The day in conversation for 2026-11-02

discussion

Transcript

The discussion: Vera: The uncertainty in the Leonid space predictions stemming from solar flux forecasts really highlights how brittle our long-term extrapolations are, especially when we consider the dependency on those specific solar cycle models. Jocelyn: That dependency is exactly what frustrates me; we spend so much time refining the orbital mechanics, but if the input for the flux model drifts by a small amount, that translates directly into a significant uncertainty in those two-and-a-half day accuracy claims. Have you looked at how those forecasts integrate with historical solar variability data from earlier days? Subrahmanyan: I was looking at how we applied Monte Carlo simulations to the timing errors last week, and it seems the main issue isn't just the forecast error itself, but rather how aggressively we weight past solar cycles versus the immediate, short-term predictions. It makes me wonder if we should be focusing less on predicting absolute lifetimes and more on quantifying the covariance structure between those two variables. Vera: Quantifying covariance is a good pivot point. Speaking of data pipelines, I’m interested in how the turboPDZ framework tackles that noise inherent in photometric redshift distributions; is it primarily helping to stabilize the reliability

More episodes

← Home