Much of the paper is devoted to details of data analysis, which I will spare you. But I think the gist is interesting, and not too hard to summarize.
First of all, J1023 is a millisecond pulsar, that is, a magnetized neutron star spinning 540 times per second. This rotating magnetic field produces, through an embarrassingly mysterious process, a beam of radio waves that sweeps past the Earth 540 times per second, and which we detected in a sky survey. This pulsar is spinning so quickly because matter from its companion, an ordinary-looking star, spiraled down onto its surface and spun it up. One reason the system is exciting is that while we see the neutron star spinning away busily now, we think that in 2001 people actually detected matter spiraling down towards its surface. People have thought that this is how radio pulsars get to be spinning so fast, but this is the closest we've come to actually seeing it happen. Unfortunately, while matter was actually spiralling down, nobody even knew there was a neutron star involved. So there aren't many observations of it in that phase. I'm hoping it does it again soon. But there are still lots of interesting questions about what J1023's doing right now.
First of all, if the companion was dumping material towards the pulsar in 2001, why did it stop? Why isn't the companion still transferring material? Or is it? One of the things we found in our first paper on the source was that it doesn't look like the transferred material actually fell on the surface of J1023. Where did it go, then?
Of course, this is not how one goes about writing a scientific paper. Maybe in an ideal world, but it's really tough to find out anything at all about a ten-kilometer ball of neutrons that's three thousand light-years away. So in practice a project usually starts with me coming up with some clever new way to observe J1023, then figuring out whatever I can from it. In this case, we managed to get some good X-ray observations, and it turns out that the above are the questions I can partially answer. But my answers come in two parts: what we see, and what I think is going on.
What we see is: J1023's a modest X-ray source. The emission looks like it's "non-thermal", i.e., not coming from dense material that's so hot it glows in X-rays (the way a hot stove glows in the visible). Instead it shows a very "hard" distribution of photons that looks more like something coming from a shock front or particle acceleration. There's maybe a hint of some thermal emission (lower dashed curves in the plot on the left), but if so it's something like 6% of the X-ray luminosity. When we look at the emission as a function of position in its orbit, we find that there's some variability: when the neutron star's on the far side, the X-rays are fainter, and when it's on the near side they're brighter (plot on the right); it also looks like maybe the emission's a bit "softer" when it's fainter. More interestingly, when we "fold" the X-ray photons at the pulsar period we find what looks like pulsations: that is, as the neutron star rotates, the X-rays increase and decrease (next plot, on the left, showing radio and X-ray pulse profiles).
So what do I think is going on?
Well, the most certain thing is that some of the emission is coming from the pulsar. After all, we see variability as the pulsar rotates. There are a couple of ways this can be generated. Since we know the pulsar emits in radio, we know there are great streams of particles flowing along field lines in its magnetosphere. In other, easier-to-understand millisecond pulsars, these stream down onto the pulsar surface and heat up the magnetic polar caps so much that we see thermal X-rays from them. These thermal X-rays are pulsed, as the neutron star rotates them into and out of view, but they're not very pulsed, since the neutron star's gravity is so strong it bends the light you can see much of the way around to the back. Since we see something like a 10% modulation of the X-rays, and the thermal component, if any, is only about 6%, it seems like there must be something else going on. But those streams of particles that produce the radio can also produce X-rays directly, and these have non-thermal spectra. This process also tends to produce sharper pulses. Unfortunately, we have just barely enough photons to detect pulsations at all, so we can't tell what their spectrum or shape is.
Bogdanov et al. 2005). So it looks like that's what's going on here too. Maybe.
There's another possibility that might account for some or all of the X-rays: the pulsar wind, and any material from the companion it carries with it, presumably go streaming out of the system. When they crash into the interstellar medium, you expect a shock. This too could produce the sort of non-thermal emission we see. Since it's well out of the system, you wouldn't expect any modulation, and in fact the majority of the emission is not modulated. What you might expect, though, is to see interesting shapes: it turns out that in many systems, the pulsar wind streams far enough out of the system that you can actually resolve the shape of the shock. What you usually see is an arc-like "bow shock" ahead of the pulsar as it zips supersonically through the interstellar medium, plus a "trail" where the outflowing material is channeled back along the pulsar's path. Unfortunately we don't see that in J1023, but we looked using XMM-Newton, which has kind of a high background and limited spatial resolution. We have more observations with the Chandra X-ray Observatory, which will hopefully resolve the issue when we get them analyzed.
So, what about those questions?
Well, I think the companion is still dumping material towards the pulsar, but right now the pulsar wind is so strong that it sweeps the material out of the system. In 2001, though, something (perhaps a temporary increase in mass transfer) caused the material to work its way into the pulsar's magnetosphere and short it out. When that happens, the pulsar wind shuts off, and material can freely fall towards the pulsar. Only if the amount of mass transfer falls so much that the pulsar wind can turn back on will the system go back into quiescence. (This isn't just my crazy theory; it's an idea called radio ejection phases that was proposed before J1023 was discovered.) I think the material that's been swept out of the system probably streams out into a bow shock that's just too small or too faint for us to see in our observations.
In any case, this system is clearly a close cousin to both the radio millisecond pulsars that show gas-caused "eclipses" and the quiescent low-mass X-ray binaries that are presumably their progenitors, so it's worth looking for either to behave like the other. Maybe J1023 isn't so unique after all...