Infinities
A replacement for the transfer function normprod.m in the MATLAB NN toolbox. With this transfer function, the NaNs in the simulation of RBF neural networks can be avoided.The function can handly zeros, and positive and negative infinite values in the weight vector. Also combinations of positive...
Platforms: Java
License: Freeware | Size: 10 KB | Download (47): nprod.m version Download |
log(cos(Z)) and log(sin(Z)) return infinities for abs(imag(Z)) > 711 in double precision, or abs(imag(Z)) > 90 in single precision. The overflow occurs in the trigonometric function; the correct final result has about the same magnitude as Z, and can be readily computed.Using logcos(Z) or...
Platforms: Matlab
License: Shareware | Cost: $0.00 USD | Size: 10 KB | Download (48): Logarithm of complex sine and cosine avoiding overflow Download |
Randomly pick n from size(h)>=nelements, biased with linear weights asgiven in h, without replacement.Works with infinity and zero weightingentries, but always picks themsequentially in this case.Syntax: Y=randweightedpick(h,n)Example:randweightedpick([2,0,5,Inf,3,Inf,0],7)returns [4,6,...,2,7]...
Platforms: Matlab
License: Shareware | Cost: $0.00 USD | Size: 10 KB | Download (43): Random Weighted Selection Download |
bigfloat is a Python library that provides arbitrary-precision correctly-rounded floating-point arithmetic. It is implemented as a wrapper around the MPFR library. Here are some key features of "bigfloat": ?*A* correct rounding on all operations; precisely defined semantics compatible with the...
Platforms: *nix
License: Freeware | Size: 102.4 KB | Download (32): bigfloat Download |