(j3.2006) Rationale for RANDOM_INIT?

[Steve Lionel]

On 1/3/2018 6:51 PM, Steven G. Kargl wrote:
> Thanks for the pointer to the papers.  I understand the
> need for RANDOM_INIT as I wrote gfortran's RANDOM_SEED
> years ago.  I made the decision that 'CALL RANDOM_SEED()'
> would re-initialize the prng with the same set of seeds.
> That decision was not popular with some users who expected
> the call to initialize the prng to a new stream

Intel's original implementation did the same as yours, but then changed 
to initialize to a "new stream". I initially proposed to have 
RANDOM_SEED with no arguments do the "new stream", but there were 
objections and we decided a new intrinsic was needed.

Steve