#define protected public

#include <iostream>
#include <sys/types.h>
#include <signal.h>
#include <string>
#include <sstream>
#include <cmath>

#include "gtest/gtest.h"
#include "trick/ExecutiveException.hh"
#include "trick/Executive.hh"
#include "trick/MonteCarlo.hh"
#include "trick/MonteVar.hh"
#include "trick/MonteVarCalculated.hh"
#include "trick/MonteVarFile.hh"
#include "trick/MonteVarFixed.hh"
#include "trick/MonteVarRandom.hh"
#include "trick/montecarlo_c_intf.h"
#include "trick/exec_proto.h"
#include "trick/exec_proto.hh"
#include "trick/SimObject.hh"
#include "trick/MemoryManager.hh"
#include "trick/memorymanager_c_intf.h"
#include "trick/rand_generator.h"
//#include "trick/RequirementScribe.hh"

void sig_hand(int sig) ;
void child_handler(int sig) ;
#if (__APPLE__ | __CYGWIN__ | __INTERIX )
void fpe_sig_handler(int sig) ;
#else
void fpe_sig_handler(int sig, siginfo_t * sip, void *uap) ;
#endif

namespace Trick {

class testSimObject : public Trick::SimObject {
    public:

        int default_data_1() {
            return 0 ;
        }

        int initialization_1() {
            return 0 ;
        }

        int return_error() {
            return -1 ;
        }

        int throw_exception() {
            return exec_terminate("throw_exception", "exec_terminate called") ;
        }

        testSimObject() {
            int ii = 0 ;
            add_job(0, ii++, "monte_master_init", NULL, 1, "master_init", "TRK") ;
            add_job(0, ii++, "monte_master_pre", NULL, 1, "master_pre", "TRK") ;
            add_job(0, ii++, "monte_master_post", NULL, 1, "master_post", "TRK") ;
            add_job(0, ii++, "monte_master_shutdown", NULL, 1, "master_shutdown", "TRK") ;
            add_job(0, ii++, "monte_slave_init", NULL, 1, "slave_init", "TRK") ;
            add_job(0, ii++, "monte_slave_pre", NULL, 1, "slave_pre", "TRK") ;
            add_job(0, ii++, "monte_slave_post", NULL, 1, "slave_post", "TRK") ;
            add_job(0, ii++, "monte_slave_shutdown", NULL, 1, "slave_shutdown", "TRK") ;
        }

        virtual int call_function( Trick::JobData * curr_job ) ;
        virtual double call_function_double( Trick::JobData * curr_job ) { (void)curr_job ; return 0.0 ; } ;
} ;

int testSimObject::call_function( __attribute__((unused)) Trick::JobData * curr_job ) {
    int trick_ret = 0 ;

    return trick_ret ;
} ;

class MonteCarloTest : public ::testing::Test {

    protected:
        Trick::MonteCarlo exec;
        Trick::Executive executive;
        testSimObject so1 ;
        //Trick::RequirementScribe req;
        Trick::MemoryManager mm ;

        MonteCarloTest() { exec_register_scheduler(&exec) ; }
        ~MonteCarloTest() {  }
        virtual void SetUp() {}
        virtual void TearDown() {}

        int get_class_map_value(std::string job_class) { return exec.class_map[job_class] ; }
        Trick::ScheduledJobQueue * get_master_init_queue() { return &exec.master_init_queue ; }
        Trick::ScheduledJobQueue * get_master_pre_queue() { return &exec.master_pre_queue ; }
        Trick::ScheduledJobQueue * get_master_post_queue() { return &exec.master_post_queue ; }
        Trick::ScheduledJobQueue * get_master_shutdown_queue() { return &exec.master_shutdown_queue ; }
        Trick::ScheduledJobQueue * get_slave_init_queue() { return &exec.slave_init_queue ; }
        Trick::ScheduledJobQueue * get_slave_pre_queue() { return &exec.slave_pre_queue ; }
        Trick::ScheduledJobQueue * get_slave_post_queue() { return &exec.slave_post_queue ; }
        Trick::ScheduledJobQueue * get_slave_shutdown_queue() { return &exec.slave_shutdown_queue ; }
} ;

