No one answered my question.
How does this thing achieve yaw?
Differential airbrakes.
Deploy them on one side only, you yaw to that side.
(SWAG on my part--it's the only way I can see to do it)
Wow. Can you imagine the computers in that thing. No wonder it costs over a billion dollars.
At least it has ailerons (sort of)!
Aircraft designers learned many years ago how to get around conventional control surfaces. Look at the delta wing aircraft with no ailerons!
I'd guess R. Mackey knows more about this design, but I'll offer a few comments until he chimes in with specifics.
Like rwguinn, I suspect the B-2 has spoilers on top of the wing to assist in roll control. The elevons you can see on the wing extend symmetrically for pitch control and asymmetrically for roll control. The Flight Control System, which is computer controlled based (partially) on pilot input is miraculous.
The computers are not as sophisticated as one would think, but the firmware/software is quite sophisticated.

You guys overestimate me... I haven't done any work on the B-2 Spirit, though I've seen them flying often.
Your suspicions are all correct. Yaw control is achieved via split rudders -- the ailerons have independent upper and lower surfaces. The aircraft can thereby induce drag at the wingtips, and this is how it yaws and maintains yaw stability.
The split rudder is really its own brand of control surface. In some respects it resembles a simple aileron, but in others its more like a spoiler in that it deliberately induces a controlled flow separation.
Yaw stability is also the biggest challenge. Compare the B-2 to earlier
flying wings -- these development aircraft relied on vertical surfaces, despite the dreams of Jack Northrop for a truly clean aircraft. In the early versions, the engines and their housings acted as vertical stabilizers, required to get cooling air over the piston engines. When the later versions mounted jets, actual vertical stabilizers were added, and even this wasn't always enough.
A third possibility, sometimes used in the development vehicles, was differential thrust. One tries to avoid that in a large aircraft, however, since it leads to inefficiency and isn't as responsive as control surfaces. Nonetheless, the generally more massive flying wing design lends itself to mounting the engines wide, since there's relatively little penalty in terms of increased yaw moment.
In the case of the B-2, however, Northrop's "clean" ideal was a design requirement, not merely an aesthetic. Engines above and below the planform, large vertical stabilizers, etc. are anathema to low observability. This led to much higher demands on the split rudders -- even the earliest development aircraft had them, but the old design was inherently more stable, and a pilot could use them without electronic assistance. The new aircraft, on the other hand, needs to adjust them almost constantly to keep itself well inside its stability box. So high rate flight computers send minute corrections to the split rudders constantly, enforcing a straight track. Fortunately, the aircraft is rather large and massive, so a small excursion doesn't instantly lead to a roll and spin. Its size and inertia mean there is time to return before getting into an uncontrollable spin or stall condition.
The flight computers are not particularly complicated, but the air data sensors are. An uncommanded yaw needs to be sensed with more accuracy and at a higher rate. This places emphasis on inertial reference and actual air probes; the latter also have to be "stealthy" as they protrude outside the aicraft, and this is one of the more complex challenges of the entire vehicle.
These sensors are expensive. So are the advanced materials, and the low observability features are highly maintenance intensive. Other than that, however, the basic principles are fairly straightforward. Northrop's original dream was a good one, simply ahead of available technology. For its intended mission and requirements, the B-2 is a jolly good aircraft.