Embeddings of the line in the plane. (English) Zbl 0332.14004

Let \(k\) be a field. Let us think of the affine plane \(\mathbb A^2\) as the algebraic variety with coordinate ring \(k[Y,X]\) and the affine line \(\mathbb A^1\) as the algebraic variety with coordinate ring \(k[Z]\). (If \(k\) is algebraically closed it is safe to think \(k^2\) and \(k\).) Then the zero-set \(C_1\) of the polynomial \(Y\) is a closed subvariety of \(\mathbb A^2\) isomorphic to \(\mathbb A^1\), and so is the zero-set \(C_2\) of the (non-linear) polynomial \(Y+X^2\). Here we obviously have an automorphism of \(\mathbb A^2\) that sends \(C_2\) to \(C_1\), and it is natural to ask whether this is always the case. Along the way it is natural to inquire about the nature of the group of all automorphisms of \(\mathbb A^2\).
The paper “Embeddings of the line in the plane” by S. S. Abhyankar and T.-T. Moh [AM1] deals with these questions. For reasons that will become clear, it is usually called the “Epimorphism Paper”. The earlier paper, “Newton-Puiseux expansion and generalized Tschirnhausen transformation” [AM2] will be termed the “Expansions Paper”. It gives a comprehensive treatment, very novel at the time, of “plane curves with one place at infinity”. It is an essential predecessor to [AM1] and has found far reaching other applications, such as the Jacobian problem, for instance. Both papers were very influential.
Many of the topics of this review were intensely discussed in Abhyankar’s circle during the Séminaire de Mathématiques Supérieures of 1970 at the Université de Montréal. The Séminaire is vividly remembered by the participants for the great intellectual ferment surrounding it, mathematical and otherwise. The lecturers were Shreeram Abhyankar, Michael Artin, Alexandre Grothendieck and Masayoshi Nagata. This was a time of great abstraction in Algebraic Geometry, but also a time of newly heightened interest in concrete and seemingly elementary problems, some very famous for being as hard to solve as easy to formulate.
Both the “Jacobian Problem” and the “Lines in the Plane Problem” belong in that class. For the lines it was accepted that the case of “one characteristic pair” (relatively prime \(X\)- and \(Y\)-degrees) was doable, and Oscar Zariski had suggested to Abhyankar that the case of bi-degrees having a prime GCD, or a bit more generally, those with “two characteristic pairs”, be seriously considered. The brilliant full solution coming soon after, not only settled a fundamental mathematical problem, but it also led to a veritable explosion in related research activity. The result clearly fascinated a large number of mathematicians. Many additional proofs were published with techniques from topology, algebraic surface theory, complex variables, alternate arrangements bypassing or modifying the Newton-Puiseux expansions of [AM2], and so on. Some of these are mentioned below when needed as a reference, but we did not try to present a full list. Of course, all alternate proofs are valuable and may lead to new developments. The interested reader can easily locate variant proofs by a simple data base search.
Shreeram Abhyankar was our friend, guru, mentor and teacher. We dedicate this review to his inspiring mathematical career.
The Epimorphism Paper
Let \(k\) be a field of characteristic \(\pi\) and let \(X, Y, Z\) be indeterminates over \(k\).
Main Theorem. Let \(u\) and \(v\) be non-constant polynomials of degree \(m\) and \(n\) in \(Z\) with coefficients in \(k\). Assume that \(k[Z]=k[u,v]\). Assume also that either \(m\) or \(n\) is not divisible by \(\pi\). Then either \(m\) divides \(n\) or \(n\) divides \(m\).
We will also describe the conclusion of the theorem by the phrase “\((m,n)\) is principal”.
Using the theorem, we can deduce that if \(m \geq n\) there is a \(c\in k\) so that \(u'=u-cv^{m/n} \in k[Z]\) with \(\deg(u')<\deg(u)\). Clearly \(k[Z]=k[u',v]\). If \(m<n\), then we find \(c\) so that \(k[Z]=k[u, v'], v'=v-cu^{n/m}, \deg(v') <\deg(v)\). By induction on \(\deg(u)+\deg(v)\), this gives the
Epimorphism Theorem. Let \(\gamma: k[Y,X] \rightarrow k[Z]\) be the \(k\)-epimorphism with \(\gamma(X)=0\) and \(\gamma(Y)=Z\). Let \(\alpha: k[Y,X] \rightarrow k[Z]\) be any \(k\)-epimorphism such that at least one of \(\deg_Z\alpha(X)\) and \(\deg_Z\alpha(Y)\) is not divisible by \(\pi\). Then there exists an automorphism \(\delta: k[Y,X] \rightarrow k[Y,X]\) such that \(\gamma=\alpha \delta\).