TEST_F(MonteCarloTest , Job_Queue_Master_Init) {
    //req.add_requirement("722132127");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_master_init_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.master_init") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Master_Pre) {
    //req.add_requirement("587551115");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_master_pre_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.master_pre") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Master_Post) {
    //req.add_requirement("4165308678");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_master_post_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.master_post") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Master_Shutdown) {
    //req.add_requirement("3461634900");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_master_shutdown_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.master_shutdown") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Slave_Init) {
    //req.add_requirement("1412318284");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_slave_init_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.slave_init") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Slave_Pre) {
    //req.add_requirement("3301658297");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_slave_pre_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.slave_pre") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Slave_Post) {
    //req.add_requirement("3882184138");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_slave_post_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.slave_post") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Job_Queue_Slave_Shutdown) {
    //req.add_requirement("350185460");

    Trick::JobData * curr_job ;

    EXPECT_EQ(exec_get_time_tic_value() , 1000000) ;

    exec_add_sim_object(&so1 , "so1") ;

    curr_job = get_slave_shutdown_queue()->get_next_job() ;
    ASSERT_FALSE( curr_job == NULL ) ;
    EXPECT_STREQ( curr_job->name.c_str() , "so1.slave_shutdown") ;

    EXPECT_TRUE(executive.check_all_jobs_handled() == 0 ) ;
}

TEST_F(MonteCarloTest , Good_Initialization) {
    //req.add_requirement("1452306647");

    exec_add_sim_object(&so1 , "so1") ;
    EXPECT_EQ(exec.execute_monte() , 0 ) ;
}

TEST_F(MonteCarloTest , TestDefaultValues) {

    EXPECT_EQ(exec.get_enabled(), false) ;
    EXPECT_EQ(exec.get_dry_run(), false) ;
    EXPECT_EQ(exec.get_localhost_as_remote(), false) ;
    EXPECT_EQ(exec.get_custom_slave_dispatch(), false) ;
    EXPECT_EQ(exec.get_timeout(), 120) ;
    EXPECT_EQ(exec.get_max_tries(), 2) ;
    EXPECT_EQ(exec.get_verbosity(), exec.MC_INFORMATIONAL) ;
    EXPECT_EQ(exec.get_num_runs(), 0) ;
    EXPECT_EQ(exec.get_slave_id(), 0) ;
    EXPECT_EQ(exec.actual_num_runs, 0) ;
    EXPECT_EQ(exec.num_results, 0) ;

}

TEST_F(MonteCarloTest, TestSettingValues) {
    //req.add_requirement("2840823120");

    exec.set_enabled(true) ;
    EXPECT_EQ(exec.get_enabled(), true) ;
    exec.set_dry_run(true) ;
    EXPECT_EQ(exec.get_dry_run(), true) ;
    exec.set_localhost_as_remote(true) ;
    EXPECT_EQ(exec.get_localhost_as_remote(), true) ;
    exec.set_custom_slave_dispatch(true) ;
    EXPECT_EQ(exec.get_custom_slave_dispatch(), true) ;
    exec.set_timeout(60) ;
    EXPECT_EQ(exec.get_timeout(), 60) ;
    exec.set_max_tries(4) ;
    EXPECT_EQ(exec.get_max_tries(), 4) ;
    exec.set_verbosity(exec.MC_NONE) ;
    EXPECT_EQ(exec.get_verbosity(), exec.MC_NONE) ;
    exec.set_verbosity(exec.MC_ERROR) ;
    EXPECT_EQ(exec.get_verbosity(), exec.MC_ERROR) ;
    exec.set_verbosity(exec.MC_INFORMATIONAL) ;
    EXPECT_EQ(exec.get_verbosity(), exec.MC_INFORMATIONAL) ;
    exec.set_verbosity(exec.MC_ALL) ;
    EXPECT_EQ(exec.get_verbosity(), exec.MC_ALL) ;

