Пример #1
0
attribute_hidden
double pbeta_raw(double x, double a, double b, int lower_tail, int log_p)
{
    // treat limit cases correctly here:
    if(a == 0 || b == 0 || !R_FINITE(a) || !R_FINITE(b)) {
	// NB:  0 < x < 1 :
	if(a == 0 && b == 0) // point mass 1/2 at each of {0,1} :
	    return (log_p ? -M_LN2 : 0.5);
	if (a == 0 || a/b == 0) // point mass 1 at 0 ==> P(X <= x) = 1, all x > 0
	    return R_DT_1;
	if (b == 0 || b/a == 0) // point mass 1 at 1 ==> P(X <= x) = 0, all x < 1
	    return R_DT_0;
	// else, remaining case:  a = b = Inf : point mass 1 at 1/2
	if (x < 0.5) return R_DT_0; else return R_DT_1;
    }
    // Now:  0 < a < Inf;  0 < b < Inf

    double x1 = 0.5 - x + 0.5, w, wc;
    int ierr;
    //====
    bratio(a, b, x, x1, &w, &wc, &ierr, log_p); /* -> ./toms708.c */
    //====
    // ierr in {10,14} <==> bgrat() error code ierr-10 in 1:4; for 1 and 4, warned *there*
    if(ierr && ierr != 11 && ierr != 14)
	MATHLIB_WARNING4(_("pbeta_raw(%g, a=%g, b=%g, ..) -> bratio() gave error code %d"),
			x, a,b, ierr);
    return lower_tail ? w : wc;
} /* pbeta_raw() */
Пример #2
0
double pnbinom_mu(double x, double size, double mu, int lower_tail, int log_p)
{
#ifdef IEEE_754
    if (ISNAN(x) || ISNAN(size) || ISNAN(mu))
	return x + size + mu;
    if(!R_FINITE(mu))	ML_ERR_return_NAN;
#endif
    if (size < 0 || mu < 0)	ML_ERR_return_NAN;

    /* limiting case: point mass at zero */
    if (size == 0)
        return (x >= 0) ? R_DT_1 : R_DT_0;

    if (x < 0) return R_DT_0;
    if (!R_FINITE(x)) return R_DT_1;
    if (!R_FINITE(size)) // limit case: Poisson
	return(ppois(x, mu, lower_tail, log_p));

    x = floor(x + 1e-7);
    /* return
     * pbeta(pr, size, x + 1, lower_tail, log_p);  pr = size/(size + mu), 1-pr = mu/(size+mu)
     *
     *= pbeta_raw(pr, size, x + 1, lower_tail, log_p)
     *            x.  pin   qin
     *=  bratio (pin,  qin, x., 1-x., &w, &wc, &ierr, log_p),  and return w or wc ..
     *=  bratio (size, x+1, pr, 1-pr, &w, &wc, &ierr, log_p) */
    {
	int ierr;
	double w, wc;
	bratio(size, x+1, size/(size+mu), mu/(size+mu), &w, &wc, &ierr, log_p);
	if(ierr)
	    MATHLIB_WARNING(_("pnbinom_mu() -> bratio() gave error code %d"), ierr);
	return lower_tail ? w : wc;
    }
}
Пример #3
0
LDOUBLE attribute_hidden
pnbeta_raw(double x, double o_x, double a, double b, double ncp)
{
    /* o_x  == 1 - x  but maybe more accurate */

    /* change errmax and itrmax if desired;
     * original (AS 226, R84) had  (errmax; itrmax) = (1e-6; 100) */
    const static double errmax = 1.0e-9;
    const int    itrmax = 10000;  /* 100 is not enough for pf(ncp=200)
				     see PR#11277 */

    double a0, lbeta, c, errbd, x0, temp, tmp_c;
    int j, ierr;

    LDOUBLE ans, ax, gx, q, sumq;

    if (ncp < 0. || a <= 0. || b <= 0.) ML_ERR_return_NAN;

    if(x < 0. || o_x > 1. || (x == 0. && o_x == 1.)) return 0.;
    if(x > 1. || o_x < 0. || (x == 1. && o_x == 0.)) return 1.;

    c = ncp / 2.;

	/* initialize the series */

    x0 = floor(fmax2(c - 7. * sqrt(c), 0.));
    a0 = a + x0;
    lbeta = lgammafn(a0) + lgammafn(b) - lgammafn(a0 + b);
    /* temp = pbeta_raw(x, a0, b, TRUE, FALSE), but using (x, o_x): */
    bratio(a0, b, x, o_x, &temp, &tmp_c, &ierr, FALSE);

    gx = exp(a0 * log(x) + b * (x < .5 ? log1p(-x) : log(o_x))
	     - lbeta - log(a0));
    if (a0 > a)
	q = exp(-c + x0 * log(c) - lgammafn(x0 + 1.));
    else
	q = exp(-c);

    sumq = 1. - q;
    ans = ax = q * temp;

	/* recurse over subsequent terms until convergence is achieved */
    j = (int) x0;
    do {
	j++;
	temp -= (double) gx;
	gx *= x * (a + b + j - 1.) / (a + j);
	q *= c / j;
	sumq -= q;
	ax = temp * q;
	ans += ax;
	errbd = (double)((temp - gx) * sumq);
    }
    while (errbd > errmax && j < itrmax + x0);

    if (errbd > errmax)
	ML_ERROR(ME_PRECISION, "pnbeta");
    if (j >= itrmax + x0)
	ML_ERROR(ME_NOCONV, "pnbeta");

    return ans;
}
Пример #4
0
void cumfnc(double *f,double *dfn,double *dfd,double *pnonc,
	    double *cum,double *ccum)
/*
**********************************************************************
 
               F -NON- -C-ENTRAL F DISTRIBUTION
 
 
 
                              Function
 
 
     COMPUTES NONCENTRAL F DISTRIBUTION WITH DFN AND DFD
     DEGREES OF FREEDOM AND NONCENTRALITY PARAMETER PNONC
 
 
                              Arguments
 
 
     X --> UPPER LIMIT OF INTEGRATION OF NONCENTRAL F IN EQUATION
 
     DFN --> DEGREES OF FREEDOM OF NUMERATOR
 
     DFD -->  DEGREES OF FREEDOM OF DENOMINATOR
 
     PNONC --> NONCENTRALITY PARAMETER.
 
