The investigators will study the initiation mechanism of coronal mass ejections (CME) in order to enable physics based predictions of coronal mass ejection occurrence. The main effort is to extend calculation of a mechanism termed "magnetic breakout" in 2.5 and 3 dimensions using a numerical technique commonly referred to as adaptive mesh refinement. If this model is found to be viable, it will result in a true physics-based scheme for predicting coronal mass ejections. The model has been designed to achieve superior computational performance on highly parallel computer systems expected to be in use for space weather predictions in the coming decade.