Background: Regularized regression methods such as principal component or partial least squares regression perform well in learning tasks on high dimensional spectral data, but cannot explicitly eliminate irrelevant features. The random forest classifier with its associated Gini feature importance, on the other hand, allows for an explicit feature elimination, but may not be optimally adapted to spectral data due to the topology of its constituent classification trees which are based on orthogonal splits in feature space.

We describe a new type of magnetic trap whose time-averaged, orbiting potential (TOP) supplies tight and harmonic confinement of atoms. The TOP trap allows for long storage times even for cold atom samples by suppressing the loss due to nonadiabatic spin Hips which limits the storage time in an ordinary magnetic quadrupole trap. In preliminary experiments on evaporative cooling of ' Rb atoms in the TOP trap, we obtain a phase-space density enhancement of up to 3 orders of magnitude and temperatures as low as 200 nK. PACS numbers: 32.80.Pj, 42.50.Vk, 85.70.Nk Laser cooling and trapping of neutral atoms can provide up to 15 orders of magnitude phase-space density compression in alkali vapors [I]; however, optical processes limiting density [2] and temperature [3] have obstructed efforts to optically cool an atom sample directly to the Bose-Einstein condensation (BEC) phase transition. Evaporative cooling [4] in a purely magnetic trap offers an appealing, nonoptical cooling mechanism that can enhance the phase-space density of an optically precooled sample by orders of magnitude [5]. The resulting atom