the probability of success grows exponentially with the number of trials.
The choice of which outcomes of rand() you reject is immaterial, provided that you map an equal number of them to each integer below 6. The code at cppreference.com makes a different choice, because of the first problem above—that nothing is guaranteed about the distribution or independence of outputs of rand(), and in practice the low-order bits exhibited patterns that don't ‘look random enough’ (never mind that the next output is a deterministic function of the previous one).
Exercise for the reader: Prove that the code at cppreference.com yields a uniform distribution on die rolls if rand() yields a uniform distribution on 0, 1, 2, …, RAND_MAX.
Exercise for the reader: Why might you prefer one or the other subsets to reject? What computation is needed for each trial in the two cases?
A third problem is that the seed space is so small that even if the seed is uniformly distributed, an adversary armed with knowledge of your program and one outcome but not the seed can readily predict the seed and subsequent outcomes, which makes them seem not so random after all. So don't even think about using this for cryptography.
You can go the fancy overengineered route and C++11's std::uniform_int_distribution class with an appropriate random device and your favorite random engine like the ever-popular Mersenne twister std::mt19937 to play at dice with your four-year-old cousin, but even that is not going to be fit for generating cryptographic key material—and the Mersenne twister is a terrible space hog too with a multi-kilobyte state wreaking havoc on your CPU's cache with an obscene setup time, so it is bad even for, e.g., parallel Monte Carlo simulations with reproducible trees of subcomputations; its popularity likely arises mainly from its catchy name. But you can use it for toy dice rolling like this example!
Another approach is to use a simple cryptographic pseudorandom number generator with a small state, such as a simple fast key erasure PRNG, or just a stream cipher such as AES-CTR or ChaCha20 if you are confident (e.g., in a Monte Carlo simulation for research in the natural sciences) that there are no adverse consequences to predicting past outcomes if the state is ever compromised.
I'm not an experienced C++ user by any means, but was interested to see if the other answers regarding
std::rand()/((RAND_MAX + 1u)/6) being less biased than 1+std::rand()%6 actually holds true. So I wrote a test program to tabulate the results for both methods (I haven't written C++ in ages, please check it). A link for running the code is found here. It's also reproduced as follows:
// Example program
#include <cstdlib>
#include <iostream>
#include <ctime>
#include <string>
int main()
{
std::srand(std::time(nullptr)); // use current time as seed for random generator
// Roll the die 6000000 times using the supposedly unbiased method and keep track of the results
int results[6] = {0,0,0,0,0,0};
// roll a 6-sided die 20 times
for (int n=0; n != 6000000; ++n) {
int x = 7;
while(x > 6)
x = 1 + std::rand()/((RAND_MAX + 1u)/6); // Note: 1+rand()%6 is biased
results[x-1]++;
}
for (int n=0; n !=6; n++) {
std::cout << results[n] << ' ';
}
std::cout << "\n";
// Roll the die 6000000 times using the supposedly biased method and keep track of the results
int results_bias[6] = {0,0,0,0,0,0};
// roll a 6-sided die 20 times
for (int n=0; n != 6000000; ++n) {
int x = 7;
while(x > 6)
x = 1 + std::rand()%6;
results_bias[x-1]++;
}
for (int n=0; n !=6; n++) {
std::cout << results_bias[n] << ' ';
}
}
I then took the output of this and used the chisq.test function in R to run a Chi-square test to see if the results are significantly different than expected. This stackexchange question goes into more detail of using the chi-square test to test die fairness: How can I test whether a die is fair?. Here are the results for a few runs:
> ?chisq.test
> unbias <- c(100150, 99658, 100319, 99342, 100418, 100113)
> bias <- c(100049, 100040, 100091, 99966, 100188, 99666 )
> chisq.test(unbias)
Chi-squared test for given probabilities
data: unbias
X-squared = 8.6168, df = 5, p-value = 0.1254
> chisq.test(bias)
Chi-squared test for given probabilities
data: bias
X-squared = 1.6034, df = 5, p-value = 0.9008
> unbias <- c(998630, 1001188, 998932, 1001048, 1000968, 999234 )
> bias <- c(1000071, 1000910, 999078, 1000080, 998786, 1001075 )
> chisq.test(unbias)
Chi-squared test for given probabilities
data: unbias
X-squared = 7.051, df = 5, p-value = 0.2169
> chisq.test(bias)
Chi-squared test for given probabilities
data: bias
X-squared = 4.319, df = 5, p-value = 0.5045
> unbias <- c(998630, 999010, 1000736, 999142, 1000631, 1001851)
> bias <- c(999803, 998651, 1000639, 1000735, 1000064,1000108)
> chisq.test(unbias)
Chi-squared test for given probabilities
data: unbias
X-squared = 7.9592, df = 5, p-value = 0.1585
> chisq.test(bias)
Chi-squared test for given probabilities
data: bias
X-squared = 2.8229, df = 5, p-value = 0.7273
In the three runs that I did, the p-value for both methods was always greater than typical alpha values used to test significance (0.05). This means that we wouldn't consider either of them to be biased. Interestingly, the supposedly unbiased method has consistently lower p-values, which indicates that it might actually be more biased. The caveat being that I only did 3 runs.
UPDATE: While I was writing my answer, Konrad Rudolph posted an answer that takes the same approach, but gets a very different result. I don't have the reputation to comment on his answer, so I'm going to address it here. First, the main thing is that the code he uses uses the same seed for the random number generator every time it's run. If you change the seed, you actually get a variety of results. Second, if you don't change the seed, but change the number of trials, you also get a variety of results. Try increasing or decreasing by an order of magnitude to see what I mean. Third, there is some integer truncation or rounding going on where the expected values aren't quite accurate. It probably isn't enough to make a difference, but it's there.
Basically, in summary, he just happened to get the right seed and number of trials that he might be getting a false result.
One can think of a random number generator as working on a stream of binary digits. The generator turns the stream into numbers by slicing it up into chunks. If the std:rand function is working with a RAND_MAX of 32767, then it is using 15 bits in each slice.
When one takes the modules of a number between 0 and 32767 inclusive one finds that 5462 '0's and '1's but only 5461 '2's, '3's, '4's, and '5's. Hence the result is biased. The larger the RAND_MAX value is, the less bias there will be, but it is inescapable.
What is not biased is a number in the range [0..(2^n)-1]. You can generate a (theoretically) better number in the range 0..5 by extracting 3 bits, converting them to an integer in the range 0..7 and rejecting 6 and 7.
One hopes that every bit in the bit stream has an equal chance of being a '0' or a '1' irrespective of where it is in the stream or the values of other bits. This is exceptionally difficult in practice. The many different implementations of software PRNGs offer different compromises between speed and quality. A linear congruential generator such as std::rand offers fastest speed for lowest quality. A cryptographic generator offers highest quality for lowest speed.

Generating random integers with a difference constraint

I have the following problem:
Generate M uniformly random integers from the range 0-N, where N >> M, and where no pair has a difference less than K. where M >> K.
At the moment the best method I can think of is to maintain a sorted list, then determine the lower bound of the current generated integer and test it with the lower and upper elements, if it's ok to then insert the element in between. This is of complexity O(nlogn).
Would there happen to be a more efficient algorithm?
An example of the problem:
Generate 1000 uniformly random integers between zero and 100million where the difference between any two integers is no less than 1000
A comprehensive way to solve this would be to:
Determine all the combinations of n-choose-m that satisfy the constraint, lets called it set X
Select a uniformly random integer i in the range [0,|X|).
Select the i'th combination from X as the result.
