TomoSphero: Fast Differentiable Projector for Planetary and Solar Tomography on Spherical Grids

To understand the invisible, scientists often rely on a technique called tomography. Imagine trying to figure out the shape of a hidden object inside a sealed box not by opening it, but by shining light through it from many different angles and measuring how much light gets through. By combining these measurements, a computer can reconstruct a three-dimensional map of what lies inside. This method is familiar to anyone who has had a medical CT scan, where X-rays reveal the bones and organs within a human body. However, while this technology works beautifully for flat, box-like grids used in hospitals, it faces a unique challenge when applied to the universe itself. Planets and stars are spheres, and their atmospheres do not fit neatly into the straight lines and square boxes that most computer programs use to calculate these images. When scientists try to map the thin, gaseous envelopes surrounding worlds like Earth or the Sun, the standard tools often struggle, creating blurry or inaccurate pictures because the geometry of the object does not match the geometry of the math.

A team of researchers has addressed this mismatch by creating a new software tool called TomoSphero, designed specifically to see the world in curves. The authors, working at the University of Illinois, developed a system that treats the object being studied as a collection of spherical shells rather than a block of cubes. This approach allows the software to trace rays of light as they travel through the curved layers of a planetary atmosphere, calculating exactly how much gas or dust they encounter along the way. The tool is built to work with modern graphics cards, the same powerful chips found in gaming computers, which allows it to perform these complex calculations incredibly fast. More importantly, the software is "differentiable," a technical term meaning it can automatically figure out how to adjust its own internal settings to improve the final image. This feature turns the reconstruction process into a smooth, automated search for the best possible answer, rather than a manual, trial-and-error effort.

The researchers tested their new tool by simulating a mission to study the Earth's exosphere, the outermost layer of our atmosphere where gas particles are so sparse they rarely collide. This region is critical for understanding how our planet loses water over billions of years, as hydrogen atoms escape into space. To do this, they imagined a spacecraft called the Carruthers Geocorona Observatory, which would orbit far from Earth to take continuous pictures of the sunlit hydrogen glow. Using their new software, the team fed simulated measurements from this hypothetical spacecraft into the system. The software successfully reconstructed a three-dimensional map of the hydrogen density, revealing where the gas was thick and where it was thin. In these simulations, the tool proved highly accurate, with errors remaining well below the thresholds required for scientific missions, even in areas where the view was partially blocked by the Earth itself.

What makes this development significant is not just that it works, but how it changes the way scientists can approach these problems. Before this, researchers often had to write complex, custom code for every new type of spherical problem they encountered, a process that was slow and prone to human error. TomoSphero acts as a universal building block that can be plugged into various reconstruction algorithms, allowing scientists to focus on the science rather than the math. It supports different types of measurement setups, from simple parallel beams to more complex cone-shaped views, and can handle grids that are not perfectly uniform, which is useful when scientists need higher detail in specific regions. The software also includes tools to visualize the setup, letting researchers see exactly how their virtual cameras are positioned before they ever start processing data.

The paper confirms that this tool is ready for real-world use, having passed rigorous tests against known mathematical solutions. The researchers showed that the software can handle the massive amount of data required for high-resolution 3D maps without crashing, though they noted that extremely large problems might still require special handling if they exceed the memory of a single computer card. By making the process of mapping spherical objects faster and more flexible, TomoSphero opens the door to more detailed studies of our own atmosphere and those of other worlds. It represents a shift toward using the natural shape of the universe as the foundation for our calculations, rather than forcing the universe to fit into the rigid grids of the past. As missions like the Carruthers Geocorona Observatory prepare to launch, tools like this will be essential for turning raw data into a clear picture of the invisible layers that surround our planet.