C++ default_random_engine::seed方法代码示例

/***********************************************************
 *	Generate a random number between 0 and 1.
 ****/
double RandomNum(bool reseed) {
  static bool firstTime = true;
  static std::default_random_engine gen;
  static std::uniform_real_distribution<double> unif(0.0, 1.0);

  if (firstTime || reseed) {
    std::random_device rd;
    gen.seed(rd());
    firstTime = false;
  }  
  return unif(gen);
}

Vertex::Vertex()
{
    myWeight = 1;
    parent = 0;
    isDraggedAlong = false;
    isAbsorbed = false;
    noOfChild = 0;
    ExtraWeight = 0;
    myRealCommunity = -1;
    gen.seed(QTime::currentTime().msec());

}

TEST_P(DequeTest, TestCorrectOperations) {
	std::uniform_int_distribution <int> getOperation(0, 5);
	generator.seed(5);
	std::deque <int> stldeq;
	mydeque <int> mydeq;
	for(size_t i = 0; i < GetParam(); ++i) {
		Operations thisOper = static_cast<Operations>  (getOperation(generator));	
		switch(thisOper) {
		case PUSH_BACK: {
		    int tmpelem = dist(generator);
		    mydeq.push_back(tmpelem);
		    stldeq.push_back(tmpelem);
		}
		break;
		case PUSH_FRONT: {
		    int tmpelem = dist(generator);
		    mydeq.push_back(tmpelem);
		    stldeq.push_back(tmpelem);
		}
		break;
		case POP_BACK: {
		    if(stldeq.size()) {
			mydeq.pop_back();
			stldeq.pop_back();
			
    		    }
		}
		break;
		case POP_FRONT: {
		    if(stldeq.size()) {
			mydeq.pop_front();
			stldeq.pop_front();
    		    }
		}
		break;
		case SIZE_EQUAL: {
		    int size_stl = stldeq.end() - stldeq.begin();
		    int size_my = mydeq.end() - mydeq.begin();
		    ASSERT_EQ(size_stl, size_my);
		}
		break;
		case EQUAL: {
		    ASSERT_TRUE(std::equal(mydeq.begin(), mydeq.end(), stldeq.begin()));

		}
		break;
		default: ASSERT_TRUE(false);
		break;
		}
	}
	SUCCEED();
}

Vector2 Ball::genRandomVelocity() const
{
	Vector2 vec;

	static std::default_random_engine engine;
	engine.seed(time(nullptr));
	static std::bernoulli_distribution distribution;

	vec.setX(distribution(engine) ? 1 : -1);
	vec.setY(distribution(engine) ? 1 : -1);

	return vec;
}

    void doInit()
    {
        mutex.lock();

        printf("Connecting with game server\n");
        Network::connect(port.getName(), SERVER_NAME);

        myLife=6;

        randengine.seed(0);
    
        mutex.unlock();
    }

void InitRandom() {
  Clock clock;
  TimeStamp now;
  clock.CurrentTime(&now);

  uint64_t seed = (static_cast<uint64_t>(now.MicroSeconds()) << 32) +
                   static_cast<uint64_t>(getpid());
#ifdef HAVE_RANDOM
  generator_.seed(seed);
#elif defined(_WIN32)
  srand(seed);
#else
  srandom(seed);
#endif
}

int randomInteger(int max, int min) {
    if (max == min)
        return max;

    orAssertGreaterThan(max, min);

    if (!engineIsSeeded) {
        engine.seed(std::chrono::system_clock::now().time_since_epoch().count());
        engineIsSeeded = true;
    }

    int range = max - min;
    auto elem = distributions.find(range);
    if (elem == distributions.end()) {
        distributions[range] = std::uniform_int_distribution<int>(0, range);
        return distributions[range](engine) + min;
    } else {
        return std::get<1>(*elem)(engine) + min;
    }
}

std::string EasterEggFactory::GetRandomEggTextureFilepath()
{
    // MAKE SURE THAT THE RANDOM NUMBER GENERATOR HAS BEEN INITIALIZED.
    static bool randomNumberGeneratorInitialized = false;
    static std::default_random_engine randomNumberGenerator;
    if (!randomNumberGeneratorInitialized)
    {
        // Initialize the random number generator with a seed based on the current time.
        unsigned long seed = static_cast<unsigned long>(std::chrono::system_clock::now().time_since_epoch().count());
        randomNumberGenerator.seed(seed);

        randomNumberGeneratorInitialized = true;
    }

    // SELECT A RANDOM EGG TEXTURE.
    // An random index is selected that is restricted to the valid range of indices
    // into the egg texture array.
    unsigned int randomEggTextureIndex = ( randomNumberGenerator() % EGG_TEXTURE_FILEPATHS.size() );
    return EGG_TEXTURE_FILEPATHS[randomEggTextureIndex];
}