This solution is problematic when the n-choose-m is large, as enumerating and storing all possible combinations will be extremely costly. Hence an efficient online generating solution is sought.
Note: The following is a C++ implementation of the solution provided by pentadecagon
std::vector<int> generate_random(const int n, const int m, const int k)
{
if ((n < m) || (m < k))
return std::vector<int>();
std::random_device source;
std::mt19937 generator(source());
std::uniform_int_distribution<> distribution(0, n - (m - 1) * k);
std::vector<int> result_list;
result_list.reserve(m);
for (int i = 0; i < m; ++i)
{
result_list.push_back(distribution(generator));
}
std::sort(std::begin(result_list),std::end(result_list));
for (int i = 0; i < m; ++i)
{
result_list[i] += (i * k);
}
return result_list;
}
http://ideone.com/KOeR4R
.
EDIT: I adapted the text for the requirement to create ordered sequences, each with the same probability.
Create random numbers a_i for i=0..M-1 without duplicates. Sort them. Then create numbers
b_i=a_i + i*(K-1)
Given the construction, those numbers b_i have the required gaps, because the a_i already have gaps of at least 1. In order to make sure those b values cover exactly the required range [1..N], you must ensure a_i are picked from a range [1..N-(M-1)*(K-1)]. This way you get truly independent numbers. Well, as independent as possible given the required gap. Because of the sorting you get O(M log M) performance again, but this shouldn't be too bad. Sorting is typically very fast. In Python it looks like this:
import random
def random_list( N, M, K ):
s = set()
while len(s) < M:
s.add( random.randint( 1, N-(M-1)*(K-1) ) )
res = sorted( s )
for i in range(M):
res[i] += i * (K-1)
return res
First off: this will be an attempt to show that there's a bijection between the (M+1)- compositions (with the slight modification that we will allow addends to be 0) of the value N - (M-1)*K and the valid solutions to your problem. After that, we only have to pick one of those compositions uniformly at random and apply the bijection.
Bijection:
Let
Then the xi form an M+1-composition (with 0 addends allowed) of the value on the left (notice that the xi do not have to be monotonically increasing!).
From this we get a valid solution
by setting the values mi as follows:
We see that the distance between mi and mi + 1 is at least K, and mM is at most N (compare the choice of the composition we started out with). This means that every (M+1)-composition that fulfills the conditions above defines exactly one valid solution to your problem. (You'll notice that we only use the xM as a way to make the sum turn out right, we don't use it for the construction of the mi.)
To see that this gives a bijection, we need to see that the construction can be reversed; for this purpose, let
be a given solution fulfilling your conditions. To get the composition this is constructed from, define the xi as follows:
Now first, all xi are at least 0, so that's alright. To see that they form a valid composition (again, every xi is allowed to be 0) of the value given above, consider:
The third equality follows since we have this telescoping sum that cancels out almost all mi.
So we've seen that the described construction gives a bijection between the described compositions of N - (M-1)*K and the valid solutions to your problem. All we have to do now is pick one of those compositions uniformly at random and apply the construction to get a solution.
Picking a composition uniformly at random
Each of the described compositions can be uniquely identified in the following way (compare this for illustration): reserve N - (M-1)*K spaces for the unary notation of that value, and another M spaces for M commas. We get an (M+1)- composition of N - (M-1)*K by choosing M of the N - (M-1)*K + M spaces, putting commas there, and filling the rest with |. Then let x0 be the number of | before the first comma, xM+1 the number of | after the last comma, and all other xi the number of | between commas i and i+1. So all we have to do is pick an M-element subset of the integer interval[1; N - (M-1)*K + M] uniformly at random, which we can do for example with the Fisher-Yates shuffle in O(N + M log M) (we need to sort the M delimiters to build the composition) since M*K needs to be in O(N) for any solutions to exist. So if N is bigger than M by at least a logarithmic factor, then this is linear in N.
Note: #DavidEisenstat suggested that there are more space efficient ways of picking the M-element subset of that interval; I'm not aware of any, I'm afraid.
You can get an error-proof algorithm out of this by doing the simple input validation we get from the construction above that N ≥ (M-1) * K and that all three values are at least 1 (or 0, if you define the empty set as a valid solution for that case).
Why not do this:
for (int i = 0; i < M; ++i) {
pick a random number between K and N/M
add this number to (N/M)* i;
Now you have M random numbers, distributed evenly along N, all of which have a difference of at least K. It's in O(n) time. As an added bonus, it's already sorted. :-)
EDIT:
Actually, the "pick a random number" part shouldn't be between K and N/M, but between min(K, [K - (N/M * i - previous value)]). That would ensure that the differences are still at least K, and not exclude values that should not be missed.
Second EDIT:
Well, the first case shouldn't be between K and N/M - it should be between 0 and N/M. Just like you need special casing for when you get close to the N/M*i border, we need special initial casing.
Aside from that, the issue you brought up in your comments was fair representation, and you're right. As my pseudocode is presented, it currently completely misses the excess between N/M*M and N. It's another edge case; simply change the random values of your last range.
Now, in this case, your distribution will be different for the last range. Since you have more numbers, you have slightly less chance for each number than you do for all the other ranges. My understanding is that because you're using ">>", this shouldn't really impact the distribution, i.e. the difference in size in the sample set should be nominal. But if you want to make it more fair, you divide the excess equally among each range. This makes your initial range calculation more complex - you'll have to augment each range based on how much remainder there is divided by M.
There are lots of special cases to look out for, but they're all able to be handled. I kept the pseudocode very basic just to make sure that the general concept came through clearly. If nothing else, it should be a good starting point.
Third and Final EDIT:
For those worried that the distribution has a forced evenness, I still claim that there's nothing saying it can't. The selection is uniformly distributed in each segment. There is a linear way to keep it uneven, but that also has a trade-off: if one value is selected extremely high (which should be unlikely given a very large N), then all the other values are constrained:
int prevValue = 0;
int maxRange;
for (int i = 0; i < M; ++i) {
maxRange = N - (((M - 1) - i) * K) - prevValue;
int nextValue = random(0, maxRange);
prevValue += nextValue;
store previous value;
prevValue += K;
}
This is still linear and random and allows unevenness, but the bigger prevValue gets, the more constrained the other numbers become. Personally, I prefer my second edit answer, but this is an available option that given a large enough N is likely to satisfy all the posted requirements.
Come to think of it, here's one other idea. It requires a lot more data maintenance, but is still O(M) and is probably the most fair distribution:
What you need to do is maintain a vector of your valid data ranges and a vector of probability scales. A valid data range is just the list of high-low values where K is still valid. The idea is you first use the scaled probability to pick a random data range, then you randomly pick a value within that range. You remove the old valid data range and replace it with 0, 1 or 2 new data ranges in the same position, depending on how many are still valid. All of these actions are constant time other than handling the weighted probability, which is O(M), done in a loop M times, so the total should be O(M^2), which should be much better than O(NlogN) because N >> M.
Rather than pseudocode, let me work an example using OP's original example:
0th iteration: valid data ranges are from [0...100Mill], and the weight for this range is 1.0.
1st iteration: Randomly pick one element in the one element vector, then randomly pick one element in that range.
If the element is, e.g. 12345678, then we remove the [0...100Mill] and replace it with [0...12344678] and [12346678...100Mill]
If the element is, e.g. 500, then we remove the [0...100Mill] and replace it with just [1500...100Mill], since [0...500] is no longer a valid range. The only time we will replace it with 0 ranges is in the unlikely event that you have a range with only one number in it and it gets picked. (In that case, you'll have 3 numbers in a row that are exactly K apart from each other.)
