π（yx）=π（μ（x）y）x,y∈H.PDFVIP

HOPF MODULES AND INTEGRALS: NAKAYAMA AUTOMORPHISMS ROLF FARNSTEINER We adopt the notation from our previous lectures [3, 4] and assume in particular that H is a ?nite dimensional Hopf algebra over a ?eld k. As noted in [4, Cor. 1], H is a Frobenuis algebra, ? with any non-zero left integral π ∈ H? giving rise to a non-degenerate associative form ( , )π. Since H is a Frobenius algebra,R the Nakayama functor on mod H is just the twist by any Nakayama automorphism H, see [2]. Recall that the Nakayama automorphism μ relative to ( , )π is de?ned via π(yx) = π(μ(x)y) ? x, y ∈ H. In this lecture, we relate μ to the so-called left modular function of H, thereby showing in particular that μ has ?nite order. The relevant results can be found in [7] and [5]. r r Given x ∈ H , we have hx ∈ H for every h ∈ H. Consequently, there exists an algebra homomorphismR ζr : H ?→ k with R r hx = ζr(h)x ? h ∈ H, x ∈ . ZH ? r This homomorphism is referred to as the right modular function of H. Clearly, H = H if and only if ζr = ε coincides with the counit of H. In that case H is called unimodularR. TheR left modular function ζ? is de?ned analogously. By [4, Thm.], we have ?1 ζ? = ζr ? η . If γ : H ?→ k is a homomorphism of k-algebras, then the convolution idH ?γ : H ?→ H ; h 7→ h(1)γ(h(2)) X(h) is an automorphism of the algebra H, whose inverse is given by idH ?(γ ? η). ?2 Theorem. The automorphism η ?(idH ?ζ?)