[arcticblue]

I think the following bit of code has a bug. You are updating w and w0 every time through the loop and then comparing the two w's at the end. I believe you should be comparing the value of w from before the loop to the value of w at the end of the loop.

So the first time you have w=[0,0,0] then you loop through all points updating the weight. Then after that you compare w=[0,0,0] with weight you have at the end of the loop. Then you use the updated weight as the initial weight for the next time through the loop.

for (j in permuted.ind) {
  w0 = w;
  v = (-YIn[j] * XIn[j,]) / (1 + exp(YIn[j] * (XIn[j,] %*% w0)));
  w = w0 - eta * v;
}
# Stop gradient descent as soon as delta(w, w0) < 0.01
if (sqrt(sum((w - w0)^2)) < 0.01) break()