}

TEST_F(MonteCarloTest, TestRanges) {
    exec.set_num_runs(25) ;
    EXPECT_EQ(exec.get_num_runs(), 25) ;
    EXPECT_EQ(exec.actual_num_runs, 25) ;
    exec.add_range(3, 5) ;
    EXPECT_EQ(exec.actual_num_runs, 3) ;
    exec.add_range(20) ;
    EXPECT_EQ(exec.actual_num_runs, 4) ;
    EXPECT_EQ(exec.in_range(exec.runs[3]) , true) ;
    EXPECT_EQ(exec.in_range(exec.runs[10]), false) ;
}

TEST_F(MonteCarloTest, TestSlaves) {
    //req.add_requirement("1098748189");

    Trick::MonteSlave slave0("localhost") ;
    Trick::MonteSlave slave1("WonderWoman") ;
    EXPECT_EQ(exec.slaves.empty(), true) ;
    EXPECT_EQ(exec.get_slave_index(1), -1) ;
    exec.add_slave(&slave0) ;
    EXPECT_EQ(exec.slaves.empty(), false) ;
    EXPECT_EQ(exec.slaves.size(), 1) ;
    exec.add_slave(&slave1) ;
    EXPECT_EQ(exec.slaves.size(), 2) ;
    EXPECT_EQ(slave0.id, 1) ;
    EXPECT_EQ(slave1.id, 2) ;
    EXPECT_EQ(exec.get_slave_index(1), 0) ;
    EXPECT_EQ(exec.get_slave_index(2), 1) ;
    EXPECT_EQ(slave0.state, Trick::MonteSlave::MC_UNINITIALIZED) ;
    EXPECT_EQ(slave1.state, Trick::MonteSlave::MC_UNINITIALIZED) ;
    exec.disable_slave("WonderWoman", true) ;
    EXPECT_EQ(slave1.state, Trick::MonteSlave::MC_STOPPED) ;
    EXPECT_EQ(exec.slaves.size(), 2) ;
    exec.disable_slave("WonderWoman", false) ;
    EXPECT_EQ(slave1.state, Trick::MonteSlave::MC_UNINITIALIZED) ;
    EXPECT_EQ(slave0.machine_name, "localhost") ;
    EXPECT_EQ(slave1.machine_name, "WonderWoman") ;
}

TEST_F(MonteCarloTest, MonteVarFile) {
    //req.add_requirement("3932595803");

    // Test MonteVarFile
    Trick::MonteVarFile var0("time_to_fire_1", "M_jet_firings_inline", 2) ;
    EXPECT_EQ(exec.variables.size(), 0) ;
    exec.add_variable(&var0) ;
    EXPECT_EQ(var0.get_next_value(), "time_to_fire_1 = 1.0000") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
}

TEST_F(MonteCarloTest, MonteVarFileWithEOFLine) {
    Trick::MonteVarFile var0("time_to_fire_1", "MonteCarlo_eof_line", 2) ;
    EXPECT_EQ(exec.variables.size(), 0) ;
    exec.add_variable(&var0) ;
    EXPECT_EQ(var0.get_next_value(), "time_to_fire_1 = 1.0000") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
    EXPECT_EQ(var0.get_next_value(), "EOF") ;
}

TEST_F(MonteCarloTest, MonteVarFileWithLineEndedEOF) {
    Trick::MonteVarFile var0("time_to_fire_1", "MonteCarlo_line_ended_eof", 2) ;
    EXPECT_EQ(exec.variables.size(), 0) ;
    exec.add_variable(&var0) ;
    EXPECT_EQ(var0.get_next_value(), "time_to_fire_1 = 1.0000") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
    EXPECT_EQ(var0.get_next_value(), "EOF") ;
}

