Local residues of holomorphic 1-forms on an isolated surface singularity (Q1849573)

The following theorem on \((\mathbb{C}^n, 0)\) has been generalized: Let \(f_1, \ldots, f_n\in O_{\mathbb{C}^n,0}\) be a regular sequence, \[ f=(f_1, \ldots, f_n), \quad J_f=\det\left(\frac{\partial f_i}{\partial x_j}\right), \quad Q_f=O_{\mathbb{C}^n, 0}/\langle f_1, \ldots, f_n\rangle \] then \[ \text{ind}_{\mathbb{C}^n,0} (f)=\dim_\mathbb{C} Q_f=\text{res}_{\mathbb{C}^n, 0}(J_f|f). \] Here the local residue of \(J_f\) with respect to \(f\) is \(\text{res}_{\mathbb{C}^n, 0} (J_f|f)=\int_\Sigma \frac{J_f}{f_1\cdots f_n} dx_1\wedge\ldots\wedge dx_n\) with \(\Sigma=\{|f_1|=\cdots=|f_n|=\delta \}\) and the Poincaré-Hopf-index, \(\text{ind}_{\mathbb{C}^n, 0}(f)\) is the degree of the map \(\frac{f}{\|f\|}\) at \(0\). The theorem is generalized to a germ of a \(2\)-dimensional complete intersection singularity \((V,0) \subset(\mathbb{C}^n, 0)\). The generalization is based on results of Ebeling and Gusein-Zade: Let \((V,0)\in(\mathbb{C}^n,0)\) be an isolated complete intersection singularity defined by \(f_1, \ldots, f_q\in \mathcal{O}_{\mathbb{C}^n,0}\) and \(\omega=\sum_{i=1}^n \omega_i dx_i\) be the germ of a holomorphic \(1\)-form on \((\mathbb{C}^n,0)\). The index \(\text{ind}_{V,0}(\omega)\) is defined as the degree of the map \((\omega, df_1, \ldots, df_q):V\cap S_\delta\to\text{Grass}(q+1, \mathbb{C}^n), S_\delta\) a small sphere around the origin in \(\mathbb{C}^n\). Let \(I\) be the ideal in \(\mathcal{O}_{\mathbb{C}^n,0}\) generated by \(f_1, \ldots, f_q\) and the \((q+1)\)-minors of the matrix \[ \begin{pmatrix} \frac{\partial f_1}{\partial x_1} & \cdots & \frac{\partial f_1}{\partial x_n}\\ \vdots & & \\ \frac{\partial f_q}{\partial x_1} & \cdots & \frac{\partial f_q}{\partial x_n}\\ \omega_1 & & \omega_n \end{pmatrix} \] and \(A=\mathcal{O}_{\mathbb{C}^n,0}/I\) then \(\dim_\mathbb{C} A=\text{ind}_{V,0}(\omega)\). The author proves that there exist (in the \(2\)--dimensional case) a residue map \(\text{res}_{V,0}:A\to\mathbb{C}\), the so-called relative residue form of \(\omega\), and a \(\sigma\in A\) such that \(\text{ind}_{V,0}(\omega)=\text{res}_{V,0}(\sigma)\), \(\sigma\) and \(\text{res}_{V,0}\) are explicitly constructed: Let \(M_i\) be the matrix obtained from \[ \begin{pmatrix} \frac{\partial f_1}{\partial x_1} & \cdots & \frac{\partial f_1}{\partial x_n} \\ \vdots & & \\ \frac{\partial f_{n-2}}{\partial x_1} & \cdots & \frac{\partial f_{n-2}}{\partial x_n} \\ \omega_1 & & \omega_n \end{pmatrix} \] by cancelling the \(i\)-th column, \(m_i=\det(M_i)\) and \(M=\det((-1)^i\frac{\partial m_i}{\partial x_j})\) then \(\sigma\) is the coefficient of \(t^{n-2}\) in the characteristic polynomial of \(M\). If \(m_1, m_2\) is a regular \(\mathcal{O}_{V,0}\)-sequence (which can be achieved after linear coordinate change) then \[ \text{res}_{V,0}(h|m_1, m_2)=\frac{1}{(2\pi i)^2} \int_\Sigma \frac{h dx_1\wedge dx_2}{m_1 m_2} \] with \(\Sigma=\{f_1=\ldots, f_{n-z}=0, |m_1|=|m_2|=\delta\}\).