The weight for the ranges are their length over the total length, e.g. 12344678/(12344678 + (100Mill - 12346678)) and (100Mill - 12346678)/(12344678 + (100Mill - 12346678))
In the next iterations, you do the same thing: randomly pick a number between 0 and 1 and determine which of the ranges that scale falls into. Then randomly pick a number in that range, and replace your ranges and scales.
By the time it's done, we're no longer acting in O(M), but we're still only dependent on the time of M instead of N. And this actually is both uniform and fair distribution.
Hope one of these ideas works for you!

What is the most efficient way to generate unique pseudo-random numbers? [duplicate]

Duplicate:
Unique random numbers in O(1)?
I want an pseudo random number generator that can generate numbers with no repeats in a random order.
For example:
random(10)
might return
5, 9, 1, 4, 2, 8, 3, 7, 6, 10
Is there a better way to do it other than making the range of numbers and shuffling them about, or checking the generated list for repeats?
Edit:
Also I want it to be efficient in generating big numbers without the entire range.
Edit:
I see everyone suggesting shuffle algorithms. But if I want to generate large random number (1024 byte+) then that method would take alot more memory than if I just used a regular RNG and inserted into a Set until it was a specified length, right? Is there no better mathematical algorithm for this.
You may be interested in a linear feedback shift register.
We used to build these out of hardware, but I've also done them in software. It uses a shift register with some of the bits xor'ed and fed back to the input, and if you pick just the right "taps" you can get a sequence that's as long as the register size. That is, a 16-bit lfsr can produce a sequence 65535 long with no repeats. It's statistically random but of course eminently repeatable. Also, if it's done wrong, you can get some embarrassingly short sequences. If you look up the lfsr, you will find examples of how to construct them properly (which is to say, "maximal length").
A shuffle is a perfectly good way to do this (provided you do not introduce a bias using the naive algorithm). See Fisher-Yates shuffle.
If a random number is guaranteed to never repeat it is no longer random and the amount of randomness decreases as the numbers are generated (after nine numbers random(10) is rather predictable and even after only eight you have a 50-50 chance).
I understand tou don't want a shuffle for large ranges, since you'd have to store the whole list to do so.
Instead, use a reversible pseudo-random hash. Then feed in the values 0 1 2 3 4 5 6 etc in turn.
There are infinite numbers of hashes like this. They're not too hard to generate if they're restricted to a power of 2, but any base can be used.
Here's one that would work for example if you wanted to go through all 2^32 32 bit values. It's easiest to write because the implicit mod 2^32 of integer math works to your advantage in this case.
unsigned int reversableHash(unsigned int x)
{
x*=0xDEADBEEF;
x=x^(x>>17);
x*=0x01234567;
x+=0x88776655;
x=x^(x>>4);
x=x^(x>>9);
x*=0x91827363;
x=x^(x>>7);
x=x^(x>>11);
x=x^(x>>20);
x*=0x77773333;
return x;
}
If you don't mind mediocre randomness properties and if the number of elements allows it then you could use a linear congruential random number generator.
A shuffle is the best you can do for random numbers in a specific range with no repeats. The reason that the method you describe (randomly generate numbers and put them in a Set until you reach a specified length) is less efficient is because of duplicates. Theoretically, that algorithm might never finish. At best it will finish in an indeterminable amount of time, as compared to a shuffle, which will always run in a highly predictable amount of time.
Response to edits and comments:
If, as you indicate in the comments, the range of numbers is very large and you want to select relatively few of them at random with no repeats, then the likelihood of repeats diminishes rapidly. The bigger the difference in size between the range and the number of selections, the smaller the likelihood of repeat selections, and the better the performance will be for the select-and-check algorithm you describe in the question.
What about using GUID generator (like in the one in .NET). Granted it is not guaranteed that there will be no duplicates, however the chance getting one is pretty low.
This has been asked before - see my answer to the previous question. In a nutshell: You can use a block cipher to generate a secure (random) permutation over any range you want, without having to store the entire permutation at any point.
If you want to creating large (say, 64 bits or greater) random numbers with no repeats, then just create them. If you're using a good random number generator, that actually has enough entropy, then the odds of generating repeats are so miniscule as to not be worth worrying about.
For instance, when generating cryptographic keys, no one actually bothers checking to see if they've generated the same key before; since you're trusting your random number generator that a dedicated attacker won't be able to get the same key out, then why would you expect that you would come up with the same key accidentally?
Of course, if you have a bad random number generator (like the Debian SSL random number generator vulnerability), or are generating small enough numbers that the birthday paradox gives you a high chance of collision, then you will need to actually do something to ensure you don't get repeats. But for large random numbers with a good generator, just trust probability not to give you any repeats.
As you generate your numbers, use a Bloom filter to detect duplicates. This would use a minimal amount of memory. There would be no need to store earlier numbers in the series at all.
The trade off is that your list could not be exhaustive in your range. If your numbers are truly on the order of 256^1024, that's hardly any trade off at all.
(Of course if they are actually random on that scale, even bothering to detect duplicates is a waste of time. If every computer on earth generated a trillion random numbers that size every second for trillions of years, the chance of a collision is still absolutely negligible.)
I second gbarry's answer about using an LFSR. They are very efficient and simple to implement even in software and are guaranteed not to repeat in (2^N - 1) uses for an LFSR with an N-bit shift-register.
There are some drawbacks however: by observing a small number of outputs from the RNG, one can reconstruct the LFSR and predict all values it will generate, making them not usable for cryptography and anywhere were a good RNG is needed. The second problem is that either the all zero word or the all one (in terms of bits) word is invalid depending on the LFSR implementation. The third issue which is relevant to your question is that the maximum number generated by the LFSR is always a power of 2 - 1 (or power of 2 - 2).
The first drawback might not be an issue depending on your application. From the example you gave, it seems that you are not expecting zero to be among the answers; so, the second issue does not seem relevant to your case.
The maximum value (and thus range) problem can solved by reusing the LFSR until you get a number within your range. Here's an example:
Say you want to have numbers between 1 and 10 (as in your example). You would use a 4-bit LFSR which has a range [1, 15] inclusive. Here's a pseudo code as to how to get number in the range [1,10]:
x = LFSR.getRandomNumber();
while (x > 10) {
x = LFSR.getRandomNumber();
}
You should embed the previous code in your RNG; so that the caller wouldn't care about implementation.
Note that this would slow down your RNG if you use a large shift-register and the maximum number you want is not a power of 2 - 1.
This answer suggests some strategies for getting what you want and ensuring they are in a random order using some already well-known algorithms.
There is an inside out version of the Fisher-Yates shuffle algorithm, called the Durstenfeld version, that randomly distributes sequentially acquired items into arrays and collections while loading the array or collection.
One thing to remember is that the Fisher-Yates (AKA Knuth) shuffle or the Durstenfeld version used at load time is highly efficient with arrays of objects because only the reference pointer to the object is being moved and the object itself doesn't have to be examined or compared with any other object as part of the algorithm.
I will give both algorithms further below.
If you want really huge random numbers, on the order of 1024 bytes or more, a really good random generator that can generate unsigned bytes or words at a time will suffice. Randomly generate as many bytes or words as you need to construct the number, make it into an object with a reference pointer to it and, hey presto, you have a really huge random integer. If you need a specific really huge range, you can add a base value of zero bytes to the low-order end of the byte sequence to shift the value up. This may be your best option.
