Linear stability analysis is applied to pre-transitional boundary layer flow fields extracted from direct numerical simulations (DNS) of bypass transition (Hack & Zaki, J. Fluid Mech., 2014a). The presence of broadband free-stream forcing in the DNS leads to the formation of a spectrum of high-amplitude streaks inside the boundary layer. The streaky base flow becomes susceptible to high-frequency modal instabilities which are localized on single streaks and which eventually induce breakdown to turbulence. It is shown that linear stability theory can capture the attributes of the instability modes and predict which particular streaks will cause breakdown. Representative examples of outer and inner modes of streak instabilities are examined. The consideration of a large number of instabilities provides statistically relevant results for the properties of the instabilities. Favorable pressure gradient is established to support the outer mode whereas adverse pressure gradient favors the inner instability.

The last part of the presentation focuses on a novel approach to the prediction of streak breakdown based on machine learning. The method is shown to be highly accurate at identifying unstable streaks while the associated computational effort is several orders of magnitude lower than that of the solution of the eigenvalue problem intrinsic to classical stability analysis.