LASSO solver is right up to scale

Author: jonathan-taylor

tests/test_QP.R

@@ -9,7 +9,8 @@ lam = 2
 
 soln1 = selectiveInference:::fit_randomized_lasso(X, Y, lam, 0, 0)$soln
 G = glmnet(X, Y, intercept=FALSE, standardize=FALSE)
-soln2 = coef(G, s=1/n, exact=TRUE, x=X, y=Y)[-1]
+soln2 = coef(G, s=lam/n, exact=TRUE, x=X, y=Y)[-1]
 
 print(soln1)
-print(soln2)
+print(soln2)
+plot(soln1, soln2)