The Ak and Bk approximate CI methods. Comment on a paper by Maynau and Heully