“…This is clearly related to the notion of 'complete blow-up' ( [16], [12]), but it is not the same. If w 0 has a direction of mean convergence which is equivalent to say the solution has a least blow-up time, (i.e., the blow-up time T * agrees with the blow-up time of a spatially constant solution with an initial data sup w 0 ), then the blow-up profile w(x, T * ) has a blow-up direction (see [3]). As remarked in Appendix the w(x, T * ) fulfills (3).…”