Call an automorphism \(\tau:k[Y,X] \rightarrow k[Y,X]\) elementary if \(\tau(Y,X)=(X,Y)\) or \(\tau(Y,X) = (bY+f(X),aX)\) where \(f(X)\in k[X]\) and \(a,b\in k\) are non zero. Call \(\tau\) tame, if it is a composite of elementary automorphisms.
It is then deduced that:
Addendum to the Epimorphism Theorem. \(\delta\) can be chosen to be a tame automorphism.
The following lemma is crucial in linking the epimorphism problem to the expansion techniques of the earlier paper [AM2]. In [AM1] it is given a very elementary treatment in the spirit of High School Algebra. It will reappear later in a more sophisticated form in a general discussion of plane curves with one place at infinity.
We will call a polynomial \(F(Y,X)\) pre-monic in \(Y\) of degree \(n\) if \(F(Y,X)=a_0Y^n+a_1Y^{n-1}+\cdots \in k[X][Y]\) with \(0\neq a_0\in k\).
Lemma. Let \(u\in k[Z]\) be of \(Z\)-degree \(n>0\). Let \(v\in k[Z]\) be such that \(k[Z]=k[u,v]\). Let \(\alpha: k[Y,X]\rightarrow k[Z]\) be the \(k\) homomorphism with \(\alpha(Y)=v\) and \(\alpha(X)=u\). Then a generator \(F(Y,X)\) of \(Ker(\alpha)\) is pre-monic of degree \(n\) in \(Y\) and irreducible in \(k((X^{-1}))[Y]\). This is also described as, \(F(Y,X)\) is a curve with one rational place at infinity.
We can, of course, choose \(F\) to be monic in \(Y\), but let us note that unless \(v=0\) and hence \(\deg(u)=1\), \(F\) is also pre-monic in \(X\) of some degree \(m\geq 0\).
The following is a more geometric, and equivalent, version of the Epimorphism Theorem. It explains the title of the Epimorphism Paper.
Embedding Theorem. Let \(C\) be a closed curve in the affine plane \(\mathbb A^2_k=\mathrm{Spec}(k[Y,X])\). Assume that \(C\) is biregularly isomorphic to the affine line \(\mathbb A^1_k=\mathrm{Spec}(k[Z])\). Let \(F\) be a generator of its ideal in \(k[Y,X]\) and assume that either \(\deg_YF\) or \(\deg_XF\) is not divisible by \(\pi\). Then \(F\) is a variable in \(k[Y,X]\), i.e., there exists \(G\in k[Y,X]\) such that \(k[Y,X]=k[F,G]\). Moreover, the automorphism \(\delta:k[Y,X]\rightarrow k[Y,X]\) defined by \(\delta(Y)=F, \delta(X)=G\) is tame.
The first counterexample to the above theorems in case \(\pi>0\) and the gcd of \(\deg_YF\) and \( \deg_XF\) is divisible by \(\pi\) was, to our knowledge, given by B. Segre in [Se]. Abhyankar and Moh give a family of counterexamples of the same type. Let us describe the simplest one:
Consider \(u=Z-Z^{r\pi}\) and \(v=Z^{\pi^2}\) in \(k[Z]\). Then \(Z=u+(u^{\pi}+v^r)^r\). Let \(\alpha: k[Y,X]\rightarrow k[Z]\) be the \(k\) homomorphism with \(\alpha(Y)=v\) and \(\alpha(X)=u\). Then \(F(Y,X)=Y^{r\pi}-Y+X^{\pi^2}\) is a generator of \(Ker(\alpha)\), which is monic in \(Y\). Taking \(r>1\) and prime to \(\pi\) we have a counterexample to the Main Theorem.
Interest in the automorphism group of \(k[Y,X]\) is older than the above results. We have the following theorem, proved by H. W. E. Jung in 1942 [J] for \(\pi=0\) and by W. van der Kulk in general in 1953 [VdK]. It can now be deduced from the embedding theorem when \(\pi=0\).
Automorphism Theorem. Any automorphism of \(k[Y,X]\) is tame.
Van der Kulk also proves a uniqueness theorem for the decomposition of a given automorphism into elementary ones. This was further elaborated by M. Nagata in [Na2].
