Numerical relativity: maximizing the scientific payoff from gravitational wave detection

For many of the proposed LIGO sources, we are currently unable to produce reliable theoretical waveforms or event rates. Usually the difficulty is in performing numerical simulations. These problems seriously undermine our ability to extract science from LIGO, or in some cases even to detect sources at all. We describe two sources where these problems are present: the merger of a binary black hole system, and a newborn neutron star unstable to r-modes. We explain the origin of the difficulties, and how they might be overcome.