Generating random numbers C++

Question

Hi, I’m writing a simulation in C++ and I have to generate some random numbers. To do this, initialize ( srand(time(NULL)) once and I have a method as follows:

void Supermercado::geraTempoProximoCliente() {
    int intervaloProximoCliente = rand()%(2*tempoMedioChegada-1) +1;

    printf("Intervalo para chegar outro cliente %d\n",intervaloProximoCliente);

    tempoChegadaProximoCliente = getRelogio() + intervaloProximoCliente;
}

Would anyone like to tell me why every time my program generates a new interval, it is always the same throughout the entire execution of the program? Builder where seed is generated:

Supermercado::Supermercado(string nome_,int tempoSimulacao_,inttempoMedioChegada_,
    int numeroDeCaixas_, int tamanhoMaximoDasFilas, Caixa caixas[]) {
listaDeCaixas = new CircularList<Caixa>();
tempoMedioChegada = tempoMedioChegada_;
relogio = 0;
clientesDesistentes = 0;
valorDesistentes = 0;
numeroDeCaixas = numeroDeCaixas_;
tempoSimulacao = tempoSimulacao_;
tempoChegadaProximoCliente = 0;
srand(time(NULL));
for(int i = 0; i < numeroDeCaixas; i++)
    listaDeCaixas -> push_back(caixas[i]);
}

I urge the supermarket once only on Main and call the method run from it.

The method perform:

void Supermercado::executa() {
    bool continua = true;
    Caixa caixa;
    geraTempoProximoCliente();
    while(relogio != tempoSimulacao*3600) {
        if(relogio == tempoChegadaProximoCliente)
            geraTempoProximoCliente();
    }
    relogio++;
}

Answer 1

C++11 can already be considered standard, and introduced a standard library for random numbers, <random>, even if it’s not the source of your problem, I’ll introduce you.

The main problem of using rand() and truncate the bits with % is that the distribution of bits in rand is not guaranteed to be uniform. Even if funcione™ in your case, I recommend using the new libraries that ensure the standard implementation of the pseudo-random generator.

Basically, you need to declare a number generator. One of the ones provided by the standard library is a 32-bit Mersenne Twister:

#include <random>
std::mt19937 mt(42); //gerador mersene twist, inicializado com seed = 42

Then you need to define how you want to map its pseudo-random output (much better than improvising with the operator %). If you want integers between zero and one hundred, evenly distributed, for example:

//distribuição linear de inteiros entre 0 e 100
std::uniform_int_distribution<int> linear_i(0, 100);

You already want real numbers linearly distributed between 0.0 and 100.0, for example:

//distribuição linear de números reais entre 0 e 100
std::uniform_real_distribution<float> linear_r(0.f, 100.f);

Once you have initialized the generator and the distribution (you can save them as a class member), every time you need a number, just call the desired distribution to map a number generated by the generator:

int   aleatorio_i = linear_i( mt ); //inteiro pseudo-aleatório
float aleatorio_f = linear_r( mt ); //float pseudo-aleatório

You can test the example I created here.

The best of the new library is that it has several ready-made distributions. The case you present - interval between arrival of customers - for example, can be modeled by a normal distribution around the average.

A while ago I made an example code of distributions. Maybe I’ll be the guide.

Answer 2

Use srand(time(0)) instead of srand(time(NULL)). See if it works.

Answer 3

Always set function srand() before function Rand(). What do you call the Rand() in

void Supermercado::geraTempoProximoCliente()

Then put the srand() before the code

int intervaloProximoCliente = rand()%(2*tempoMedioChegada-1) +1;

Thus ending:

void Supermercado::geraTempoProximoCliente() {
    srand(time(NULL));
    int intervaloProximoCliente = rand()%(2*tempoMedioChegada-1) +1;
    printf("Intervalo para chegar outro cliente %d\n",intervaloProximoCliente);
    tempoChegadaProximoCliente = getRelogio() + intervaloProximoCliente;
}

In doing so, you ensure that every time you generate a new number, the program captures a new system time to use in the function Rand().

I hope I’ve helped!