Quantifying the maximum possible improvement in 2k experiments Nandan Sudarsanam1 · Anusha Kumar2 · Daniel D. Frey3