Physics Personal Statementfor Oxford, Cambridge & Imperial

When I first read about GW150914, the signal recorded by LIGO on 14 September 2015, I was less interested in the black holes than in how anyone knew the signal was real. The graph looked so slight that noise seemed the more obvious explanation. Reading about laser interferometry changed that. I began with phase difference and interference, then realised that the discovery depended just as much on statistics, calibration and comparison between detectors as it did on Einstein's theory. Physics stopped looking like a set of finished equations and started looking like an argument about whether a measurement deserves to be believed. That is why I want to study it at university. I am especially drawn to the point where theory, mathematical modelling and measurement meet, particularly in astrophysics, because I want to get better at judging what makes weak-signal evidence strong enough to trust.

My studies have given me the language to pursue that question more seriously. A-level waves made it clearer why a Michelson interferometer can turn a tiny path difference into a measurable change in light intensity, while simple harmonic motion and damping helped me see LIGO's suspended mirrors as part of the physics rather than background engineering. I liked that ideas from class reappeared inside an instrument built to test general relativity. That shift also changed how I approached problems in school. I became much more careful about checking dimensions and limiting cases before trusting neat algebra, which slowed me down at first but made me less likely to mistake tidy working for sound reasoning. I wanted to see what that looked like outside textbook questions, so I based my EPQ on how gravitational-wave signals are extracted from detector noise using public data from the Gravitational Wave Open Science Center. Using Python, NumPy, SciPy and Matplotlib, I downloaded strain data around GW150914 from the Hanford and Livingston detectors, applied a band-pass filter, compared the traces and used a Fourier transform to look at their frequency content. My first plots were too convincing. By narrowing the filter range aggressively, I could produce something chirp-like in stretches that were mostly noise. That mistake became the most useful part of the project because it forced me to treat my own analysis more sceptically, compare the two detectors more carefully and think harder about coincidence and signal-to-noise. I did not reproduce the event with the precision of published work, but I finished with a better sense of the difference between uncovering a pattern and imposing one.