     CUM <-- CUMULATIVE NONCENTRAL F DISTRIBUTION
 
     CCUM <-- COMPLIMENT OF CUMMULATIVE
 
 
                              Method
 
 
     USES FORMULA 26.6.20 OF REFERENCE FOR INFINITE SERIES.
     SERIES IS CALCULATED BACKWARD AND FORWARD FROM J = LAMBDA/2
     (THIS IS THE TERM WITH THE LARGEST POISSON WEIGHT) UNTIL
     THE CONVERGENCE CRITERION IS MET.
 
     FOR SPEED, THE INCOMPLETE BETA FUNCTIONS ARE EVALUATED
     BY FORMULA 26.5.16.
 
 
               REFERENCE
 
 
     HANDBOOD OF MATHEMATICAL FUNCTIONS
     EDITED BY MILTON ABRAMOWITZ AND IRENE A. STEGUN
     NATIONAL BUREAU OF STANDARDS APPLIED MATEMATICS SERIES - 55
     MARCH 1965
     P 947, EQUATIONS 26.6.17, 26.6.18
 
 
                              Note
 
 
     THE SUM CONTINUES UNTIL A SUCCEEDING TERM IS LESS THAN EPS
     TIMES THE SUM (OR THE SUM IS LESS THAN 1.0E-20).  EPS IS
     SET TO 1.0E-4 IN A DATA STATEMENT WHICH CAN BE CHANGED.
 
**********************************************************************
*/
{
#define qsmall(x) (int)(sum < 1.0e-20 || (x) < eps*sum)
#define half 0.5e0
#define done 1.0e0
static double eps = 1.0e-4;
static double dsum,dummy,prod,xx,yy,adn,aup,b,betdn,betup,centwt,dnterm,sum,
    upterm,xmult,xnonc;
static int i,icent,ierr;
static double T1,T2,T3,T4,T5,T6;
/*
     ..
     .. Executable Statements ..
*/
    if(!(*f <= 0.0e0)) goto S10;
    *cum = 0.0e0;
    *ccum = 1.0e0;
    return;
S10:
    if(!(*pnonc < 1.0e-10)) goto S20;
/*
     Handle case in which the non-centrality parameter is
     (essentially) zero.
*/
    cumf(f,dfn,dfd,cum,ccum);
    return;
S20:
    xnonc = *pnonc/2.0e0;
/*
     Calculate the central term of the poisson weighting factor.
*/
    icent = (long)(xnonc);
    if(icent == 0) icent = 1;
/*
     Compute central weight term
*/
    T1 = (double)(icent+1);
    centwt = exp(-xnonc+(double)icent*log(xnonc)-alngam(&T1));
/*
     Compute central incomplete beta term
     Assure that minimum of arg to beta and 1 - arg is computed
          accurately.
*/
    prod = *dfn**f;
    dsum = *dfd+prod;
    yy = *dfd/dsum;
    if(yy > half) {
        xx = prod/dsum;
        yy = done-xx;
    }
    else  xx = done-yy;
    T2 = *dfn*half+(double)icent;
    T3 = *dfd*half;
    bratio(&T2,&T3,&xx,&yy,&betdn,&dummy,&ierr);
    adn = *dfn/2.0e0+(double)icent;
    aup = adn;
    b = *dfd/2.0e0;
    betup = betdn;
    sum = centwt*betdn;
/*
     Now sum terms backward from icent until convergence or all done
*/
    xmult = centwt;
    i = icent;
    T4 = adn+b;
    T5 = adn+1.0e0;
    dnterm = exp(alngam(&T4)-alngam(&T5)-alngam(&b)+adn*log(xx)+b*log(yy));
S30:
    if(qsmall(xmult*betdn) || i <= 0) goto S40;
    xmult *= ((double)i/xnonc);
    i -= 1;
    adn -= 1.0;
    dnterm = (adn+1.0)/((adn+b)*xx)*dnterm;
    betdn += dnterm;
    sum += (xmult*betdn);
    goto S30;
S40:
    i = icent+1;
/*
     Now sum forwards until convergence
*/
    xmult = centwt;
    if(aup-1.0+b == 0) upterm = exp(-alngam(&aup)-alngam(&b)+(aup-1.0)*log(xx)+
      b*log(yy));
    else  {
        T6 = aup-1.0+b;
        upterm = exp(alngam(&T6)-alngam(&aup)-alngam(&b)+(aup-1.0)*log(xx)+b*
          log(yy));
    }
    goto S60;
S50:
    if(qsmall(xmult*betup)) goto S70;
S60:
    xmult *= (xnonc/(double)i);
    i += 1;
    aup += 1.0;
    upterm = (aup+b-2.0e0)*xx/(aup-1.0)*upterm;
    betup -= upterm;
    sum += (xmult*betup);
    goto S50;
S70:
    *cum = sum;
    *ccum = 0.5e0+(0.5e0-*cum);
    return;
#undef qsmall
#undef half
#undef done
}