In a letter dated 25 April 1971, at the time the news of the solution of the epimorphism problem first spread, Abhyankar was informed by G.M. Bergman that as a consequence of the commutative result the epimorphism theorem also holds in the non-commutative case. We have
Non-Commutative Epimorphism Theorem. Let \(k\langle X,Y \rangle\) and \(k\langle Z \rangle =k[Z]\) be the free associative algebras. Then the Epimorphism Theorem holds with \(k\langle X,Y \rangle\) and \(k\langle Z \rangle\) replacing \(k[Y,X]\) an \(k[Z]\).
The Expansions Paper
Curves with One Place at Infinity
A place at infinity of an affine irreducible curve with coordinate ring \(A\) and function field \(K\) is a valuation ring \(V\) of \(K/k\) which does not contain \({A}\).
We say that \(A\) has one place at infinity if there is exactly one place at infinity and moreover the residue field of the valuation ring coincides with \(k\), in other words, the valuation is “residually rational”. Important examples are what Abhyankar called polynomial curves, i.e., curves parametrized by polynomials in one variable.
For a curve defined by a polynomial \(F(Y,X)\in k[Y,X]\) a place at infinity is concretely described by what Abhyankar, tongue in cheek, called Newton’s theorem on Puiseux expansions, or “meromorphic branches” of \(F(Y,X)\). We will call them Newton-Puiseux (NP) expansions. These are substitutions \((X,Y)=(\tau^{-n}, \eta(\tau))\), where \(\eta(\tau) \in k^*((\tau))\) and (i) \(F(\eta(\tau),\tau^{-n})=0\), (ii) \(k^*\) is a finite algebraic extension of \(k\) and (iii) \(n\) is not divisible by the characteristic the gcd of \(n\) and the support of \(\eta(\tau)\) is \(1\). This defines the valuation ring \(V\) consisting of all rational functions \(g(Y,X)\) for which \(\mathrm{ord}_{\tau}g(\eta(\tau),\tau^{-n})\geq 0\).
For a plane curve defined by \(F(Y,X)\) with, say, \(\deg_Y F(Y,X)\) not divisible by the characteristic, it can be deduced that it has one rational place at infinity (sometimes shortened to just “\(F\) is a one place curve”), if and only if it has an NP expansion with \(k^*=k\) and \(n=\deg_YF(Y,X)\).
Note that a polynomial \(F\) with one place at infinity is irreducible in \(k[Y,X]\) and irreducible as well in the larger ring \(k((X^{-1}))[Y]\). It can be shown that \(F\) is pre-monic in any choice of variables (which are not absent from \(F\)). In particular, the above equivalence is not dependent on our choice of coordinates, up to possibly switching of \(X\) and \(Y\). For simplicity we will assume hereafter that \(F(Y,X)\) is monic in \(Y\) and \(\pi\nmid n=\deg_YF\).
It is a classic procedure to associate certain “characteristic sequences” to elements of \(k[[\tau]]\). The basic tool in [AM2] is a certain clever reorganization invented by Abhyankar of such sequences associated to any meromorphic power series \(\eta(\tau) \in k((\tau))\). Abhyankar used to relate that he recorded several ideas about plane curves in his personal notes of Zariski’s lectures on curves, but always presumed that he had simply learned them in Zariski’s course. He discovered that they were his own inventions only when Zariski was not aware of them. His idea of introducing the \(q\)-sequence, for instance, made the complicated formulas involved in manipulating Newton-Puiseux expansions into much simpler statements of invariance. He wrote two separate papers entitled Inversion and Invariance of Characteristic Pairs exploring the power of these techniques.
Following [A] 6.4, we give a brief outline. Let \(\eta(\tau) \in k((\tau))\) and \(n\in \mathbb N\) be given. Let \(J\) be the support of \(\eta(\tau)\) and assume that the gcd of \(J\) with \(n\) is \(1\).
Start with \(m_1=\min(J)\) and \(d_1=n\). Set \(d_2=\gcd(m_1,d_1)\). Define \(m_2\) as the first element of \(J\) not divisible by \(d_2\). Set \(d_3=\gcd(m_2,d_2)\). Continue until \(d_{h+1}=1\) (since then no \(m_{h+1}\) can be found).
This m-sequence marks the places of gcd-drops in \(J\) and the d-sequence gives the successive gcd’s.
A sequence equivalent to the \(m\)-sequence, but more useful, is the q-sequence defined by \(q_1=m_1\) and \(q_{i+1}=m_{i+1}-m_i\) thereafter.
