結局のところ、解決策は本当に簡単でした。このアルゴリズムが言及されている多くの論文を見ると、両方の和のインデックスが「j/i = 1」ではなく「j/i = 2」から始まるので、指数関数はもはや0
l = [0 0.3078 0.27287 0 0 0 0.41933];
r = [0 0 0 0 0 0.4 0.6];
sigma = 9.8747;
mu0 = 2/sigma;
iterations = 50;
% Density evolution algorithm depiction for finding the treshold of irregular LDPC codes
syms x;
l_idle = zeros(1,length(l));
r_idle = zeros(1,length(r));
Q_1 = exp(-0.4527*x^0.86 + 0.0218);
Q_2 = sqrt(pi/x)*exp((-x/4)*(1-20/(7*x)));
mv = zeros(1,iterations+1);
for k=2:length(mv)
for i = 2:length(l_idle)
if ((mu0 + (i-1)*mv(k-1)) < 10)
l_idle(i) = double(subs(Q_1,x,(mu0 + (i-1)*mv(k-1))));
else
l_idle(i) = double(subs(Q_2,x,(mu0 + (i-1)*mv(k-1))));
end
end
lambda = l(2:length(l))*transpose(l_idle(2:length(l_idle)));
for j = 2:length(r_idle)
b = 1-(1-lambda)^(j-1);
if b < 10, r_idle(j) = subs(0.4527^(-1/0.86)*(0.0218-log(x))^(1/0.86),x,b);
else, r_idle(j) = subs(finverse(Q_2,x),x,b);
end
end
mv(k) = r(2:length(r))*transpose(r_idle(2:length(r_idle)));
end
ご支援いただきありがとうございます。素晴らしい週末をお過ごしいただけますか?