Optimal designs for testing hypothesis in multiarm clinical trials