TEST_F(MonteCarloTest, MonteVarRandom_Gaussian) {
    std::string str;
    double value;
    //req.add_requirement("3932595803");

    // Test MonteVarRandom
    // Gaussian
    Trick::MonteVarRandom var1("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    var1.set_mu(4.0) ;
    var1.set_min(-4.0) ;
    var1.set_max(4.0) ;
    var1.set_seed(12345);
    var1.set_sigma_range(0) ;
    exec.add_variable(&var1) ;
    var1.get_next_value();
    str = var1.value;

    std::stringstream sstream(str);
    sstream >> value;

    EXPECT_LE(value, 8);
    EXPECT_GE(value, 0);

    EXPECT_EQ(var1.name, "time_to_fire_1") ;
    EXPECT_EQ(var1.distribution, 0) ;
    EXPECT_EQ(var1.randist.mu, 4.0) ;
    EXPECT_EQ(var1.randist.min, -4.0) ;
    EXPECT_EQ(var1.randist.max, 4.0) ;
    EXPECT_EQ(var1.unit, "") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
}

TEST_F(MonteCarloTest, MonteVarRandom_Poisson) {
    std::string str;
    double value;
    //req.add_requirement("3932595803");

    // Test MonteVarRandom
    // Poisson
    Trick::MonteVarRandom var2("time_to_fire_1", Trick::MonteVarRandom::POISSON, "s") ;
    var2.set_mu(4.0) ;
    var2.set_min(-4.0) ;
    var2.set_max(4.0) ;
    var2.set_min_is_relative(0);
    var2.set_max_is_relative(0);
    var2.set_seed(12345);
    var2.set_sigma_range(0) ;
    exec.add_variable(&var2) ;
    var2.get_next_value();
    str = var2.value;

    std::stringstream sstream(str);
    sstream >> value;

#if _HAVE_GSL
    EXPECT_GE(value, 0);
    EXPECT_LE(value, 4);
#else
    EXPECT_EQ(value, 0);
#endif

    EXPECT_EQ(var2.name, "time_to_fire_1") ;
    EXPECT_EQ(var2.distribution, 2) ;
    EXPECT_EQ(var2.unit, "s") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
}

TEST_F(MonteCarloTest, MonteVarRandom_Flat) {
    std::string str;
    double value;
    //req.add_requirement("3932595803");

    // Test MonteVarRandom
    // Flat
    Trick::MonteVarRandom var3("time_to_fire_4", Trick::MonteVarRandom::FLAT) ;
    var3.set_mu(4.0) ;
    var3.set_min(-4.0) ;
    var3.set_max(4.0) ;
    var3.set_seed(12345);
    var3.set_sigma_range(0) ;
    exec.add_variable(&var3) ;
    var3.get_next_value();
    str = var3.value;

    std::stringstream sstream(str);
    sstream >> value;

    EXPECT_GE(value, 0);
    EXPECT_LE(value, 8);

    EXPECT_EQ(var3.name, "time_to_fire_4") ;
    EXPECT_EQ(var3.distribution, 1) ;
    EXPECT_EQ(var3.unit, "") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
}

TEST_F(MonteCarloTest, MonteVarRandom_NonGSL) {
    //req.add_requirement("3932595803");

    Trick::MonteVarRandom var4("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    var4.set_mu(4.0) ;
    var4.set_min(-4.0) ;
    var4.set_max(4.0) ;
    var4.set_seed(12345);
    // for this test, let sigma_range default to 1   var4.set_sigma_range(0) ;

    exec.add_variable(&var4) ;
    // Test MonteVarRandom
    // non GSL library variables
    EXPECT_EQ(var4.randist.uniform, LCG1) ;
    var4.set_uniform_generator(TRIPLE) ;
    EXPECT_EQ(var4.randist.uniform, 1) ;
    var4.set_uniform_generator(LCG1) ;
    EXPECT_EQ(var4.randist.uniform, LCG1) ;
    EXPECT_EQ(var4.randist.sigma_range, 1) ;
    var4.set_sigma_range(8) ;
    EXPECT_EQ(var4.randist.sigma_range, 8) ;
}

