Yes, you can make a wire winding that creates a field approximating a halbach array. Boeing has a patent on this-- US Patent 7,598,646. Unfortunately, it is only an interesting academic exercise. It turns out that this "halbach winding" will actually be less efficient than a conventional winding.
It would take a bit to explain in detail, but the basic idea is that you can approximate a magnet by a sheet of current running around the outside of the magnet i.e. the field of a cylindrical magnet looks like the field of a cylindrical solenoid coil. Thus you could make a "halbach array" winding by putting correctly oriented strands of windings down the "edges" of all the magnets in the halbach array.
What this fails to appreciate is that while a halbach array is the optimum (in some sense of the word) use of permanent magnets, wires for windings can be arranged in ways that magnets can not. The "back" sides of the "halbach winding" can be removed-- these "back sides" actually generate a small torque in the wrong direction (look at the direction of the current and the direction of the field from the PM relative to the "front side" of the winding), so removing them will only increase the torque of the motor and reduce winding losses.
Obviously, this can not be done with a magnet, since you can't have only "one side" of a magnet, but with a winding it is easy becuase you make a longer end turn and return the current in a different "pole" of the motor. This is where the analogy of magnet->winding breaks down; you can do thing with windings that you can't do with permanent magnets, and thus the "optimum" PM arrangment is not the "optimum" winding arrangment.