We define the s-sequence and r-sequence by \(s_i = \sum_{j=1}^i q_j d_j \) and \(r_i=s_i/d_i\). These play a crucial role in computing intersection numbers, as exemplified by the following kind of computation which plays an important role in the Abhyankar-Moh theory:
Consider a deformed initial part of the series \(\eta(\tau)\), namely \(u(\tau) = c_1\tau^{m_1}+\cdots + Z\tau^{m_i}\), agreeing with \(\eta(\tau)\) in all terms preceding \(\tau^{m_i}\) and with \(Z\) an indeterminate. Then it is easy to calculate that the initial term of the product \(\prod_\omega \left(u(\tau)-\eta(\omega\tau)\right)\), where \(\omega\) ranges over the \(n\)-th roots of unity, is of the form \(c(Z)\tau^{s_i}\), where \(c(Z)\) is a non-zero polynomial in \(Z\).
Where necessary, we will indicate the dependence on \(\eta\) and \(n\) in these definitions by a suitable notation.
Note that each of the \(m\)-, \(q\)- and \(r\)-sequence have the associated \(d\)-sequence as sequence of successive gcd’s.
One Place and Expansions
Let \(F(Y,X)\) be a one place curve with NP expansion \((X,Y)=(\tau^{-n}, \eta(\tau))\). We get the well-known induced factorization \(F(Y,\tau^{-n}) = \prod_{j=1}^n (Y-\eta(\omega^j\tau)) \), where \(\omega\) is a primitive \(n\)-th root of unity. In particular, all roots of \(F(Y,\tau^{-n})\) have the same support and the characteristic sequences we defined depend on \(F\) only. We note that \(m_1(-n,\eta(\tau))=-\deg_XF(Y,X)\).
We mention two essential ingredients of the Abhyankar-Moh theory which are responsible for most of its successes.
{The Irreducibility Criterion of Abhyankar and Moh. Assume that \(n=\deg_{Y} {F(Y,X)}\not\equiv 0 \mod{\pi}\) and \(F(Y,X)\) is monic in \(Y\). Then \(F(Y,X) \) has one place at infinity iff there is a “test series” \(u(\tau) \in k((\tau)) \) such that \( \mathrm{ord}_{\tau}{(F(u(\tau),\tau^{-n}))} > s_h(-n,u(\tau)). \)
Moreover, given any series \(u(\tau)\) passing this test, there is a “root” \(y(\tau)\) satisfying \(F( y(\tau), \tau^{-n}) = 0 \) and \(\mathrm{ord}_{\tau}{(y(\tau)-u(\tau))}>m_h(-n,u(\tau)) \).

This Lemma, originally in [AM2], was later reproved by Abhyankar in greater detail in [A2].}
{The Innovation of the Approximate Roots. Let \(F=F(Y,X)\) have one place at infinity, arranged to be monic in \(Y\). Consider the characteristic sequences as described above for an NP expansion \((X,Y)=(\tau^{-n},\eta(\tau))\). Then for each \(d_i\), \(i=1,2,\dots , h\), we get approximate roots \(G_i(Y,X)\) defined by
For \(i=1\), \(G_1=Y\) and for \(i>1\), \(G_i(Y,X)\) is monic in \(Y\) of degree \(n/d_i\) and
\(\deg_Y(F-G_i^{d_i})< n-n/d_i\).
Such polynomials are easily seen to be uniquely defined by \(F\) for any factors of \(n\), but for the \(d_i\) chosen from the characteristic sequence, it is shown that each \(G_i(Y,X)\) is a curve with one place at infinity and that \(G_i(Y,X)\), taken mod \(F(Y,X)\), has value \(r_i\) in the valuation at infinity of \(F\).
This implies that an NP expansion of \(G_i\) matches that of \(F\) up to \(m_i\). The idea that this should happen for \(i=2\) was first put forward by Moh and perfected to the above form by the genius of Abhyankar.}

The following result is now recognized as describing a fundamental property of one place curves.
The One Place Theorem for Translates of a One Place Curve. If \(\pi=0\) and \(F\) has one place at infinity, then \(F+\lambda\) also has one place at infinity for any \(\lambda\in k\). Moreover, all translates have NP expansions that match through the last characteristic term. In geometric language, this means that \(F\) and \(F+\lambda\) go through each other at infinity at all the singular points in a sequence of quadratic transforms.
This is deduced from the irreducibility lemma and the explicit calculation of the initial forms in terms of the approximate roots. Over \(\mathbb C\) it implies that the fibration defined by \(F\) is topologically trivial in a neighborhood of infinity.
