Mapping dark matter with cosmic magnification

We develop a new tool to generate statistically precise dark matter maps from the cosmic magnification of galaxies with distance estimates. We show how to overcome the intrinsic clustering problem using the slope of the luminosity function, because magnification changes strongly over the luminosity function, while intrinsic clustering only changes weakly. This may allow precision cosmology beyond most current systematic limitations. The Square Kilometre Array is able to reconstruct the projected matter density map at smoothing scale similar to 10' with S/N >= 1, at the rate of 200-4000 deg(2) per year, depending on the abundance and evolution of 21 cm emitting galaxies. This power of mapping dark matter is comparable to, or even better than, that of cosmic shear from deep optical surveys or 21 cm surveys.