TEST_F(MonteCarloTest, MonteVarFixed) {
    //req.add_requirement("3932595803");

    // Test MonteVarFixed
    Trick::MonteVarFixed var5("time_to_fire_5", 3.1415) ;
    exec.add_variable(&var5) ;
    EXPECT_EQ(var5.get_next_value(), "time_to_fire_5 = 3.1415") ;
    EXPECT_EQ(var5.name, "time_to_fire_5") ;
    EXPECT_EQ(var5.unit, "") ;
    EXPECT_EQ(exec.variables.size(), 1) ;
}

///@brief check that the final distribution is correct. (Bug 6950: The non-GSL was wrong for GAUSSIAN and FLAT)
TEST_F(MonteCarloTest, MonteVarRandom_Gaussian_distributionMatchesSpec) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var6("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var6) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var6.set_seed(12345) ;
    var6.set_mu(expected_mean) ;
    var6.set_sigma(expected_sigma) ;
    var6.set_min(-10.0) ;
    var6.set_max(10.0) ;
    var6.set_sigma_range(0) ;

    // maintain running sums for calculating mean and variance
    double mean = 0.0 ;
    double sum_variance = 0.0 ;
    int    n = 0 ;
    double delta = 0.0 ;
    int    numRuns = 100 ;

    for (int ii = 0; ii <= numRuns; ++ii) {

        {
            var6.get_next_value() ;
            str = var6.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }

        n++ ;
        delta = value - mean ;
        mean += delta / (double)n ;
        sum_variance += delta * (value - mean) ;
    }

    // final statistics
    double variance = sum_variance / (double) (n - 1) ;
    double sigma = sqrt(variance) ;


    EXPECT_NEAR(expected_mean,  mean,  0.05 * expected_mean) ;
    EXPECT_NEAR(expected_sigma, sigma, 0.10 * expected_sigma) ;
}


#ifndef _HAVE_GSL

///@brief set_sigma_range(int) feature was only implemented for nonGsl with non-C++11 fallbacks
TEST_F(MonteCarloTest, MonteVarRandom_Gaussian_nonGslSigmaRangeDefaulted_maxDeviationFromMeanIs1Sigma) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var7("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var7) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var7.set_seed(12345) ;
    var7.set_mu(expected_mean) ;
    var7.set_sigma(expected_sigma) ;
    var7.set_min(-10.0) ;
    var7.set_max(10.0) ;

    // without this call, default sigma_range is 1
    // var7.set_sigma_range(0) ;

    int    num_runs = 100 ;
    double sigmas_max_deviation_from_mean = 0.0 ;
    double deviation_sigmas = 0.0 ;

    for (int ii = 0; ii <= num_runs; ++ii) {

        {
            var7.get_next_value() ;
            str = var7.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }

        deviation_sigmas = fabs( (value - expected_mean) / expected_sigma ) ;

        if (deviation_sigmas > sigmas_max_deviation_from_mean) {
            sigmas_max_deviation_from_mean = deviation_sigmas ;
        }
    }

    // Loose constraint: anything .LE. 1 will pass.
    // (Doesn't require the largest deviations to be around 1.
    // In other words, even max deviations of zero would pass.)
    //EXPECT_LE(sigmas_max_deviation_from_mean, 1.0) ;

    // tighter constraint: the max deviation with 100 runs
    // should be pretty close to but still <= 1.0.
    // This specifies the range 0.9 .. 1.0
    EXPECT_NEAR(sigmas_max_deviation_from_mean, 0.95, 0.05) ;
}

