By the spectral theorem a symmetric positive-definite matrix has , with orthogonal and . Define . It is symmetric positive-definite and . For uniqueness, any symmetric positive-definite square root commutes with , hence preserves each eigenspace of . On its -eigenspace, is symmetric positive-definite and all its eigenvalues have square , so all equal . Therefore , including repeated eigenspaces.
For , putThen and , so ; also . Any other right polar decomposition has , so uniqueness of the positive root fixes and then . The left factor is consequently fixed too.
The principal stretches are the positive eigenvalues of , equivalently the singular values of . For a unit vector in an orthonormal eigenbasis of ,Thus , attained precisely by the unit vectors in the eigenspace of belonging to . With a simple largest eigenvalue these are its two opposite unit eigenvectors; degeneracy allows every unit combination within that eigenspace.
For the simple shear,A line with nonzero direction keeps its length exactly whenHence , with arbitrary. Every segment in the cube parallel to one of these directions preserves length; there is no restriction on its starting point other than remaining in the cube.
The two in-plane stretches are , and the third is one. A maximizing direction for is , since the eigenvalue equation gives . Therefore as . At exactly zero shear all directions tie; the limit refers to the selected maximizing direction for positive shear. These are the length-preserving directions and principal directions of simple shear.