A Stochastic LBFGS Algorithm for Radio Interferometric Calibration
This work addresses the challenge of handling very large datasets in radio astronomy calibration, which is crucial for detecting transient signals like fast radio bursts, and it also has potential applications in deep learning optimization.
The authors tackled the problem of calibrating radio interferometric data at fine time and frequency resolution without requiring the entire dataset to fit into memory, which is necessary for detecting signals like fast radio bursts. They developed a stochastic LBFGS algorithm that enables calibration at such high resolution and demonstrated its applicability by training deep neural networks, showing competitive performance against first-order optimization methods.
We present a stochastic, limited-memory Broyden Fletcher Goldfarb Shanno (LBFGS) algorithm that is suitable for handling very large amounts of data. A direct application of this algorithm is radio interferometric calibration of raw data at fine time and frequency resolution. Almost all existing radio interferometric calibration algorithms assume that it is possible to fit the dataset being calibrated into memory. Therefore, the raw data is averaged in time and frequency to reduce its size by many orders of magnitude before calibration is performed. However, this averaging is detrimental for the detection of some signals of interest that have narrow bandwidth and time duration such as fast radio bursts (FRBs). Using the proposed algorithm, it is possible to calibrate data at such a fine resolution that they cannot be entirely loaded into memory, thus preserving such signals. As an additional demonstration, we use the proposed algorithm for training deep neural networks and compare the performance against the mainstream first order optimization algorithms that are used in deep learning.