///@brief test set_sigma_range feature works for STL random
TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_nonGslSigmaRangeDefaulted_maxDeviationFromMeanIs1Sigma) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var9("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN, "", Trick::MonteVarRandom::TRICK_DEFAULT_ENGINE) ;
    exec.add_variable(&var9) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var9.set_seed(12345) ;
    var9.set_mu(expected_mean) ;
    var9.set_sigma(expected_sigma) ;
    var9.set_min(-10.0) ;
    var9.set_max(10.0) ;

    // without this call, default sigma_range is 1
    // var9.set_sigma_range(0) ;

    int    num_runs = 1000 ;
    double sigmas_max_deviation_from_mean = 0.0 ;
    double deviation_sigmas = 0.0 ;

    for (int ii = 0; ii <= num_runs; ++ii) {

        {
            var9.get_next_value() ;
            str = var9.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }

        deviation_sigmas = fabs( (value - expected_mean) / expected_sigma ) ;

        if (deviation_sigmas > sigmas_max_deviation_from_mean) {
            sigmas_max_deviation_from_mean = deviation_sigmas ;
        }
    }

    // Loose constraint: anything .LE. 1 will pass.
    // (Doesn't require the largest deviations to be around 1.
    // In other words, even max deviations of zero would pass.)
    //EXPECT_LE(sigmas_max_deviation_from_mean, 1.0) ;

    // tighter constraint: the max deviation with 100 runs
    // should be pretty close to but still <= 1.0.
    // This specifies the range 0.9 .. 1.0
    EXPECT_NEAR(sigmas_max_deviation_from_mean, 0.95, 0.05) ;
}


#endif // not _HAVE_GSL

#if (defined(_HAVE_TR1_RANDOM) || defined(_HAVE_STL_RANDOM))

TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_distributionMatchesSpec) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var8("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN, "--",
                                Trick::MonteVarRandom::TRICK_DEFAULT_ENGINE) ;
    exec.add_variable(&var8) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var8.set_seed(12345) ;
    var8.set_sigma_range(0) ;
    var8.set_mu(expected_mean) ;
    var8.set_sigma(expected_sigma) ;
    var8.set_min(-10.0) ;
    var8.set_max(10.0) ;

    // maintain running sums for calculating mean and variance
    double mean = 0.0 ;
    double sum_variance = 0.0 ;
    int    n = 0 ;
    double delta = 0.0 ;
    int    numRuns = 100 ;

    for (int ii = 0; ii <= numRuns; ++ii) {

        {
            var8.get_next_value() ;
            str = var8.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }

        n++ ;
        delta = value - mean ;
        mean += delta / (double)n ;
        sum_variance += delta * (value - mean) ;
    }

    // final statistics
    double variance = sum_variance / (double) (n - 1) ;
    double sigma = sqrt(variance) ;

    EXPECT_NEAR(expected_mean,  mean,  0.05 * expected_mean) ;
    EXPECT_NEAR(expected_sigma, sigma, 0.10 * expected_sigma) ;
}


///@breif test relative min/max work as expected for STL
TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_relativeMinMaxWorks) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var6("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var6) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var6.set_seed(12345) ;
    var6.set_sigma_range(0) ;
    var6.set_mu(expected_mean) ;
    var6.set_sigma(expected_sigma) ;
    var6.set_min(-1.0) ;
    var6.set_max(1.0) ;
    var6.set_min_is_relative(true) ;
    var6.set_max_is_relative(true) ;

    // maintain running sums for calculating mean and variance
    int    numRuns = 100 ;
    double saw_min =  1000.0;
    double saw_max = -1000.0;

    for (int ii = 0; ii <= numRuns; ++ii) {
        {
            var6.get_next_value() ;
            str = var6.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }
        if (value < saw_min) {
            saw_min = value;
        }
        if (value > saw_max) {
            saw_max = value;
        }
    }

    // these conditions should be true if valid
    // relative min/max range of 10 +/- 1 is being imposed
    EXPECT_NEAR(9.05, saw_min, 0.05) ;
    EXPECT_NEAR(10.95, saw_max, 0.05) ;
}