If you need to eliminate duplicates of really huge random numbers, then that is trickier. Even with really huge random numbers, removing duplicates also makes them significantly biased and not random at all. If you have a really large set of unduplicated really huge random numbers and you randomly select from the ones not yet selected, then the bias is only the bias in creating the huge values for the really huge set of numbers from which to choose. A reverse version of Durstenfeld's version of the Yates-Fisher could be used to randomly choose values from a really huge set of them, remove them from the remaining values from which to choose and insert them into a new array that is a subset and could do this with just the source and target arrays in situ. This would be very efficient.
This may be a good strategy for getting a small number of random numbers with enormous values from a really large set of them in which they are not duplicated. Just pick a random location in the source set, obtain its value, swap its value with the top element in the source set, reduce the size of the source set by one and repeat with the reduced size source set until you have chosen enough values. This is essentiall the Durstenfeld version of Fisher-Yates in reverse. You can then use the Dursenfeld version of the Fisher-Yates algorithm to insert the acquired values into the destination set. However, that is overkill since they should be randomly chosen and randomly ordered as given here.
Both algorithms assume you have some random number instance method, nextInt(int setSize), that generates a random integer from zero to setSize meaning there are setSize possible values. In this case, it will be the size of the array since the last index to the array is size-1.
The first algorithm is the Durstenfeld version of Fisher-Yates (aka Knuth) shuffle algorithm as applied to an array of arbitrary length, one that simply randomly positions integers from 0 to the length of the array into the array. The array need not be an array of integers, but can be an array of any objects that are acquired sequentially which, effectively, makes it an array of reference pointers. It is simple, short and very effective
int size = someNumber;
int[] int array = new int[size]; // here is the array to load
int location; // this will get assigned a value before used
// i will also conveniently be the value to load, but any sequentially acquired
// object will work
for (int i = 0; i <= size; i++) { // conveniently, i is also the value to load
// you can instance or acquire any object at this place in the algorithm to load
// by reference, into the array and use a pointer to it in place of j
int j = i; // in this example, j is trivially i
if (i == 0) { // first integer goes into first location
array[i] = j; // this may get swapped from here later
} else { // subsequent integers go into random locations
// the next random location will be somewhere in the locations
// already used or a new one at the end
// here we get the next random location
// to preserve true randomness without a significant bias
// it is REALLY IMPORTANT that the newest value could be
// stored in the newest location, that is,
// location has to be able to randomly have the value i
int location = nextInt(i + 1); // a random value between 0 and i
// move the random location's value to the new location
array[i] = array[location];
array[location] = j; // put the new value into the random location
} // end if...else
} // end for
Voila, you now have an already randomized array.
If you want to randomly shuffle an array you already have, here is the standard Fisher-Yates algorithm.
type[] array = new type[size];
// some code that loads array...
// randomly pick an item anywhere in the current array segment,
// swap it with the top element in the current array segment,
// then shorten the array segment by 1
// just as with the Durstenfeld version above,
// it is REALLY IMPORTANT that an element could get
// swapped with itself to avoid any bias in the randomization
type temp; // this will get assigned a value before used
int location; // this will get assigned a value before used
for (int i = arrayLength -1 ; i > 0; i--) {
int location = nextInt(i + 1);
temp = array[i];
array[i] = array[location];
array[location] = temp;
} // end for
For sequenced collections and sets, i.e. some type of list object, you could just use adds/or inserts with an index value that allows you to insert items anywhere, but it has to allow adding or appending after the current last item to avoid creating bias in the randomization.
Shuffling N elements doesn't take up excessive memory...think about it. You only swap one element at a time, so the maximum memory used is that of N+1 elements.
Assuming you have a random or pseudo-random number generator, even if it's not guaranteed to return unique values, you can implement one that returns unique values each time using this code, assuming that the upper limit remains constant (i.e. you always call it with random(10), and don't call it with random(10); random(11).
The code doesn't check for errors. You can add that yourself if you want to.
It also requires a lot of memory if you want a large range of numbers.
/* the function returns a random number between 0 and max -1
* not necessarily unique
* I assume it's written
*/
int random(int max);
/* the function returns a unique random number between 0 and max - 1 */
int unique_random(int max)
{
static int *list = NULL; /* contains a list of numbers we haven't returned */
static int in_progress = 0; /* 0 --> we haven't started randomizing numbers
* 1 --> we have started randomizing numbers
*/
static int count;
static prev_max = 0;
// initialize the list
if (!in_progress || (prev_max != max)) {
if (list != NULL) {
free(list);
}
list = malloc(sizeof(int) * max);
prev_max = max;
in_progress = 1;
count = max - 1;
int i;
for (i = max - 1; i >= 0; --i) {
list[i] = i;
}
}
/* now choose one from the list */
int index = random(count);
int retval = list[index];
/* now we throw away the returned value.
* we do this by shortening the list by 1
* and replacing the element we returned with
* the highest remaining number
*/
swap(&list[index], &list[count]);
/* when the count reaches 0 we start over */
if (count == 0) {
in_progress = 0;
free(list);
list = 0;
} else { /* reduce the counter by 1 */
count--;
}
}
/* swap two numbers */
void swap(int *x, int *y)
{
int temp = *x;
*x = *y;
*y = temp;
}
Actually, there's a minor point to make here; a random number generator which is not permitted to repeat is not random.
Suppose you wanted to generate a series of 256 random numbers without repeats.
Create a 256-bit (32-byte) memory block initialized with zeros, let's call it b
Your looping variable will be n, the number of numbers yet to be generated
Loop from n = 256 to n = 1
Generate a random number r in the range [0, n)
Find the r-th zero bit in your memory block b, let's call it p
Put p in your list of results, an array called q
Flip the p-th bit in memory block b to 1
After the n = 1 pass, you are done generating your list of numbers
Here's a short example of what I am talking about, using n = 4 initially:
**Setup**
b = 0000
q = []
**First loop pass, where n = 4**
r = 2
p = 2
b = 0010
q = [2]
**Second loop pass, where n = 3**
r = 2
p = 3
b = 0011
q = [2, 3]
**Third loop pass, where n = 2**
r = 0
p = 0
b = 1011
q = [2, 3, 0]
** Fourth and final loop pass, where n = 1**
r = 0
p = 1
b = 1111
q = [2, 3, 0, 1]
Please check answers at
Generate sequence of integers in random order without constructing the whole list upfront
and also my answer lies there as
very simple random is 1+((power(r,x)-1) mod p) will be from 1 to p for values of x from 1 to p and will be random where r and p are prime numbers and r <> p.
I asked a similar question before but mine was for the whole range of a int see Looking for a Hash Function /Ordered Int/ to /Shuffled Int/
static std::unordered_set<long> s;
long l = 0;
for(; !l && (s.end() != s.find(l)); l = generator());
v.insert(l);
generator() being your random number generator. You roll numbers as long as the entry is not in your set, then you add what you find in it. You get the idea.
I did it with long for the example, but you should make that a template if your PRNG is templatized.
Alternative is to use a cryptographically secure PRNG that will have a very low probability to generate twice the same number.
If you don't mean poor statisticall properties of generated sequence, there is one method:
Let's say you want to generate N numbers, each of 1024 bits each. You can sacrifice some bits of generated number to be "counter".
So you generate each random number, but into some bits you choosen you put binary encoded counter (from variable, you increase each time next random number is generated).
You can split that number into single bits and put it in some of less significant bits of generated number.
That way you are sure you get unique number each time.
I mean for example each generated number looks like that:
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxyyxxxxyxyyyyxxyxx
where x is take directly from generator, and ys are taken from counter variable.
Mersenne twister
Description of which can be found here on Wikipedia: Mersenne twister
Look at the bottom of the page for implementations in various languages.
