Nilpotent self-adjoint elements, anticommutation and star transposition #
In a star monoid with zero whose star is proper (IsProperStar: star x * x = 0 forces
x = 0), such as a matrix ring over a star-ordered ring like ℝ or ℂ, a self-adjoint element
x with x * x = star x * x = 0 is zero. Since the powers of a self-adjoint element are
self-adjoint, peeling off squares shows that a nilpotent self-adjoint element is zero. This is
what rules out nonzero nilpotent ideals in an algebra of matrices closed under conjugate
transposition.
In a star ring, taking adjoints shows that an element anticommuting with a self-adjoint γ has an
adjoint anticommuting with γ, as needed for gradings. Similarly, taking stars transposes
one-sided relations with a self-adjoint element: x * q = x becomes q * x = x for self-adjoint
x, and q * u = 0 becomes star u * q = 0. For a self-adjoint p, the
element 2 p - 1 is again self-adjoint.
Main results #
IsSelfAdjoint.eq_zero_of_isNilpotent: a nilpotent self-adjoint element is zero.IsSelfAdjoint.mul_star_eq_neg_star_mul: ifγ * x = -(x * γ)for a self-adjointγ, thenγ * star x = -(star x * γ).IsSelfAdjoint.mul_eq_right_of_mul_eq_left: for self-adjointxandq,x * q = ximpliesq * x = x.IsSelfAdjoint.star_mul_eq_zero_of_mul_eq_zero: for self-adjointq,q * u = 0impliesstar u * q = 0.IsSelfAdjoint.two_mul_sub_one:2 p - 1is self-adjoint for a self-adjointp.
A nilpotent self-adjoint element is zero when the star is proper. Some such hypothesis is
needed: over ZMod 2 the symmetric matrix !![1, 1; 1, 1] squares to zero.
For self-adjoint x and q, taking stars turns x * q = x into q * x = x.
For self-adjoint q, taking stars turns q * u = 0 into star u * q = 0.
For a self-adjoint p, 2 p - 1 is self-adjoint.