Cooling Improves Cosmic Microwave Background Map-making when Low-frequency Noise is Large

In the context of Cosmic Microwave Background data analysis, we study solution to equation that transforms scanning into a map. As originally suggested in "messenger" methods for solving linear systems, split noise covariance uniform and non-uniform parts adjust their relative weights during iterative solution. With simulations, mock instrumental with different properties, find this "cooling" or perturbative approach is particularly effective when there significant low-frequency timestream. such cases, conjugate gradient algorithm applied modified system converges faster higher fidelity than standard approach. We give an analytic estimate parameter controls how gradually should change course