The problem is to select a "random" sequence of N unique numbers from the range 1..M where there is no constraint on the relationship between N and M (M could be much bigger, about the same, or even smaller than N; they may not be relatively prime).
Expanding on the linear feedback shift register answer: for a given M, construct a maximal LFSR for the smallest power of two that is larger than M. Then just grab your numbers from the LFSR throwing out numbers larger than M. On average, you will throw out at most half the generated numbers (since by construction more than half the range of the LFSR is less than M), so the expected running time of getting a number is O(1). You are not storing previously generated numbers so space consumption is O(1) too. If you cycle before getting N numbers then M less than N (or the LFSR is constructed incorrectly).
You can find the parameters for maximum length LFSRs up to 168 bits here (from wikipedia): http://www.xilinx.com/support/documentation/application_notes/xapp052.pdf
Here's some java code:
/**
* Generate a sequence of unique "random" numbers in [0,M)
* #author dkoes
*
*/
public class UniqueRandom
{
long lfsr;
long mask;
long max;
private static long seed = 1;
//indexed by number of bits
private static int [][] taps = {
null, // 0
null, // 1
null, // 2
{3,2}, //3
{4,3},
{5,3},
{6,5},
{7,6},
{8,6,5,4},
{9,5},
{10,7},
{11,9},
{12,6,4,1},
{13,4,3,1},
{14,5,3,1},
{15,14},
{16,15,13,4},
{17,14},
{18,11},
{19,6,2,1},
{20,17},
{21,19},
{22,21},
{23,18},
{24,23,22,17},
{25,22},
{26,6,2,1},
{27,5,2,1},
{28,25},
{29,27},
{30,6,4,1},
{31,28},
{32,22,2,1},
{33,20},
{34,27,2,1},
{35,33},
{36,25},
{37,5,4,3,2,1},
{38,6,5,1},
{39,35},
{40,38,21,19},
{41,38},
{42,41,20,19},
{43,42,38,37},
{44,43,18,17},
{45,44,42,41},
{46,45,26,25},
{47,42},
{48,47,21,20},
{49,40},
{50,49,24,23},
{51,50,36,35},
{52,49},
{53,52,38,37},
{54,53,18,17},
{55,31},
{56,55,35,34},
{57,50},
{58,39},
{59,58,38,37},
{60,59},
{61,60,46,45},
{62,61,6,5},
{63,62},
};
//m is upperbound; things break if it isn't positive
UniqueRandom(long m)
{
max = m;
lfsr = seed; //could easily pass a starting point instead
//figure out number of bits
int bits = 0;
long b = m;
while((b >>>= 1) != 0)
{
bits++;
}
bits++;
if(bits < 3)
bits = 3;
mask = 0;
for(int i = 0; i < taps[bits].length; i++)
{
mask |= (1L << (taps[bits][i]-1));
}
}
//return -1 if we've cycled
long next()
{
long ret = -1;
if(lfsr == 0)
return -1;
do {
ret = lfsr;
//update lfsr - from wikipedia
long lsb = lfsr & 1;
lfsr >>>= 1;
if(lsb == 1)
lfsr ^= mask;
if(lfsr == seed)
lfsr = 0; //cycled, stick
ret--; //zero is stuck state, never generated so sub 1 to get it
} while(ret >= max);
return ret;
}
}
Here is a way to random without repeating results. It also works for strings. Its in C# but the logig should work in many places. Put the random results in a list and check if the new random element is in that list. If not than you have a new random element. If it is in that list, repeat the random until you get an element that is not in that list.
List<string> Erledigte = new List<string>();
private void Form1_Load(object sender, EventArgs e)
{
label1.Text = "";
listBox1.Items.Add("a");
listBox1.Items.Add("b");
listBox1.Items.Add("c");
listBox1.Items.Add("d");
listBox1.Items.Add("e");
}
private void button1_Click(object sender, EventArgs e)
{
Random rand = new Random();
int index=rand.Next(0, listBox1.Items.Count);
string rndString = listBox1.Items[index].ToString();
if (listBox1.Items.Count <= Erledigte.Count)
{
return;
}
else
{
if (Erledigte.Contains(rndString))
{
//MessageBox.Show("vorhanden");
while (Erledigte.Contains(rndString))
{
index = rand.Next(0, listBox1.Items.Count);
rndString = listBox1.Items[index].ToString();
}
}
Erledigte.Add(rndString);
label1.Text += rndString;
}
}
For a sequence to be random there should not be any auto correlation. The restriction that the numbers should not repeat means the next number should depend on all the previous numbers which means it is not random anymore....
If you can generate 'small' random numbers, you can generate 'large' random numbers by integrating them: add a small random increment to each 'previous'.
const size_t amount = 100; // a limited amount of random numbers
vector<long int> numbers;
numbers.reserve( amount );
const short int spread = 250; // about 250 between each random number
numbers.push_back( myrandom( spread ) );
for( int n = 0; n != amount; ++n ) {
const short int increment = myrandom( spread );
numbers.push_back( numbers.back() + increment );
}
myshuffle( numbers );
The myrandom and myshuffle functions I hereby generously delegate to others :)
to have non repeated random numbers and to avoid waistingtime with checking for doubles numbers and get new numbers over and over use the below method which will assure the minimum usage of Rand:
for example if you want to get 100 non repeated random number:
1. fill an array with numbers from 1 to 100
2. get a random number using Rand function in the range of (1-100)
3. use the genarted random number as an Index to get th value from the array (Numbers[IndexGeneratedFromRandFunction]
4. shift the number in the array after that Index to the left
5. repeat from step 2 but now the the rang should be (1-99) and go on
now we have a array with different numbers!
int main() {
int b[(the number
if them)];
for (int i = 0; i < (the number of them); i++) {
int a = rand() % (the number of them + 1) + 1;
int j = 0;
while (j < i) {
if (a == b[j]) {
a = rand() % (the number of them + 1) + 1;
j = -1;
}
j++;
}
b[i] = a;
}
}

How to get 2 random (different) elements from a c++ vector

I would like to get 2 random different elements from an std::vector. How can I do this so that:
It is fast (it is done thousands of times in my algorithm)
It is elegant
The elements selection is really uniformly distributed
For elegance and simplicty:
void Choose (const int size, int &first, int &second)
{
// pick a random element
first = rand () * size / MAX_RAND;
// pick a random element from what's left (there is one fewer to choose from)...
second = rand () * (size - 1) / MAX_RAND;
// ...and adjust second choice to take into account the first choice
if (second >= first)
{
++second;
}
}
using first and second to index the vector.
For uniformness, this is very tricky since as size approaches RAND_MAX there will be a bias towards the lower values and if size exceeds RAND_MAX then there will be elements that are never chosen. One solution to overcome this is to use a binary search:
int GetRand (int size)
{
int lower = 0, upper = size;
do
{
int mid = (lower + upper) / 2;
if (rand () > RAND_MAX / 2) // not a great test, perhaps use parity of rand ()?
{
lower = mid;
}
else
{
upper = mid;
}
} while (upper != lower); // this is just to show the idea,
// need to cope with lower == mid and lower != upper
// and all the other edge conditions
return lower;
}
What you need is to generate M uniformly distributed random numbers from [0, N) range, but there is one caveat here.
One needs to note that your statement of the problem is ambiguous. What is meant by the uniformly distributed selection? One thing is to say that each index has to be selected with equal probability (of M/N, of course). Another thing is to say that each two-index combination has to be selected with equal probability. These two are not the same. Which one did you have in mind?