The Value Semigroup of a One Place Curve, Standard Basis
Let \(F(Y,X)\) and have one place at infinity with NP expansion \((X,Y)=(\tau^{-n}, \eta(\tau))\) and coordinate ring \(A=k[Y,X]/(F(Y,X))\). Let \(\alpha:k[Y,X]\rightarrow k[y,x]=A\) be the canonical homomorphism with \(\alpha(X)=x, \alpha(Y)=y\). The valuation \(V\) at infinity on the quotient field of \(A\) is defined by \(V(h(y,x))= \mathrm{ord}_{\tau}h(\eta(\tau),\tau^{-n})\). We define the value semigroup of \(F\) as \(\Gamma_F = \{ V(h) ~|~ 0\neq h\in A\}\). Define \(g_0=x\) and put \(r_0=V(x)=-n\). For \(i=1,\dots,n\) let \(g_i=\alpha(G_i)\), where the \(G_i\) are the approximate roots introduced above and, as we said, \(V(g_i)=r_i\). Let \(\mathfrak A\) be the set of \((h+1)\)-tuples \(a=(a_0,a_1,\dots, a_h)\) of integers with \(a_0 \geq 0\) and \(0\leq a_i < d_i/d_{i+1} \) for \(i=1,2,\dots ,h\). The following results are the technical core of the Abhyankar-Moh theory:
\(\Gamma_F=\{a_0 r_0 + a_1r_1 + \dots +a_h r_h| (a_0, \dots, a_h)\in \mathfrak A\}\). These “restricted expansions” are unique.
The “standard monomials” \(\{g^a = \prod_{i=0}^h g_i^{a_i}|a\in \mathfrak A\}\) provide a \(k\)-basis of \(A\). Note that in each linear combination over \(k\) of standard monomials there is a unique term whose valuation at infinity is the valuation of the sum.

It can be deduced from the above that if \(d_2\) belongs to \(\Gamma_F\), then \((r_0,r_1)\) is principal and hence one of \(m,n\) divides the other. We obtain a proof of the Epimorphism Theorem since in that case \(\Gamma_F\) is \(\{0,1,2,\dots \}\).
We remark that expansion techniques can also be applied to irreducible elements \(F(U,V)\) of the power series ring \(k[[U,V]]\), in particular when it is the completion of of the local ring at the point at infinity in \(\mathbb P^2\) of a one-place curve. If one does not insist that the expansion be of NP type (with one of the variables a power of the parameter \(\tau\)), the basic definitions can be made without explicit reference to the characteristic \(\pi\). This is useful in some applications, see [Ru3].
Suzuki’s Proof A proof of the Embedding Theorem contemporary with that of Abhyankar and Moh was given by M. Suzuki [Su]. It is very different in spirit, and the Embedding Theorem nowadays is often cited as the AMS-Theorem. Suzuki’s paper also was very influential. It uses methods of complex analysis, in particular the theory of pluri-subharmonic functions, to study polynomial maps \(F: \mathbb C^2\rightarrow \mathbb C\), where \(F\) is an irreducible polynomial. A key result, now usually referred to as Suzuki’s formula, is that the topological Euler characteristic of any singular (special) fiber is at least as big as that of a regular (general) fiber. (This generalizes a fact well known in the case of proper maps.) In case \(F_0=F^{-1}(0)\simeq \mathbb C\), Suzuki then goes on to show that \(F_0\) is in fact a regular fiber. His methods apply to morphisms \(\Phi:X\rightarrow \mathbb C\) for surfaces more general than \(\mathbb C^2\), in particular all smooth affine surfaces. His results have been extended and sharpened by M. Zaidenberg [Za], and a proof of Suzuki’s formula relying on geometric methods rather than complex analysis, or, let us say, more accessible to algebraic geometers, has been given by R. Gurjar [Gu2].
Positive Characteristic The counterexamples to the Embedding Theorem in positive characteristic \(\pi\) cited above left open the question whether at least the “One Place Theorem” (see 3.2) still holds. This question was raised by Abhyankar in [A]. A partially positive, partially negative answer was given by R. Ganong [Ga1]. For simplicity we assume that \(k\) is algebraically closed.
The Generic One Place Theorem. Let \(F\in k[Y,X]\) have one place at infinity. Then the generic member \(F-t\), \(t\) transcendental over \(k\), of the pencil \(F-\lambda, \lambda \in k\), has one place at infinity with residue field purely inseparable over \(k(t)\). Moreover, for almost all \(\lambda \in k\), \(F-\lambda\) has one place at infinity with multiplicity sequence at infinity the same as that of \(F-t\) over the algebraic closure of \(k(t)\).
