Warning: f_tol is deprecated.

[olivierlabayle]

We are now getting a deprecation warning when using some of the linear models in this package, e.g. from the example:

n = 500
p = 5
X = randn(n, p)
y = 2 * (rand(n) .< 0.5) .- 1   # entries are +-1
λ = 0.5
logistic = LogisticRegression(λ)
theta = fit(logistic, X, y)

leads to:

Warning: f_tol is deprecated. Use f_abstol or f_reltol instead. The provided value (0.0001) will be used as f_reltol.

[tlienart]

Thanks for the flag, this must come from the call to the Optim package, should be a straightforward replacement, I can try to take a look this week