If M is considerably smaller than N, the classic algorithm for selecting M numbers out of [0, N) range is Bob Floyd algorithm that can be found in Bentley's "Programming Peals" book. It looks as follows (a sketch)
for (int j = N - M; i < N; ++j) {
int rand = random(0, j); // generate a random integer in range [0, j]
if (`rand` has not been generated before)
output rand;
else
output j;
}
In order to implement the check of whether rand has already been generated or not for relatively high M some kind of implementation of a set is necessary, but in your case M=2 it is straightforward and easy.
Note that this algorithm distributes the sets of M numbers uniformly. Also, this algorithm requires exactly M iterations (attempts) to generate M random numbers, i.e. it doesn't follow that flawed "trial-and-error" approach often used in various ad-hoc algorithms intended to solve the same problem.
Adapting the above to your specific situation, the correct algorithm will look as follows
first = random(0, N - 2);
second = random(0, N - 1);
if (second == first)
second = N - 1;
(I leave out the internal details of random(a, b) as an implementation detail).
It might not be immediately obvious why the above works correctly and produces a truly uniform distribution, but it really does :)
How about using a std::queue and doing std::random_shuffle on them. Then just pop til your hearts content?
Not elegant, but extreamly simple: just draw a random number in [0, vector.size()[ and check it's not twice the same.
Simplicity is also in some way elegance ;)
What do you call fast ? I guess this can be done thousands of times within a millisecond.
Whenever need something random, you are going to have various questions about the random number properties regarding uniformity, distribution and so on.
Assuming you've found a suitable source of randomness for your application, then the simplest way to generate pairs of uncorrelated entries is just to pick two random indexes and test them to ensure they aren't equal.
Given a vector of N+1 entries, another option is to generate an index i in the range 0..N. element[i] is choice one. Swap elements i and N. Generate an index j in the range 0..(N-1). element[j] is your second choice. This slowly shuffles your vector which may be problematical, but it can be avoided by using a second vector which holds indexes into the first, and shuffling that. This method trades a swap for the index comparison and tends to be more efficient for small vectors (a dozen or fewer elements, typically) as it avoids having to do multiple comparisons as the number of collisions increase.
You might wanna look into the gnu scientific library. There are some pretty nice random number generators in there that are guaranteed to be random down to the bit level.

Create Random Number Sequence with No Repeats

Duplicate:
Unique random numbers in O(1)?
I want an pseudo random number generator that can generate numbers with no repeats in a random order.
For example:
random(10)
might return
5, 9, 1, 4, 2, 8, 3, 7, 6, 10
Is there a better way to do it other than making the range of numbers and shuffling them about, or checking the generated list for repeats?
Edit:
Also I want it to be efficient in generating big numbers without the entire range.
Edit:
I see everyone suggesting shuffle algorithms. But if I want to generate large random number (1024 byte+) then that method would take alot more memory than if I just used a regular RNG and inserted into a Set until it was a specified length, right? Is there no better mathematical algorithm for this.
You may be interested in a linear feedback shift register.
We used to build these out of hardware, but I've also done them in software. It uses a shift register with some of the bits xor'ed and fed back to the input, and if you pick just the right "taps" you can get a sequence that's as long as the register size. That is, a 16-bit lfsr can produce a sequence 65535 long with no repeats. It's statistically random but of course eminently repeatable. Also, if it's done wrong, you can get some embarrassingly short sequences. If you look up the lfsr, you will find examples of how to construct them properly (which is to say, "maximal length").
A shuffle is a perfectly good way to do this (provided you do not introduce a bias using the naive algorithm). See Fisher-Yates shuffle.
If a random number is guaranteed to never repeat it is no longer random and the amount of randomness decreases as the numbers are generated (after nine numbers random(10) is rather predictable and even after only eight you have a 50-50 chance).
I understand tou don't want a shuffle for large ranges, since you'd have to store the whole list to do so.
Instead, use a reversible pseudo-random hash. Then feed in the values 0 1 2 3 4 5 6 etc in turn.
There are infinite numbers of hashes like this. They're not too hard to generate if they're restricted to a power of 2, but any base can be used.
Here's one that would work for example if you wanted to go through all 2^32 32 bit values. It's easiest to write because the implicit mod 2^32 of integer math works to your advantage in this case.
unsigned int reversableHash(unsigned int x)
{
x*=0xDEADBEEF;
x=x^(x>>17);
x*=0x01234567;
x+=0x88776655;
x=x^(x>>4);
x=x^(x>>9);
x*=0x91827363;
x=x^(x>>7);
x=x^(x>>11);
x=x^(x>>20);
x*=0x77773333;
return x;
}
If you don't mind mediocre randomness properties and if the number of elements allows it then you could use a linear congruential random number generator.
A shuffle is the best you can do for random numbers in a specific range with no repeats. The reason that the method you describe (randomly generate numbers and put them in a Set until you reach a specified length) is less efficient is because of duplicates. Theoretically, that algorithm might never finish. At best it will finish in an indeterminable amount of time, as compared to a shuffle, which will always run in a highly predictable amount of time.
Response to edits and comments:
If, as you indicate in the comments, the range of numbers is very large and you want to select relatively few of them at random with no repeats, then the likelihood of repeats diminishes rapidly. The bigger the difference in size between the range and the number of selections, the smaller the likelihood of repeat selections, and the better the performance will be for the select-and-check algorithm you describe in the question.
What about using GUID generator (like in the one in .NET). Granted it is not guaranteed that there will be no duplicates, however the chance getting one is pretty low.
This has been asked before - see my answer to the previous question. In a nutshell: You can use a block cipher to generate a secure (random) permutation over any range you want, without having to store the entire permutation at any point.
If you want to creating large (say, 64 bits or greater) random numbers with no repeats, then just create them. If you're using a good random number generator, that actually has enough entropy, then the odds of generating repeats are so miniscule as to not be worth worrying about.
For instance, when generating cryptographic keys, no one actually bothers checking to see if they've generated the same key before; since you're trusting your random number generator that a dedicated attacker won't be able to get the same key out, then why would you expect that you would come up with the same key accidentally?
Of course, if you have a bad random number generator (like the Debian SSL random number generator vulnerability), or are generating small enough numbers that the birthday paradox gives you a high chance of collision, then you will need to actually do something to ensure you don't get repeats. But for large random numbers with a good generator, just trust probability not to give you any repeats.
As you generate your numbers, use a Bloom filter to detect duplicates. This would use a minimal amount of memory. There would be no need to store earlier numbers in the series at all.
The trade off is that your list could not be exhaustive in your range. If your numbers are truly on the order of 256^1024, that's hardly any trade off at all.
(Of course if they are actually random on that scale, even bothering to detect duplicates is a waste of time. If every computer on earth generated a trillion random numbers that size every second for trillions of years, the chance of a collision is still absolutely negligible.)
I second gbarry's answer about using an LFSR. They are very efficient and simple to implement even in software and are guaranteed not to repeat in (2^N - 1) uses for an LFSR with an N-bit shift-register.
There are some drawbacks however: by observing a small number of outputs from the RNG, one can reconstruct the LFSR and predict all values it will generate, making them not usable for cryptography and anywhere were a good RNG is needed. The second problem is that either the all zero word or the all one (in terms of bits) word is invalid depending on the LFSR implementation. The third issue which is relevant to your question is that the maximum number generated by the LFSR is always a power of 2 - 1 (or power of 2 - 2).
The first drawback might not be an issue depending on your application. From the example you gave, it seems that you are not expecting zero to be among the answers; so, the second issue does not seem relevant to your case.