int main(int argc, char** argv) {
  // Google tools initialization
  google::InitGoogleLogging(argv[0]);
  google::SetUsageMessage(
      "A Connect Four game based on Minimax with Alpha-Beta prunning.");
  google::ParseCommandLineFlags(&argc, &argv, true);
  // Random seed
  PRNG.seed(FLAGS_seed);
  // Show used options
  LOG(INFO) << "-o " << FLAGS_o;
  LOG(INFO) << "-rows " << FLAGS_rows;
  LOG(INFO) << "-cols " << FLAGS_cols;
  LOG(INFO) << "-seed " << FLAGS_seed;
  LOG(INFO) << "-ai " << FLAGS_ai;
  LOG(INFO) << "-max_depth " << FLAGS_max_depth;
  LOG(INFO) << "-wh " << FLAGS_wh;
  LOG(INFO) << "-random " << FLAGS_random;
  // Play!
  Game game;
  game.Play();
  return 0;
}

int main(int argc, char* argv[])
{


    fstream myfile, gradfile;
        gradfile.open("gradsRNN.txt", ios::trunc | ios::out);  



    double dt = 1.0;
    double tau = 30.0;

    int PHASE=LEARNING;
    int RANDW = 0;
    if (argc > 1)
        for (int nn=1; nn < argc; nn++)
        {
            if (strcmp(argv[nn], "test") == 0) { PHASE = TESTING; cout << "Test mode. " << endl; }
            if (strcmp(argv[nn], "RANDW") == 0) { RANDW = 1; } // Randomize weights. Only useful for 'test' mode.
            if (strcmp(argv[nn], "METHOD") == 0) { METHOD = argv[nn+1]; }
            if (strcmp(argv[nn], "MODULTYPE") == 0) { MODULTYPE = argv[nn+1]; }
            if (strcmp(argv[nn], "SQUARING") == 0) { SQUARING = atoi(argv[nn+1]); }
            if (strcmp(argv[nn], "DEBUG") == 0) { DEBUG = atoi(argv[nn+1]); }
            if (strcmp(argv[nn], "G") == 0) { G = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "ETA") == 0) { ETA = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "TAU") == 0) { tau = atof(argv[nn+1]); }
            //if (strcmp(argv[nn], "INPUTMULT") == 0) { INPUTMULT = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "ALPHAMODUL") == 0) { ALPHAMODUL = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "PROBAMODUL") == 0) { PROBAMODUL = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "ALPHATRACE") == 0) { ALPHATRACE = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "ALPHATRACEEXC") == 0) { ALPHATRACEEXC = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "RNGSEED") == 0) { RNGSEED = atof(argv[nn+1]); }
            if (strcmp(argv[nn], "MAXDW") == 0) { MAXDW = atof(argv[nn+1]); }
        }

    string SUFFIX = "_G" + to_string(G) + "_MAXDW" + to_string(MAXDW) + "_ETA" + to_string(ETA) + "_ALPHAMODUL" + to_string(ALPHAMODUL) + "_PROBAMODUL" + to_string(PROBAMODUL) + "_SQUARING" +to_string(SQUARING) + "_MODULTYPE-" + MODULTYPE +   
        "_ALPHATRACE" + to_string(ALPHATRACE) + "_METHOD-" + METHOD +  "_ATRACEEXC" + to_string(ALPHATRACEEXC) + "_TAU" + to_string(tau) + 
        //"_INPUTMULT" +to_string(INPUTMULT) + 
        "_RNGSEED" + 
        to_string(RNGSEED);
    cout << SUFFIX << endl;

    myrng.seed(RNGSEED);
    srand(RNGSEED);


    int trialtype;


    int NBTRIALS = 507;
    int TRIALTIME = 300;
    int TIMEINPUT = 100;
    int TIMEMODUL = 210;
    int TIMERESP = 200;

    VectorXi modulmarker(NBNEUR); modulmarker.setZero();

    if (PHASE == TESTING) 
    {
        NBTRIALS = 20*NBPATTERNS;
        //NBTRIALS = 40*NBPATTERNS;
        //TRIALTIME = 1500;
    }



    MatrixXd dJ(NBNEUR, NBNEUR); dJ.setZero();
    MatrixXd win(NBNEUR, NBIN); win.setRandom(); //win.row(0).setZero(); // Input weights are uniformly chosen between -1 and 1, except possibly for output cell (not even necessary). No plasticity for input weights.

    
    win *= .2;



    MatrixXd J(NBNEUR, NBNEUR);


    cout << Uniform(myrng) << endl;
    randJ(J); // Randomize recurrent weight matrix, according to the Sompolinsky method (Gaussian(0,1), divided by sqrt(ProbaConn*N) and multiplied by G - see definition of randJ() below).

