Challenges in cosmic magnification reconstruction by magnification response

Cosmic magnification on the observed galaxy overdensity is a promising weak gravitational lensing tracer. Current cosmic magnification reconstruction algorithms, the analytical method of blind separation (ABS) and constrained internal linear combination (cILC), intend to disentangle the weak lensing signal using the magnification response in various flux bins. In this work, we reveal an unrecognized systematic bias arising from the difference between galaxy bias and the galaxy-lensing cross-correlation bias, due to the mismatch between the weak lensing kernel and the redshift distribution of photometric objects. It results into a galaxy-lensing degeneracy, which invalidates ABS as an exact solution. Based on the simulated cosmoDC2 galaxies, we verify that the recovered weak lensing amplitude by ABS is biased low by $\ensuremath{\sim}10%$. cILC, including a modified version proposed here, also suffers from systematic bias of comparable amplitude. Combining flux and color information leads to significant reduction in statistical errors, but fails to eliminate the aforementioned bias. With the presence of this newly found systematic, it remains a severe challenge in blindly and robustly separating the cosmic magnification from the galaxy intrinsic clustering.