Search for: All records

This paper presents a comparative performance analysis of three popular Python data manipulation libraries—Pandas, Polars, and Dask—within the context of deep learning training pipelines. The existing studies in this area do not embed the libraries inside a full deep-learning training pipeline where data loading, preprocessing, and batch feeding interact tightly with GPU workloads. To bridge this gap, we integrate Pandas, Polars, and Dask into representative deep learning training and inference pipelines and conduct experiments across a wide range of various machine learning models and datasets, measuring key performance indicators such as runtime, memory usage, disk usage, and energy consumption (CPU and GPU). Our comprehensive analysis reveals that Polars consistently minimizes CPU energy consumption on larger workloads, while Pandas remains competitive for moderate sizes. Dask’s overhead can lead to higher energy usage on small to moderate datasets. All three libraries achieve similar runtimes for heavy GPU workloads (ResNet, Mask R-CNN). Polars and Pandas maintain lower CPU memory footprints than Dask, but Dask o"ers easier scalability if data truly exceeds available RAM. Polars shows marginal energy savings on the CPU during preprocessing. 

Abstract We present a census of 100 pulsars, the largest below 100 MHz, including 94 normal pulsars and six millisecond pulsars, with the Long Wavelength Array (LWA). Pulse profiles are detected across a range of frequencies from 26–88 MHz, including new narrowband profiles facilitating profile evolution studies, and breaks in pulsar spectra at low frequencies. We report mean flux density, spectral index, curvature, and low-frequency turnover-frequency measurements for 97 pulsars, including new measurements for 61 sources. Multifrequency profile widths are presented for all pulsars, including component spacing for 27 pulsars with two components. Polarized emission is detected from 27 of the sources (the largest sample at these frequencies) in multiple frequency bands, with one new detection. We also provide new timing solutions for five recently discovered pulsars. Low-frequency observations with the LWA are especially sensitive to propagation effects arising in the interstellar medium. We have made the most sensitive measurements of pulsar dispersion measures (DMs) and rotation measures, with median uncertainties of 2.9 × 10⁻⁴pc cm⁻³and 0.01 rad m⁻², respectively, and can track their variations over almost a decade, along with other frequency-dependent effects. This allows for stringent limits on average magnetic fields, with no variations detected above ∼20 nG. Finally, the census yields some interesting phenomena in individual sources, including the detection of frequency- and time-dependent DM variations in B2217+47, and the detection of highly circularly polarized emission from J0051+0423. 

We establish the first finite-time logarithmic regret bounds for the self-tuning regulation problem. We introduce a modified version of the certainty equivalence algorithm, which we call PIECE, that clips inputs in addition to utilizing probing inputs for exploration. We show that it has a ClogT upper bound on the regret after T time-steps for bounded noise, and Clog3T in the case of sub-Gaussian noise, unlike the LQ problem where logarithmic regret is shown to be not possible. The PIECE algorithm is also designed to address the critical challenge of poor initial transient performance of reinforcement learning algorithms for linear systems. Comparative simulation results illustrate the improved performance of PIECE.