What boundary conditions would allow failure to occur in the gauge length and not at or near the clamps? One is not allowed (in suggesting ways of overcoming stress concentation at the clamps) to create defects in the nanotube or nanowire, to configure the region where failure will occur.  Thus, it is not possible (or is it?)  to create an analog of dog-bone specimens by, e.g., milling away part of the nanowire with a focused ion beam, etc., because this creates defects in the nanowire.