Prob

NAME

Prob - Probabilities for SRILM

SYNOPSIS

 #include <Prob.h>

DESCRIPTION

Prob is a collection of types, constants and utility functions for handling probabilities in the SRILM library.

TYPES

Prob: A floating point number representing a probability.
LogP: Logarithm to base 10 of a probability.

CONSTANTS

LogP_Zero: Log of probability 0.
LogP_Inf: Log of probability infinity (not a legal probability, of course).
LogP_One: Log of probability 1.
LogP_Precision: The number of significant digits in a LogP
Prob_Epsilon: A positive value close to 0; probability sums less than this should be considered effectively zero.

FUNCTIONS

Boolean parseLogP(const char *string, LogP &prob): Converts a floating point string representation into a LogP. Returns true iff the number was parsed correctly. This function should be much faster than generic C library functions for floating point parsing. Also, it parses singular LogP's (plus/minus infinity) correctly.
Prob LogPtoPPL(LogP prob): Converts a LogP into a perplexity (PPL).
ProbToLogP(Prob \fPprob\fP): Converts a probability into a LogP.
LogP MixLogP(LogP prob1, LogP prob2, double lambda): Computes the LogP resulting from interpolating two LogP's. If p1 and p2 are probabilities corresponding to prob1 and prob2, respectively, then the result is the LogP corresponding to lambda * p1 + (1 - lambda) * p2.

The following functions deal with bytelogs. Bytelogs are logarithms scaled to represent probabilties and likelihoods as a short integer in SRI's DECIPHER(TM) recognizer (bytelog(p) = log(p) * 10000.5 / 1024).

double ProbToBytelog(Prob prob): Converts a probability to a bytelog.
double LogPtoBytelog(LogP prob): Convert a LogP to a bytelog.
LogP BytelogToLogP(double bytelog): Convert a bytelog to a LogP.