The maximum value (and thus range) problem can solved by reusing the LFSR until you get a number within your range. Here's an example:
Say you want to have numbers between 1 and 10 (as in your example). You would use a 4-bit LFSR which has a range [1, 15] inclusive. Here's a pseudo code as to how to get number in the range [1,10]:
x = LFSR.getRandomNumber();
while (x > 10) {
x = LFSR.getRandomNumber();
}
You should embed the previous code in your RNG; so that the caller wouldn't care about implementation.
Note that this would slow down your RNG if you use a large shift-register and the maximum number you want is not a power of 2 - 1.
This answer suggests some strategies for getting what you want and ensuring they are in a random order using some already well-known algorithms.
There is an inside out version of the Fisher-Yates shuffle algorithm, called the Durstenfeld version, that randomly distributes sequentially acquired items into arrays and collections while loading the array or collection.
One thing to remember is that the Fisher-Yates (AKA Knuth) shuffle or the Durstenfeld version used at load time is highly efficient with arrays of objects because only the reference pointer to the object is being moved and the object itself doesn't have to be examined or compared with any other object as part of the algorithm.
I will give both algorithms further below.
If you want really huge random numbers, on the order of 1024 bytes or more, a really good random generator that can generate unsigned bytes or words at a time will suffice. Randomly generate as many bytes or words as you need to construct the number, make it into an object with a reference pointer to it and, hey presto, you have a really huge random integer. If you need a specific really huge range, you can add a base value of zero bytes to the low-order end of the byte sequence to shift the value up. This may be your best option.
If you need to eliminate duplicates of really huge random numbers, then that is trickier. Even with really huge random numbers, removing duplicates also makes them significantly biased and not random at all. If you have a really large set of unduplicated really huge random numbers and you randomly select from the ones not yet selected, then the bias is only the bias in creating the huge values for the really huge set of numbers from which to choose. A reverse version of Durstenfeld's version of the Yates-Fisher could be used to randomly choose values from a really huge set of them, remove them from the remaining values from which to choose and insert them into a new array that is a subset and could do this with just the source and target arrays in situ. This would be very efficient.
This may be a good strategy for getting a small number of random numbers with enormous values from a really large set of them in which they are not duplicated. Just pick a random location in the source set, obtain its value, swap its value with the top element in the source set, reduce the size of the source set by one and repeat with the reduced size source set until you have chosen enough values. This is essentiall the Durstenfeld version of Fisher-Yates in reverse. You can then use the Dursenfeld version of the Fisher-Yates algorithm to insert the acquired values into the destination set. However, that is overkill since they should be randomly chosen and randomly ordered as given here.
Both algorithms assume you have some random number instance method, nextInt(int setSize), that generates a random integer from zero to setSize meaning there are setSize possible values. In this case, it will be the size of the array since the last index to the array is size-1.
The first algorithm is the Durstenfeld version of Fisher-Yates (aka Knuth) shuffle algorithm as applied to an array of arbitrary length, one that simply randomly positions integers from 0 to the length of the array into the array. The array need not be an array of integers, but can be an array of any objects that are acquired sequentially which, effectively, makes it an array of reference pointers. It is simple, short and very effective
int size = someNumber;
int[] int array = new int[size]; // here is the array to load
int location; // this will get assigned a value before used
// i will also conveniently be the value to load, but any sequentially acquired
// object will work
for (int i = 0; i <= size; i++) { // conveniently, i is also the value to load
// you can instance or acquire any object at this place in the algorithm to load
// by reference, into the array and use a pointer to it in place of j
int j = i; // in this example, j is trivially i
if (i == 0) { // first integer goes into first location
array[i] = j; // this may get swapped from here later
} else { // subsequent integers go into random locations
// the next random location will be somewhere in the locations
// already used or a new one at the end
// here we get the next random location
// to preserve true randomness without a significant bias
// it is REALLY IMPORTANT that the newest value could be
// stored in the newest location, that is,
// location has to be able to randomly have the value i
int location = nextInt(i + 1); // a random value between 0 and i
// move the random location's value to the new location
array[i] = array[location];
array[location] = j; // put the new value into the random location
} // end if...else
} // end for
Voila, you now have an already randomized array.
If you want to randomly shuffle an array you already have, here is the standard Fisher-Yates algorithm.
type[] array = new type[size];
// some code that loads array...
// randomly pick an item anywhere in the current array segment,
// swap it with the top element in the current array segment,
// then shorten the array segment by 1
// just as with the Durstenfeld version above,
// it is REALLY IMPORTANT that an element could get
// swapped with itself to avoid any bias in the randomization
type temp; // this will get assigned a value before used
int location; // this will get assigned a value before used
for (int i = arrayLength -1 ; i > 0; i--) {
int location = nextInt(i + 1);
temp = array[i];
array[i] = array[location];
array[location] = temp;
} // end for
For sequenced collections and sets, i.e. some type of list object, you could just use adds/or inserts with an index value that allows you to insert items anywhere, but it has to allow adding or appending after the current last item to avoid creating bias in the randomization.
Shuffling N elements doesn't take up excessive memory...think about it. You only swap one element at a time, so the maximum memory used is that of N+1 elements.
Assuming you have a random or pseudo-random number generator, even if it's not guaranteed to return unique values, you can implement one that returns unique values each time using this code, assuming that the upper limit remains constant (i.e. you always call it with random(10), and don't call it with random(10); random(11).
The code doesn't check for errors. You can add that yourself if you want to.
It also requires a lot of memory if you want a large range of numbers.
/* the function returns a random number between 0 and max -1
* not necessarily unique
* I assume it's written
*/
int random(int max);
/* the function returns a unique random number between 0 and max - 1 */
int unique_random(int max)
{
static int *list = NULL; /* contains a list of numbers we haven't returned */
static int in_progress = 0; /* 0 --> we haven't started randomizing numbers
* 1 --> we have started randomizing numbers
*/
static int count;
static prev_max = 0;
// initialize the list
if (!in_progress || (prev_max != max)) {
if (list != NULL) {
free(list);
}
list = malloc(sizeof(int) * max);
prev_max = max;
in_progress = 1;
count = max - 1;
int i;
for (i = max - 1; i >= 0; --i) {
list[i] = i;
}
}
/* now choose one from the list */
int index = random(count);
int retval = list[index];
/* now we throw away the returned value.
* we do this by shortening the list by 1
* and replacing the element we returned with
* the highest remaining number
*/
swap(&list[index], &list[count]);
/* when the count reaches 0 we start over */
if (count == 0) {
in_progress = 0;
free(list);
list = 0;
} else { /* reduce the counter by 1 */
count--;
}
}
/* swap two numbers */
void swap(int *x, int *y)
{
int temp = *x;
*x = *y;
*y = temp;
}
Actually, there's a minor point to make here; a random number generator which is not permitted to repeat is not random.
Suppose you wanted to generate a series of 256 random numbers without repeats.
Create a 256-bit (32-byte) memory block initialized with zeros, let's call it b
Your looping variable will be n, the number of numbers yet to be generated
Loop from n = 256 to n = 1
Generate a random number r in the range [0, n)
Find the r-th zero bit in your memory block b, let's call it p
Put p in your list of results, an array called q
Flip the p-th bit in memory block b to 1
After the n = 1 pass, you are done generating your list of numbers
Here's a short example of what I am talking about, using n = 4 initially:
**Setup**
b = 0000
q = []
**First loop pass, where n = 4**
r = 2
p = 2
b = 0010
q = [2]
**Second loop pass, where n = 3**
r = 2
p = 3
b = 0011
q = [2, 3]
**Third loop pass, where n = 2**
r = 0
p = 0
b = 1011
q = [2, 3, 0]
** Fourth and final loop pass, where n = 1**
r = 0
p = 1
b = 1111
q = [2, 3, 0, 1]
Please check answers at
Generate sequence of integers in random order without constructing the whole list upfront
and also my answer lies there as
very simple random is 1+((power(r,x)-1) mod p) will be from 1 to p for values of x from 1 to p and will be random where r and p are prime numbers and r <> p.
