Dear Anne,
Secondly, the answer to your question depends on what kind of SPM beamforming you are using. If you are using the Bayesian beamformer in the MSP framework (EBB) I don't think you should do anything special because that framework includes automatic dimensionality reduction prior to inverse computation. If you are using the DAiSS toolbox for SPM12 you could do an explicit dimensionality reduction there and make sure your dimensionality is smaller than the minimal number of data components you might have. If you use the old SPM8 beamforming tools, there the only option is to regularise with some high enough coefficient to make sure there is no problem. I can't say what is 'high enough' you need to play with it.
Best,
Vladimir