Stability of the Mazo False Minima in Various Blind Equalization AlgorithmsCornell University, 1992 - 230 pages |
Contents
Prediction of Local Stability | 4 |
B Calculation of the GODARD2 Algorithm Hessian | 64 |
Computer Listings | 69 |
Copyright | |
1 other sections not shown
Common terms and phrases
adaptive gain alpha Bennett Documented Bennett's MS Thesis Bernoulli distributed blind equalization algorithm channel equalization channel-equalizer space Condition converge Convex Combination correct channel inverse cost function December 16 eigenvalues elseif selection elseif sum_param_err2(numpts end end Equation 3.1 equilibria error function associated error^2 evaluates the error expectation operator fails to find find the correct fprintf function evaluates GODARD-2 algorithm GODARD2ALG gradient descent algorithm halt Hessian matrix hold initially started input signal KINK algorithm local minima Mazo false minima meta minimum numpts Observation 4.1.1 param parameter trajectories Pauline plot n plot yxline,xxline plot(xxline,yxline poor behavior positive definite predict routine SATO algorithm Section sign(zk simulation spemin SSPE stability analysis stochastic gradient descent Stop-and-Go Algorithm sum_param_err3 sum_param_err3(k)=sum(param_err_sq3 tanh(m theta theta,numpts title('Projection trivial channel update term various blind equalization vector Written xk=ak xlabel('F xlabel(’B xxline y(ak ylabel(’Theta2 zero zk Figure Zk values µƐ