LogitNormal code review

[denadai2]

Hello everybody,
I would like to write the LogitNormal function (suggested from here #2066) and add it to pymc3 as well, but I have some problems with ADVI which finds NaN. This could be caused by my model, or by this code where I need some help. Sorry to bother
ref: https://en.wikipedia.org/wiki/Logit-normal_distribution

class Logitnormal(PositiveContinuous):

    def __init__(self, mu=0.5, tau=None, *args, **kwargs):
        super(Logitnormal, self).__init__(*args, **kwargs)

        self.mu = mu = tt.as_tensor_variable(mu)
        self.tau = tau = tt.as_tensor_variable(tau)
        
        self.median = logit(mu)
        assert_negative_support(tau, 'tau', 'Logitnormal')

    def logp(self, value):
        mu = self.mu
        tau = self.tau
        return bound(-0.5 * tau * (logit(value) - mu)**2
                     + 0.5 * tt.log(tau / (2. * np.pi))
                     - tt.log(value*(1-value)),value > 0, value <1, tau > 0)

[denadai2]

closed in favour to #2100