Use SLSQP by default for sequential optimization (#124)

Summary:
Change is necessary b/c of pytorch/botorch#207
Pull Request resolved: #124

Reviewed By: bletham

Differential Revision: D16210786

Pulled By: Balandat

fbshipit-source-id: 4493c969c7d334fe741ce9611f64f01503bca8f8

ax/models/torch/botorch_defaults.py

@@ -198,6 +198,9 @@ def scipy_optimizer(
         optimize = joint_optimize
     else:
         optimize = sequential_optimize
+        # use SLSQP by default for small problems since it yields faster wall times
+        if "method" not in kwargs:
+            kwargs["method"] = "SLSQP"
     return optimize(
         acq_function=acq_function,
         bounds=bounds,