This is a “best possible” result: Ganong also gives examples of one place curves \(F\) with with a special member having more than one place at infinity, or with general member having a multiplicity sequence at infinity different from that of \(F\). Following the lead of M. Nagata [Na1] and M. Miyanishi [Mi1], Ganong investigates special properties of the pencil obtained by eliminating the base points at infinity of the pencil \(F-\lambda\). Here the fiber at infinity is simply connected (it is a tree of curves isomorphic to the projective line), and a key result is a positive characteristic version of a lemma of Kodaira [Kod] on the global multiplicity of such fibers. We remark that in case the place at infinity of \(F-t\) is rational over \(k(t)\) (i.e., the residue field is \(k(t)\), e.g., if \(\pi =0\)), then all \(F-\lambda\) have the same multiplicity sequence and the same infinitely near multiple points at infinity. In particular, the Embedding Theorem follows if \(F\) is a line.
Lines in the plane in positive characteristic are still poorly understood. Some contributions to the question have been made by [Da2, Da3, Ga1, Mo2]. The following closely related conjectures seem to have been made by several researchers. An overview of relevant results is given in [Ga3], see also [Mi2].
Lines Conjectures in Characteristic \(\pi\).
If \(F\in k[Y,X]\) is a line, then all \(F-\lambda, \lambda \in k\) are lines.
If \(F\in k[Y,X]\) is a line, then the relative Frobenius w.r.t. \(F\) is a plane, i.e., \(k[X^{\pi}, Y^{\pi},F]\) is a polynomial ring.

Further Developments from the Expansions Paper The Jacobian Problem Inspired by their new machinery, Abhyankar and Moh produced independently several papers attacking the famous Jacobian Problem. In dimension two, it asks if polynomials \(f_1,f_2\) in the polynomial ring \(k[X,Y]\) over a field \(k\) of characteristic zero with Jacobian determinant \(1\) generate \(k[X,Y]\).
Indeed, this problem was rejuvenated and popularized by Abhyankar along with several other problems in Affine Geometry of two and three dimensions as a way to attract new students to important but accessible problems in Algebraic Geometry.
Abhyankar and Moh propose to consider \(f_1,f_2\) as elements of \(k(X)[Y]\), that is as defining a polynomial curve over \(k(X)\) with \(Y\) serving as parameter, see [A] for details. They quickly translated the Jacobian condition into conditions on the resulting NP expansion and produced the following striking result, among others:
The Two Point Theorem. The Jacobian condition implies that \(f_1,f_2\) have at most two points at infinity, i.e., their top degree forms in \(X,Y\) have at most two non-associate factors. Moreover, if it can be deduced that the Jacobian condition implies that \(f_1,f_2\) have at most one point at infinity, then the Jacobian problem has an affirmative answer.
For brevity, we stop here, but the problem has a long colorful history and many results (even in higher dimensions) are available at the touch of a key stroke!
Finiteness of Embeddings of One Place Curves One way of stating the epimorphism theorem is to say that there is only one equivalence class of embeddings of an affine line in the affine plane up to automorphisms of the plane. Abhyankar raised the corresponding question for general one place curves: Suppose that \(\alpha, \beta\) are two epimorphisms from \(k[Y,X]\) onto the coordinate ring \(A\) of a plane curve with one rational place at infinity. Does it follow that \(\alpha\) and \(\beta\) are equivalent? If not, is it at least true that the number of equivalence classes is finite?
Let \(F\) be a generator of \(\ker(\alpha)\), say. We can then arrange by an automorphism of \(k[X,Y]\) that \((r_0,r_1)=(-\deg_YF, -\deg_XF)\) is non-principal. This gives that \(d_2=\gcd(r_0,r_1)\) is a number not in the value semigroup, by the non principal condition (see 3.3). In [ASi] Abhyankar and Singh prove the following striking result: Two embeddings are equivalent if and only if the corresponding \(d_2\)’s are equal. This, combined with the fact that there are only finitely many negative numbers not in the value-semigroup, gives the finiteness of embeddings with a very explicit bound on the number. (We have tacitly assumed \(\pi=0\). Otherwise even the line has infinitely many inequivalent embeddings [Ga1].)
Planar Semigroups The properties of the characteristic \(r\)-sequence in the Abhyankar-Moh theory can be codified abstractly and semigroups generated by an \(r\)-sequence as in 3.3 have been called planar semigroups by Sathaye. (Sathaye actually preferred to work with the negative of an \(r\)-sequence.) It was announced in [Sa2] and shown in [SS] that every planar semi-group is the value semi-group of a one place curve. See [A2] as well. The irreducibility criterion 3.2.1 plays a significant role here.
An important question, originally raised by Abhyankar himself, is to characterize the semi-groups at infinity, or equivalently the degree semigroups, of plane polynomial curves. This remains unsolved to date. For further calculations and conjectures about these, see [SS], [SFY] and [M-L]. An interesting special case is the Lin-Zaidenberg Theorem [LZ] which asserts that there is only one class for polynomial curves with only unibranch singularities. (The theorem proves more, namely that such curves have only one quasi-homogeneous singularity and that \(d_2=1\).)
