A few years ago, I was working on a document that would be read by US policymakers, and we had some internal discussion about how to frame our goal. Should we say that we are hoping to discover dark matter? No, we ultimately concluded, because we had, in fact, already done that. We have already discovered dark matter using multiple types of astrophysical observation, including the motions of stars in galaxies, which indicate the presence of an apparently invisible matter-like phenomenon that interacts through gravity and not noticeably through any other force. When we look at light from the dawn of time, when it flew freely through the universe, we can only match our observations with cosmological theories if we assume that dark matter exists.

We realised that what we really wanted to say is that we want to figure out what dark matter . We already know that most of the normally gravitating matter in the universe is an invisible substance that outnumbers visible matter at a rate of about five to one. While this is useful information, it hardly tells us much detail, and many particle physics theories can account for these astrophysical observations. So, the question becomes how we tell the difference between the many different dark matter models. And to sort that out, the devil is in the details.

Certainly, one approach is to look for evidence of dark matter interactions here on Earth. You may have seen recent news items about the LUX-ZEPLIN (LZ) experiment, which announced a single data point that could represent a dark matter particle interacting with atomic nuclei. If this result bears out – and it is a long way off from being statistically significant – it would represent the first evidence that dark matter interacts via nuclear forces. It would also be our most direct look at what dark matter is.

In the absence of substantive data in the lab, we remain dependent on astrophysical observations. In this case, we have to figure out how to use data that seems to be, on the surface, consistent with different hypothetical particles that could be making up dark matter. For example, LZ is looking for weakly interacting massive particles (WIMPs). Meanwhile, I spend most of my research time focused on a category of particles with a very different set of characteristics from WIMPs, called axions. While WIMPs will manifest in a particle-like way in galaxies, axions are more likely to behave as waves. This is a reflection of the principle of wave-particle duality, which tells us that all particles are also waves from the point of view of quantum physics. With axions, their wave-like nature is expected to visibly manifest across astrophysical data, including observations of dwarf galaxies and the cosmic microwave background radiation. Their signature should sometimes look different from what we would expect to see with WIMPs.

To have a chance at figuring out whether dark matter is made of axions rather than WIMPs, we first have to run computer simulations that project what the data might look like for one or the other. Only then can we see if our observations match. And we often have to make simplifying assumptions in order to do our calculations; here’s where the devil shows up.

In two new preprints that I published in September with Hoang Nhan Luu at the University of New Hampshire, we study an axion model that pushes past the simplifying assumptions that are common in our research community. Our work shows that this model – developed by an international team that includes some former colleagues from MIT – may both make simulations go faster and make them more accurate. This kind of work is unlikely to make headlines, but it does what science needs it to do. It helps create the conditions where we will finally move past discovering dark matter, to truly understanding it.