    // If in the TESTING mode, read the weights from a previously saved file:
    if (PHASE == TESTING){
        if (RANDW == 0){
        //readWeights(J, "J.dat");
        //readWeights(win, "win.dat");
        readWeights(J, "J" + SUFFIX + ".dat");
        readWeights(win, "win" + SUFFIX + ".dat"); // win doesn't change over time.
        }
        else cout << "Using random weights." << endl;
    }


//cout << J(0,0) << " " << win(1,1) << endl;

    VectorXd meanabserrs(NBTRIALS); meanabserrs.setZero();
    VectorXd lateral_input;
    VectorXd total_exc(NBNEUR), delta_total_exc(NBNEUR), delta_total_exc_sq(NBNEUR), total_exc_prev(NBNEUR);
    VectorXd delta_r(NBNEUR), delta_r_sq(NBNEUR), r_trace(NBNEUR), r_trace2(NBNEUR);
    r_trace.fill(0); r_trace2.fill(0);
    VectorXd delta_x(NBNEUR), delta_x_sq(NBNEUR), x_trace(NBNEUR), x_trace2(NBNEUR), delta_x_cu(NBNEUR);
//.........这里部分代码省略.........

	void set_seed(int s) {
		rng.seed(s);
	}

	RandomModule() {
		random_engine.seed(std::random_device{}());
	}

int main(int argc, char* argv[])
{

    fstream myfile;


    int PHASE=LEARNING;
    if (argc > 1)
       for (int nn=1; nn < argc; nn++)
       {
           if (strcmp(argv[nn], "test") == 0) { PHASE = TESTING; cout << "Test mode. " << endl; }
           if (strcmp(argv[nn], "METHOD") == 0) { METHOD = argv[nn+1]; }
           if (strcmp(argv[nn], "SUBW") == 0) { SUBW = atoi(argv[nn+1]); }
           if (strcmp(argv[nn], "G") == 0) { G = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "ALPHABIAS") == 0) { ALPHABIAS = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "ETA") == 0) { ETA = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "STIMVAL") == 0) { STIMVAL = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "ALPHAMODUL") == 0) { ALPHAMODUL = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "NBNEUR") == 0) { NBNEUR = atoi(argv[nn+1]); }
           if (strcmp(argv[nn], "PROBAMODUL") == 0) { PROBAMODUL = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "PROBAHEBB") == 0) { PROBAHEBB = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "ALPHATRACE") == 0) { ALPHATRACE = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "RNGSEED") == 0) { RNGSEED = atof(argv[nn+1]); }
           if (strcmp(argv[nn], "MAXDW") == 0) { MAXDW = atof(argv[nn+1]); }
       }

    string SUFFIX = "_G" + to_string(G) + "_MAXDW" + to_string(MAXDW) + "_ETA" + to_string(ETA) + "_ALPHAMODUL" + to_string(ALPHAMODUL) + "_PROBAMODUL" + to_string(PROBAMODUL) + "_SUBW" +to_string(SUBW) + "_ALPHATRACE" + to_string(ALPHATRACE) + "_METHOD-" + METHOD + "_ALPHABIAS" + to_string(ALPHABIAS) + "_PROBAHEBB" + to_string(PROBAHEBB) + "_NBNEUR" + to_string(NBNEUR) + "_RNGSEED" + to_string(RNGSEED);
    cout << SUFFIX << endl;

    myrng.seed(RNGSEED);

    double dt = 1.0;
    double tau = 10.0;
    int trialtype;

    int NBTRIALS = 100017; 
    int TRIALTIME = 700 ; //500; // 1100
    int STARTSTIM1 = 1, TIMESTIM1 = 500; // 200
    //int STARTSTIM2 = 400, TIMESTIM2 = 0; 

    if (PHASE == TESTING) 
        NBTRIALS = 40*NBPATTERNS;


    MatrixXd patterns[NBPATTERNS];
    MatrixXd tgtresps[NBPATTERNS];


    // Remember that input channel 0 is reserved for the unimplemented 'go' signal
    

    MatrixXd dJ(NBOUT, NBNEUR); dJ.setZero();
    MatrixXd win(NBNEUR, NBIN); randMat(win); //win.setRandom();// win.row(0).setZero(); // Uniformly between -1 and 1, except possibly for output cell (not even necessary).
    cout << win.col(0).head(5) << endl;
    MatrixXd J(NBNEUR, NBNEUR);
    
    randJ(J);
    
    if (PHASE == TESTING){
        readWeights(J, "J.dat");
        readWeights(win, "win.dat");
    }



    VectorXd meanerrs(NBTRIALS); meanerrs.setZero();
    VectorXd lateral_input;
    VectorXd dxthistrial(NBNEUR);
    MatrixXd rs(NBNEUR, TRIALTIME); rs.setZero();
    MatrixXd hebb(NBNEUR, NBNEUR);  
    VectorXd x(NBNEUR), r(NBNEUR), rprev(NBNEUR), dxdt(NBNEUR), k(NBNEUR), 
             input(NBIN), deltax(NBNEUR); 
    x.fill(0); r.fill(0);
    
    VectorXd err(TRIALTIME); 
    VectorXd meanerrtrace(NBPATTERNS);
    double meanerr;

    MatrixXd dJtmp, Jprev, Jr;

    double dtdivtau = dt / tau;



    meanerrtrace.setZero();



    for (int numtrial=0; numtrial < NBTRIALS; numtrial++)
    {

        if (PHASE == LEARNING)
            //trialtype = (int)(numtrial/2) % NBPATTERNS;
            trialtype = numtrial % NBPATTERNS;
        else 
            trialtype = numtrial % NBPATTERNS;

        
        hebb.setZero();
        dJ.setZero();
//.........这里部分代码省略.........