I asked a similar question before but mine was for the whole range of a int see Looking for a Hash Function /Ordered Int/ to /Shuffled Int/
static std::unordered_set<long> s;
long l = 0;
for(; !l && (s.end() != s.find(l)); l = generator());
v.insert(l);
generator() being your random number generator. You roll numbers as long as the entry is not in your set, then you add what you find in it. You get the idea.
I did it with long for the example, but you should make that a template if your PRNG is templatized.
Alternative is to use a cryptographically secure PRNG that will have a very low probability to generate twice the same number.
If you don't mean poor statisticall properties of generated sequence, there is one method:
Let's say you want to generate N numbers, each of 1024 bits each. You can sacrifice some bits of generated number to be "counter".
So you generate each random number, but into some bits you choosen you put binary encoded counter (from variable, you increase each time next random number is generated).
You can split that number into single bits and put it in some of less significant bits of generated number.
That way you are sure you get unique number each time.
I mean for example each generated number looks like that:
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxyyxxxxyxyyyyxxyxx
where x is take directly from generator, and ys are taken from counter variable.
Mersenne twister
Description of which can be found here on Wikipedia: Mersenne twister
Look at the bottom of the page for implementations in various languages.
The problem is to select a "random" sequence of N unique numbers from the range 1..M where there is no constraint on the relationship between N and M (M could be much bigger, about the same, or even smaller than N; they may not be relatively prime).
Expanding on the linear feedback shift register answer: for a given M, construct a maximal LFSR for the smallest power of two that is larger than M. Then just grab your numbers from the LFSR throwing out numbers larger than M. On average, you will throw out at most half the generated numbers (since by construction more than half the range of the LFSR is less than M), so the expected running time of getting a number is O(1). You are not storing previously generated numbers so space consumption is O(1) too. If you cycle before getting N numbers then M less than N (or the LFSR is constructed incorrectly).
You can find the parameters for maximum length LFSRs up to 168 bits here (from wikipedia): http://www.xilinx.com/support/documentation/application_notes/xapp052.pdf
Here's some java code:
/**
* Generate a sequence of unique "random" numbers in [0,M)
* #author dkoes
*
*/
public class UniqueRandom
{
long lfsr;
long mask;
long max;
private static long seed = 1;
//indexed by number of bits
private static int [][] taps = {
null, // 0
null, // 1
null, // 2
{3,2}, //3
{4,3},
{5,3},
{6,5},
{7,6},
{8,6,5,4},
{9,5},
{10,7},
{11,9},
{12,6,4,1},
{13,4,3,1},
{14,5,3,1},
{15,14},
{16,15,13,4},
{17,14},
{18,11},
{19,6,2,1},
{20,17},
{21,19},
{22,21},
{23,18},
{24,23,22,17},
{25,22},
{26,6,2,1},
{27,5,2,1},
{28,25},
{29,27},
{30,6,4,1},
{31,28},
{32,22,2,1},
{33,20},
{34,27,2,1},
{35,33},
{36,25},
{37,5,4,3,2,1},
{38,6,5,1},
{39,35},
{40,38,21,19},
{41,38},
{42,41,20,19},
{43,42,38,37},
{44,43,18,17},
{45,44,42,41},
{46,45,26,25},
{47,42},
{48,47,21,20},
{49,40},
{50,49,24,23},
{51,50,36,35},
{52,49},
{53,52,38,37},
{54,53,18,17},
{55,31},
{56,55,35,34},
{57,50},
{58,39},
{59,58,38,37},
{60,59},
{61,60,46,45},
{62,61,6,5},
{63,62},
};
//m is upperbound; things break if it isn't positive
UniqueRandom(long m)
{
max = m;
lfsr = seed; //could easily pass a starting point instead
//figure out number of bits
int bits = 0;
long b = m;
while((b >>>= 1) != 0)
{
bits++;
}
bits++;
if(bits < 3)
bits = 3;
mask = 0;
for(int i = 0; i < taps[bits].length; i++)
{
mask |= (1L << (taps[bits][i]-1));
}
}
//return -1 if we've cycled
long next()
{
long ret = -1;
if(lfsr == 0)
return -1;
do {
ret = lfsr;
//update lfsr - from wikipedia
long lsb = lfsr & 1;
lfsr >>>= 1;
if(lsb == 1)
lfsr ^= mask;
if(lfsr == seed)
lfsr = 0; //cycled, stick
ret--; //zero is stuck state, never generated so sub 1 to get it
} while(ret >= max);
return ret;
}
}
Here is a way to random without repeating results. It also works for strings. Its in C# but the logig should work in many places. Put the random results in a list and check if the new random element is in that list. If not than you have a new random element. If it is in that list, repeat the random until you get an element that is not in that list.
List<string> Erledigte = new List<string>();
private void Form1_Load(object sender, EventArgs e)
{
label1.Text = "";
listBox1.Items.Add("a");
listBox1.Items.Add("b");
listBox1.Items.Add("c");
listBox1.Items.Add("d");
listBox1.Items.Add("e");
}
private void button1_Click(object sender, EventArgs e)
{
Random rand = new Random();
int index=rand.Next(0, listBox1.Items.Count);
string rndString = listBox1.Items[index].ToString();
if (listBox1.Items.Count <= Erledigte.Count)
{
return;
}
else
{
if (Erledigte.Contains(rndString))
{
//MessageBox.Show("vorhanden");
while (Erledigte.Contains(rndString))
{
index = rand.Next(0, listBox1.Items.Count);
rndString = listBox1.Items[index].ToString();
}
}
Erledigte.Add(rndString);
label1.Text += rndString;
}
}
For a sequence to be random there should not be any auto correlation. The restriction that the numbers should not repeat means the next number should depend on all the previous numbers which means it is not random anymore....
If you can generate 'small' random numbers, you can generate 'large' random numbers by integrating them: add a small random increment to each 'previous'.
const size_t amount = 100; // a limited amount of random numbers
vector<long int> numbers;
numbers.reserve( amount );
const short int spread = 250; // about 250 between each random number
numbers.push_back( myrandom( spread ) );
for( int n = 0; n != amount; ++n ) {
const short int increment = myrandom( spread );
numbers.push_back( numbers.back() + increment );
}
myshuffle( numbers );
The myrandom and myshuffle functions I hereby generously delegate to others :)
to have non repeated random numbers and to avoid waistingtime with checking for doubles numbers and get new numbers over and over use the below method which will assure the minimum usage of Rand:
for example if you want to get 100 non repeated random number:
1. fill an array with numbers from 1 to 100
2. get a random number using Rand function in the range of (1-100)
3. use the genarted random number as an Index to get th value from the array (Numbers[IndexGeneratedFromRandFunction]
4. shift the number in the array after that Index to the left
5. repeat from step 2 but now the the rang should be (1-99) and go on
now we have a array with different numbers!
int main() {
int b[(the number
if them)];
for (int i = 0; i < (the number of them); i++) {
int a = rand() % (the number of them + 1) + 1;
int j = 0;
while (j < i) {
if (a == b[j]) {
a = rand() % (the number of them + 1) + 1;
j = -1;
}
j++;
}
b[i] = a;
}
}