A Sampling of Further Related Results The Epimorphism Theorem inspired a large amount of research on related questions. We can only give a brief sampling. Coefficient rings more general than fields were considered [Ba], [RS], [Ve]. More general closed embeddings of affine \(m\)-space \(\mathbb A^m\) in affine \(n\)-space \(\mathbb A^n\) were studied [Kal1], [Sr]. Abhyankar in particular advocated the study of embeddings of lines and planes in affine \(3\)-space. For some results on lines see [A3], [AS], [Cr], [BR], [Sh], and for planes see [Sa1], [Ru2], [Wr], [San]. These papers depend in a crucial way on the Epimorphism Theorem. An unexpected generalization of the Epimorphism Theorem was developed by Sathaye in [Sa3], [Sa4] and [Sa5]. It became an important tool in several studies of \(\mathbb A^2\)-fibrations over curves. The Epimorphism Theorem also is the essential ingredient in the proof of a special case of the linearization conjecture for \(\mathbb C^*\)-actions on \(\mathbb C^3\), see [KKM-LR].
It was a suggestion coming out of Abhyankar’s Purdue seminar to consider closed curves in the affine plane with several places at infinity. A once punctured affine line, \(\mathbb C^*\) when \(k=\mathbb C\), is the obvious first candidate, see [BZ], [C-NKR], [Kal2], [Ko2] for results in this case. Another suggestion was to investigate “field generators” (instead “ring generators” as in the Epimorphism Theorem), that is, polynomials \(F\) that together with a complementary rational function \(G\) generate the field \(k(X,Y)\), see [Ja], [NN], [MS], [Ru1], [Da1].
The Epimorphism Theorem gave a strong boost to Affine Algebraic Geometry, the study of algebraic varieties closely related to affine spaces, in particular affine rational surfaces. The study of affine lines on such surfaces became an important part of their classification via logarithmic Kodaira dimension [It], [GMMR], [GM1], [KK]. In turn, the classification theory has been used to prove the Epimorphism Theorem [GM2], [Gu1], [Ko1].
[A] S. S. Abhyankar, Lectures on expansion techniques in algebraic geometry. With notes by Balwant Singh. Bombay: Tata Institute of Fundamental Research (1977; Zbl 0818.14001).
[A2] S. S. Abhyankar, “Irreducibility criterion for germs of analytic functions of two complex variables”, Adv. Math. 74, No. 2, 190–257 (1989; Zbl 0683.14001).
[A3] S. S. Abhyankar, Algebraic space curves. Edited by P. Russell and A. Sathaye. Montreal: Les Presses de l’Universite de Montreal (1971; Zbl 0245.14009).
[AM1] S. S. Abhyankar and T.-T. Moh, “Embeddings of the line in the plane”, J. Reine Angew. Math. 276, 148–166 (1975; Zbl 0332.14004).
[AM2] S. S. Abhyankar and T.-T. Moh, “Newton-Puiseux expansion and generalized Tschirnhausen transformation. I. II.”, J. Reine Angew. Math. 260, 47–83 (1973); ibid. 261, 29–54 (1973; Zbl 0272.12102).
[ASi] S. S. Abhyankar and B. Singh “Embeddings of certain curves in the affine plane”, Am. J. Math. 100, 99–175 (1978; Zbl 0383.14007).
[AS] S. S. Abhyankar and A. Sathaye, Geometric theory of algebraic space curves. Berlin-Heidelberg-New York: Springer-Verlag (1974; Zbl 0357.14008).
[Ba] S. M. Bhatwadekar, “Generalized epimorphism theorem”, Proc. Indian Acad. Sci., Math. Sci. 98, No. 2–3, 109–116 (1988; Zbl 0673.13014).
[BR] S. M. Bhatwadekar and A. Roy, “Some results on embedding of a line in 3-space”, J. Algebra 142, No. 1, 101–109 (1991; Zbl 0782.14011).
[BZ] M. Borodzik and H. Żołądek, “Complex algebraic plane curves via Poincaré-Hopf formula. II. Annuli”, Isr. J. Math. 175, 301–347 (2010; Zbl 1202.14031).
[Cr] P. C. Craighero, “On Abhyankar’s conjectures on space lines”, Rend. Sem. Mat. Univ. Padova 74, 115–122 (1985; Zbl 0594.14025).
