Supplementary material for'Spike and slab variational Bayes for high dimensional logistic regression '

(Section 11). Lemma 2. Suppose the prior satisfies Lemma 3. Suppose the prior satisfies Lemma 4. Suppose the prior satisfies This is the most difficult technical step in establishing our result. Lemma 5. Consider the event We briefly explain the heuristic idea behind the proof of Lemma 5. Since the VB posterior We also work with the default parametrization of the bmlasso function of the BhGLM package, i.e. We provide five further test cases in addition to the experiment considered in Section 5. In all cases we consider Gaussian design matrices, but vary all other parameter. We ran each experiment 200 times and report the means and standard deviations of the performance metrics in Table 3.