Generate 0 and 1

Hi,

How to generate 0 and 1 with equal probability?

I wanna generate sequence like 010101 or 011001 i.e probability of 0 and 1 should be 0.5.

[gkboy]

Hey,I am trying to get the random matrix with the elements only 1 and 0 where none of columns and rows can be 0 only and also the number of 0s and 1s should be in a percentage of 30% and 70%. have to use probability

Requirements 1 Random Matrix 2. elements only 1s and 0s 3. percentage of 0s: 1s is 3:7 4. None of the columns and rows can be with only 0 and only 1. have to have mix

[James Tursa]

This is sufficiently different from the original question that I would advise you delete this comment and instead post it as a new Question.

[Srik G]

I wanna generate sequence like 010101 or 011001 with some sampling time in matlab- simulink blocks. Please suggest the block which can do this job in simulink ?

Answer 1

[Walter Roberson]

round(rand)

or

rand >= 0.5

[Ronak Sakaria]

well i want equal number of 1 and 0 to be generated in my answer. this syntax wouldnt do this.

[Walter Roberson]

T = [zeros(1,N), ones(1,N)];

T = T(randperm(2*N));

Warning: if you force the number of 1 and 0 to be equal, then that is _not_ a Uniform Random Distribution. The last bit could be completely predicted by knowing all the other bits.

[Katyayani Modi]

He i have one question I want to generate 1's or 0's but 1's will occur 70% of the time with round, ceil, fix, floor function.

[Image Analyst]

To get a specified percentage of 1's, try this:

numberOfElements = 1000; % Whatever.
percentageOfOnes = 70; % Whatever.
numberOfOnes = round(numberOfElements * percentageOfOnes / 100)
% Make initial signal with proper number of 0's and 1's.
signal = [ones(1, numberOfOnes), zeros(1, numberOfElements - numberOfOnes)];
% Scramble them up with randperm
signal = signal(randperm(length(signal)));
% Count them just to prove it
numOnes = sum(signal)

Answer 2

[Image Analyst]

Ronak, This will do the trick:

n = 10; % Total length must be even to have even # of 0s and 1s.
numberOfOnes = n/2
% Get a list of random locations, with no number repeating.
indexes = randperm(n)
% Start off with all zeros.
x = zeros(1, n);
% Now make half of them, in random locations, a 1.
x(indexes(1:numberOfOnes)) = 1

[Ronak Sakaria]

thanks buddy

Answer 3

[Geoff]

I wouldn't do this using the double version of rand. While it's not really incorrect to use rand for this, I would use randi instead. Integers are the native data type for random number generation.

This generates an m-by-n matrix of zeros and ones:

 x = randi(2,m,n) - 1

There is still a chance, however, that when generating six random numbers they will be all 1, all 0, or anything else where the distribution is not perfect. If you require an equal number of ones and zeros, but scrambled, you could do this:

% By definition, we require an EVEN number for N
N = 6;
x = repmat([0 1], 1, N/2);

% Randomly switch two individual numbers N times.
for t = randi(N, 2, N)
  x(t) = x(t([2 1]));
end

[edited: made swap code less C-like]

[Ronak Sakaria]

thats the solution I'm looking for. Thanks Geoff

[Geoff]

Oh, thanks Walter for pointing out the randperm() function. Didn't know that one. Ronak, to explain my loop (since you emailed me to ask about it):

I generate a 2xN array of indices and loop over it. Each iteration of the loop gives me a column-vector 't' containing two random indices. I then swap the values of 'x' at those two indices (even if they're the same). The x(t([2 1])) gives me the same two values of x but in reverse order (since I reversed the 't' vector).

[Daniel Shub]

I think randi uses the same random number generator as rand. As none of the random number generator give you logicals (what you want), I don't see why int8 is that much better than double.

[Walter Roberson]

Integers are (nearly always) what underlies uniform random generators. However, other distributions such as rand() may use other methods.

MATLAB does not expose the 32 bit integer generation capability of the Twister algorithm. If I recall, it uses two Twister calls to generate the 53 bit double in rand().

I do not have access to randi() to check, but because it needs to be uniform random over a variable range of integer values (rather than always over a power of 2), fairness considerations usually make it easier to use a double and multiply by the span.

[Walter Roberson]

I meant to write that randn() may use other methods.

Answer 4

[Narayan Chaudhary]

randerr(1,6,[0 3;0 1])

1x6 matrix with 0 and three non-zero entries with probability of 3 non zero entries being 1 i.e. 0,5 in overall matrix

[juan camilo lores acosta]

muchas gracias salvaste una vida :)

Answer 5

[Haoming Mai]

Hi

How to generate 0 and 1 with probability of 1/3?

[Image Analyst]

With randperm

total = 99
numOnes = round(total/3)
r = zeros(1, total); % All zeros to start.
% Now assign a third of them to 1
indexes = randperm(total, numOnes);
r(indexes) = 1
sum(r)  % Check to make sure.  Should be 33

I already showed this above in https://www.mathworks.com/matlabcentral/answers/37827-generate-0-and-1#comment_486991

Answer 6

[Walter Roberson]

A few years after this question was posted, I posted

https://www.mathworks.com/matlabcentral/answers/246999-probability-1-2-with-matlab-random-number-generators

In which I proved that it is not possible to generate 0 and 1 with exactly equal probability with any of the MATLAB random number generators.

Answer 7

[Akira Agata]

Just FYI:

To evaluate BER (Bit Error Rate) of digital communication system, PRBS (Pseudo Random Binary Sequence) has been commonly used as a test pattern for many years. Since PRBS sequence with length of bits always contains specific number of 1s and 0s (more precisely, 1s and 0s ), maybe you can use this for your purpose.

% For example, the following code generates PRBS sequence of r=4 (15bit length)
pnSequence = comm.PNSequence(...
    'Polynomial',       [4 3 0],...
    'InitialConditions',[0 0 0 1],...
    'SamplesPerFrame',  15)
prbsSequence = pnSequence();

Answer 8

[Prakash Reddy Pasham]

I just need the 1D array of zeros and Ones with our required probability of ones.

[Walter Roberson]

https://www.mathworks.com/matlabcentral/answers/37827-generate-0-and-1#comment_538887 shows how to achieve a specific density of ones -- for example, exactly 33 ones out of 99.

Note that this is not the same as the probability being 33/99 = 1/3 : probability is statistical. For example,

simulations = mean(rand(1000,1000) <= 1/3,1);
min(simulations), max(simulations)

ans = 0.2840

ans = 0.3870

histogram(simulations)

Answer 9

[Omair Mohammad Ikram]

What is the code for generation of a random sequence of bits with a specified probability of p0 of the bit 0?

[Image Analyst]

Try this:

p0 = 60; % Percent of zeros that we need in the final sequence.
numElements = 10000; % However many you want.
% Initialize a sequence of 1's.
sequence = ones(1, numElements);
numZeroes = round(numElements * p0 / 100);
% Get locations to randomly sprinkle the 0's into.
randomIndexes = randperm(numElements, numZeroes);
% Set those locations to 1.
sequence(randomIndexes) = 0;
% Verify
percentageOfOnes = 100 * sum(sequence) / numElements
percentageOfZeroes = 100 * sum(~sequence) / numElements

Adapt as needed.

[Omair Mohammad Ikram]

Thanks alot

[Walter Roberson]

Do you need probability or do you need a specific number of values in a state? If you need exactly floor(p0*samples) then use randperm like above. If you want probability then

out = rand(1,samples) > p0;

for probability p0 of a 0.

More common would be to specify the probability of a 1, in which case you would typically use <.