We study the problem of high-dimensional regression when there may be interacting variables. Approaches using sparsity-inducing penalty functions such as the Lasso (Tibshirani, 1996) can be useful for producing interpretable ...