The quaternion formalism for Möbius groups in four or fewer dimensions (Q686372)

The interpretation of Möbius geometry on the Euclidean 4-spheres as projective geometry on the projective line over the real quaternions (i.e. \(PG(2,H))\) seems to go back to \textit{E. Study} [Ein Seitenstück zur Theorie der linearen Transformationen einer komplexen Veränderlichen I, II, III, IV, Math. Z. 18, 55-86, 201-229 (1923; JFM 49.0075.01), Math. Z. 21, 45-71, 174-194 (1924; JFM 50.0668.03]. The present account is self contained and gives a thorough discussion of direct and opposite Möbius transformations viewed as projectivities and anti- projectivities of \(PG(2,H)\). The results include, e.g., an explicit formula for inverting such transformations. Moreover there is an interesting comparison of the Möbius groups of the 2-sphere (or, in other words, \(PG(2,C))\) and the 4-sphere. Well-known results from the complex case do not immediately carry over to the quaternion case, since conjugation is an automorphism of complex numbers, but an anti-automorphism of quaternions. Hence opposite Möbius transformations (corresponding to anti-projectivities) have to be seen as collineations of \(PG(2,C)\), but as dualities of \(PG(2,H)\). On the other hand direct Möbius transformations may always be viewed as collineations. This causes extra trouble when describing products and conjugates of Möbius transformations of the 4-sphere in terms of \(2 \times 2\) matrices of quaternions.