Quantum Physics Could Help Us Find Earth 2.0

When astronomers talk about directly imaging an exoplanet that is orbiting a far away star, the analogy they most commonly go with is trying to spot a fireflight next to a massive search light. An Earth-like exoplanet is incredibly dim - usually between 100 million and 10 billion times fainter than its host star. Understandably, that makes them very difficult to see. But a new paper from Hyunsoo Choi of Hanyang University in South Korea and his co-authors, which is available in pre-print on arXiv, describes a theoretical solution - use a mix of smart computer algorithms and quantum physics. Quantum physics is probably not the first thing that comes to mind when thinking about imaging something as large as a planet. In fact, the fundamental tension at the heart of physics, between quantum mechanics and relativity, doesn’t bode well for its usefulness either. But to understand how exactly it can be useful, it’s first helpful to understand the concept of the Rayleigh limit. If two objects are incredibly close to each other, and each is emitting photons, their light blurs together into a single blob if they are closer than a distance known as the Rayleigh limit. In the case of exoplanets, this would mean the planet’s light is completely subsumed into that of a star. And a normal photodetector will just detect a photon and will not be able to distinguish whether that photon came from the planet or its host star. Fraser talks about the race to find Earth 2.0 Enter quantum mechanics. In quantum mechanics, photons have a lot more information in them than just their energy level (or “brightness”). One such tidbit of information is known as the wave shape, and measuring a photon based on its wave shape is known as spatial-mode measurement. By designing a system where the photons are sorted by their wave patterns before they hit a photon detector, the system can pull out that additional information that regular cameras miss. Making this work in real time required building a continuous feedback loop into their image analysis software. First, they introduced a logarithmic scale to ensure their algorithm can track extreme differences in brightness between the planet and the star. As the algorithm begins to guess what the star system looks like (i.e., how many planets and stars there are), it calculates something called the Symmetric Logarithmic Derivative, which tells the photon sorter how to shift and adjust to make sure the maximum amount of quantum information is retained. And, crucially, they took a human-generated guess about the number of planets the algorithm should look for, replacing it with a statistical tool called the Bayesian Information Criterion. After implementing their feedback loop, the authors ran the algorithm on a simulated star system with one star and two planets - one of which was only 10,000 time dimmer than its host star, while the other was 100 million times dimmer, and both were well within the Rayleigh limit. Using Monte Carlo simulations (which slightly vary starting conditions over multiple runs of a simulation), the algorithm correctly guessed the total number of objects (i.e. 3) 72.5% of the time. Those times when the system worked, it could also locate the exact location of the planets in the simulation down to within a single pixel. It was also able to estimate the true brightness of the ultra-dim planet within a factor of two 99.7% of the time, assuming it had correctly estimated the number of objects. Suppressing starlight is one of the critical challenges of finding exoplanets, and Fraser discusses in this video Simulations are great, but they don’t always translate neatly into the real world. But the authors did their best to simulate it, intentionally messing up the alignment of the virtual telescope creating the datasets used in the simulation. The algorithm was able to adapt to this artificially introduced noise on the fly, with its success rate only dropping to 71.3%. However, there are numerous other sources of noise in real-world telescopes, and its unclear how well the algorithm can handle other “noise” at this point. Admittedly the paper describes just a computer simulation. But even so, it represents a massive leap forward in coming up with quantum imaging systems that could find planets that are up to 100 million times dimmer than their star - a massive improvement from current quantum imaging systems that can only manage at 1/1,000 contrast between their target planet and its host star. It also clearly delineates a path forward for the hardware developers. While physical implementations typically lag well behind theoretical proof, it’s only a matter of time before someone builds one of these systems and attempts to find a new exoplanet with it. It remains to be seen what other kinks will need to be worked out for that effort to be successful. But just from a purely theoretical physics standpoint, this combination of exoplanet hunting with quantum imaging is a truly unique take on one of the most fascinating areas of modern astronomy. Learn More: H. Choi et al - Exoplanet Detection Using Adaptive Quantum-Optimal Measurement UT - UCF Researchers Are Developing Tools to Help Find Habitable Planets UT - Are We Entering the Era of Quantum Telescopes? UT - A New Superconducting Camera can Resolve Single Photons