On the lifting problem for homogeneous ideals (Q5946459)

Let \(S=k[x_1, \dots, x_n]\) and \(R=k[x_0, \dots, x_n]\) be polynomial rings over a field \(k\). Let \(J\subset S\) be a homogeneous ideal. A homogeneous ideal \(I\subset R\) is a ``lifting'' of \(J\) if \(x_0\) is \(R/I\)-regular and \((I,x_0)/(x_0) \simeq J\). An important question concerning this notion is: When does \(J\) admit a radical lifting? The authors propose an elementary algorithm for finding all the liftings of a homogeneous ideal. For \(f\in S\) (not necessarily homogeneous) let \(LF(f)\) be the leading form of \(f\) and \(f^h\in R\) the homogenization of \(f\). Assume \(J=(f_1, \dots,f_r)\) with \(f_i\in S\) homogeneous. Any homogeneous lifting of \(f_i\) to \(R\) has to be of the form \(q_i^h\) where \(q_i=f_i +g_i\) with \(\deg(g_i) <\deg(f_i)\). Let \(I=(q^h_1, \dots, q^h_r)\subset R\) and \(I^*=(q_1, \dots, q_r)\subset S\). It is easy to see that \(x_0\) is \(R/I\)-regular iff \((I^*)^h=I\) iff \(LF(I^*)=J\). Now, the authors show that \(LF(I^*)=J\) iff for any homogeneous relation \(s_1f_1+ \cdots+ s_rf_r =0\) one has \(s_1g_1+ \cdots+ s_rg_r= a_1q_1+ \cdots+a_r q_r\) with \(\deg(s_1g_1 +\cdots +s_rg_r)=\max (\deg(a_1q_1), \dots,\deg (a_rq_r))\). Of course, it suffices to check this condition for a finite system of homogeneous generators of the module of relations between \(f_1,\dots, f_r\). Using their algorithm the authors produce for \(k=\mathbb{Q}\) and \(n=4\) an example of a homogeneous ideal \(J\subset S\) with \(\dim(S/J)=0\) which does not admit a radical lifting. The previously known examples (for infinite \(k)\) assumed \(n\geq 4\) and \(\dim(S/J)=1\). The authors assert that in the case \(n=3\) the question is open.