#include <CNCL/LREF.h>
Distribution Function F(x)
#include <CNCL/LREG.h>
Complementary Distribution Function G(x)
TYPE
CN_LREF
CN_LREG
CNStatistics
None
None
The CNLREF
and CNLREG
classes provide the statistical evaluation
of random sequences by the LRE algorithm. Results are the distribution
(d.f.) and the complementary distribution function (c.d.f.)
respectively, the local correlation coefficient, and the error of each
discovered point. The simulation run time is controlled by a predefined
maximum error regarding to the local correlation of the input values.
For further information refer to "Effective Control of Simulation Runs
by a New Algorithm for Correlated Random Sequences" by F. Schreiber,
AEUe Vol. 42, pp. 347-354, 1988.
Constructors:
CNLREF( CNParam * param );
CNLREF( double MIN=0.01, double MAX=0.99, double MAX_ERR=0.05, int LEVEL=100,
Scale SCALE=CNLRE::LIN, int MAXSORT= 0, const char* NAME = NIL, const char* TEXT = NIL);
CNLREF
evaluation.
MIN, MAX
MAX_ERR
LEVEL
SCALE
CNLRE::LIN
or CNLRE::LOG
)
MAXSORT
NAME
TEXT
CNLREG( CNParam * param )
CNLREG( double MIN=0.01, double MAX=0.99, double MAX_ERR=0.05, int LEVEL=100,
Scale SCALE=LIN, int MAXSORT= 0, const char* NAME = NIL, const char* TEXT = NIL)
CNLREG
evaluation. The parameters are the same as
described above.
In addition to the member functions required by CNCL and CNStatistics
,
CNLREF
and CNLREG
provide:
void set_base(double b);
void change_error(double ne);
long min_index() const;
long max_index() const;
get_result()
to acquire online evaluation during simulation.
const CNLRE::resultline *get_result(long lev);
lev
must be in the range
min_index
to max_index
. A result line is a structs with
the members x
, vf
-- holds F- or G-value ---, rho
,
sigrho
, d
-- the relative error -- and nx
-- the
number of exact hits of x
.
double cur_x_lev() const;
double cur_f_lev();
double cur_g_lev();
Go to the first, previous, next, last section, table of contents.