///@breif test relative min/max works after calling set_mu, as expected for STL
TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_callSetMu_relativeMinMaxWorks) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var11("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var11) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var11.set_seed(12345) ;
    var11.set_sigma_range(0) ;
    var11.set_mu(expected_mean + 100.0) ;
    var11.set_sigma(expected_sigma) ;
    var11.set_min(-1.0) ;
    var11.set_max(1.0) ;
    var11.set_min_is_relative(true) ;
    var11.set_max_is_relative(true) ;
    var11.set_mu(expected_mean) ;

    // maintain running sums for calculating mean and variance
    int    numRuns = 100 ;
    double saw_min =  1000.0;
    double saw_max = -1000.0;

    for (int ii = 0; ii <= numRuns; ++ii) {
        {
            var11.get_next_value() ;
            str = var11.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }
        if (value < saw_min) {
            saw_min = value;
        }
        if (value > saw_max) {
            saw_max = value;
        }
    }

    // these conditions should be true if valid
    // relative min/max range of 10 +/- 1 is being imposed
    EXPECT_NEAR(9.05, saw_min, 0.05) ;
    EXPECT_NEAR(10.95, saw_max, 0.05) ;
}

///@breif test absolute min/max works after changing from relative, as expected for STL
TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_changeToAbsolute_MinMaxWorks) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var11("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var11) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var11.set_seed(12345) ;
    var11.set_sigma_range(0) ;
    var11.set_mu(expected_mean) ;
    var11.set_sigma(expected_sigma) ;
    var11.set_min(9.0) ;
    var11.set_max(11.0) ;

    var11.set_min_is_relative(false) ;
    var11.set_max_is_relative(false) ;

    // maintain running sums for calculating mean and variance
    int    numRuns = 100 ;
    double saw_min =  1000.0;
    double saw_max = -1000.0;

    for (int ii = 0; ii <= numRuns; ++ii) {
        {
            var11.get_next_value() ;
            str = var11.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }
        if (value < saw_min) {
            saw_min = value;
        }
        if (value > saw_max) {
            saw_max = value;
        }
    }

    // these conditions should be true if valid
    // relative min/max range of 10 +/- 1 is being imposed
    EXPECT_NEAR(9.05, saw_min, 0.05) ;
    EXPECT_NEAR(10.95, saw_max, 0.05) ;
}

///@brief test that the sequence repeats after calling set_seed
TEST_F(MonteCarloTest, MonteVarRandom_StlGaussian_setSeed_TakesEffect) {
    //req.add_requirement("3932595803");
    std::string str ;
    double      value ;

    Trick::MonteVarRandom var10("time_to_fire_1", Trick::MonteVarRandom::GAUSSIAN) ;
    exec.add_variable(&var10) ;

    double expected_mean  = 10.0 ;
    double expected_sigma = 2.0 ;

    var10.set_seed(12345) ;
    var10.set_sigma_range(0) ;
    var10.set_mu(expected_mean) ;
    var10.set_sigma(expected_sigma) ;
    var10.set_min(-10.0) ;
    var10.set_max(10.0) ;
    var10.set_min_is_relative(true) ;
    var10.set_max_is_relative(true) ;

    int numRuns = 10;

    for (int ii = 0; ii < numRuns; ++ii) {
        {
            var10.get_next_value() ;
            str = var10.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }
    }
    double first_value = value;

    var10.set_seed(12345) ;

    for (int ii = 0; ii < numRuns; ++ii) {
        {
            var10.get_next_value() ;
            str = var10.value ;
            std::stringstream sstream(str) ;
            sstream >> value ;
        }
    }
    double second_value = value;

    EXPECT_NEAR(first_value, second_value, 0.000001) ;
}

#endif // _HAVE_TR1_RANDOM or _HAVE_STL_RANDOM

}