Actually the scaling power is not a law. It is more a rule of thumb based on curve fitting neutron or x-ray yields versus input current. Some literature compilations show scaling powers of 3.6 to 4.5. The 5th power scaling is probably the upper limit of efficiency predicted by LPPs theory.

Here is a link to some papers discussing the scaling models: http://www.plasmafocus.net/IPFS/2015%20papers/2015%20Publications.pdf

Here is a link to modeling dpf devices using Excel spreadsheets: http://www.plasmafocus.net/IPFS/modelpackage/File1RADPF.htm