Difference Image Analysis (DIA) is a technique for detecting transient sources and performing photometric measurements. To test the performance of the DESC DIA pipeline, we add simulated sources onto DC2 calexp images. By measuring the fraction of detected sources (detection efficiency) and photometric accuracy, we get confirmed that the DIA pipeline can produce reasonable results.
Most of detected artifacts are subtraction residuals of bright background sources. Artifacts caused by pixel saturation can be removed by applying base flag selection. The majority of remaining artifacts are dipoles. A more flexible matching kernel with high spatial degree of freedom can be useful to remove them. We also have artifacts from pixel correlation. Tracking pixel covariance matrix and applying whitening filters are helpful to reduce pixel correlations. Ring artifacts can be found when the calexp PSF is sharper than the deepCoadd PSF. Pre-convolution or convolving calexp to deepCoadd are possible methods to avoid deconvolution.