A critique of the proposed SEG vector subroutine standard and an alternative

I was not intending to discuss vector computation until artiele IX of my series on computer science for geophysicists, but I have recently been given a copy of the proposed SEG vector subroutine standard, and I was sufficiently worried that I thought I would bring the relevant parts of the discussion forward. I would normally have raised these as comments to the relevant SEG committee, but unfortunately, I only became aware of its work some considerable time after the deadline for such comments had passed and I felt that it was therefore appropriate to write this paper. To give a brief introduction, vector arithmetic computation such as is extremely common in seismic data processing involves the definition of an algorithm in terms of vector operations and the subsequent efficient implementation thereof.