[C-NKR] P. Cassou-Nogues, M. Koras and P. Russell, “Closed embeddings of \(\mathbb C^*\) in \(\mathbb C^2\). I.”, J. Algebra 322, No. 9, 2950–3002 (2009; Zbl 1218.14019).
[Da1] D. Daigle, “Birational endomorphisms of the affine plane”, J. Math. Kyoto Univ. 31, No. 2, 329–358 (1991; Zbl 0766.14006).
[Da2] D. Daigle, “A property of polynomial curves over a field of positive characteristic”, Proc. Am. Math. Soc. 109, No. 4, 887–894 (1990; Zbl 0711.14007).
[Da3] D. Daigle, “On pencils of polynomial curves”, J. Pure Appl. Algebra 111, No. 1–3, 51–57 (1996; Zbl 0856.13016).
[Ga1] R. Ganong, “On plane curves with one place at infinity”, J. Reine Angew. Math. 307–308, 173–193 (1979; Zbl 0398.14004).
[Ga3] R. Ganong, “The pencil of translates of a line in the plane”, CRM Proceedings and Lecture Notes 54, 57–71 (2011; Zbl 1263.14057).
[Gu1] R. V. Gurjar, “A new proof of the Abhyankar-Moh-Suzuki theorem”, Transform. Groups 7, No. 1, 61–66 (2002; Zbl 1054.14081).
[Gu2] R. V. Gurjar, “A new proof of Suzuki’s formula”, Proc. Indian Acad. Sci., Math.Sci. 107, No. 3, 237–242 (1997; Zbl 0920.14007).
[GM1] R. V. Gurjar and M. Miyanishi, “Algebraic surfaces with \(\overline \kappa \leq 1\)”, Kinokuniya, Tokyo, 99–124 (1988).
[GM2] R. V. Gurjar and M. Miyanishi, “On contractible curves in the complex affine plane”, Tôhoku Math. J., II. Ser. 48, No. 3, 459–469 (1996; Zbl 0876.14018).
[GMMR] R. V. Gurjar, K. Masuda, M. Miyanishi and P. Russell, “Affine lines on affine surfaces and the Makar-Limanov invariant”, Can. J. Math. 60, No. 1, 109–139 (2008; Zbl 1137.14049).
[It] S. Iitaka, “On logarithmic Kodaira dimension of algebraic varieties”, Complex Anal. algebr. Geom., Collect. Pap. dedic. K. Kodaira, 175–189 (1977; Zbl 0351.14016).
[J] H. W. E. Jung, “Über ganze birationale Transformationen der Ebene”, J. Reine Angew. Math. 184, 161–174 (1942; Zbl 0027.08503; JFM 68.0382.01).
[Ja] C.-J. Jan, “On polynomial field generators of \(k(x,y)\)”, Thesis, Purdue University (1974).
[Kal1] S. Kaliman,“Extensions of isomorphisms between affine algebraic subvarieties of \(k^n\) to automorphisms of \(k^n\)”, Proc. Am. Math. Soc. 113, No. 2, 325–334 (1991; Zbl 0743.14011).
[Kal2] S. Kaliman, “Rational polynomials with a \(\mathbb C^*\)-fiber”, Pac. J. Math. 174, No. 1, 141–194 (1996; Zbl 0868.32010).
[Kod] K. Kodaira, “On compact analytic surfaces. II.”, Ann. Math. (2) 77, 563–626 (1963; Zbl 0118.15802).
[Ko1] M. Koras, “On contractible plane curves”, Affine algebraic geometry. Osaka: Osaka University Press. 275–288 (2007; Zbl 1126.14070).
[Ko2] M. Koras, “\(\mathbb C^*\) in \(\mathbb C^2\) is birationally equivalent to a line”, CRM Proceedings and Lecture Notes 54, 165–191 (2011; Zbl 1252.14042).
[KK] T. Kishimoto and H. Kojima, “Affine lines on \(\mathbb Q\)-homology planes with logarithmic Kodaira dimension \(-\infty\)”, Transform. Groups 11, No. 4, 659–672 (2006; Zbl 1118.14064).
[KKM-LR] S. I. Kaliman, M. Koras, L. Makar-Limanov and P. Russell, “\(\mathbb C^*\)-actions on \(\mathbb C^3\) are linearizable”, Electron. Res. Announc. Am. Math. Soc. 3, 63–71 (1997; Zbl 0890.14026).
[LZ] V. Y. Lin and M. G. Zaidenberg, “An irreducible, simply connected algebraic curve in \(\mathbb C^2\) is equivalent to a quasihomogeneous curve”, Sov. Math., Dokl. 28, 200–204 (1983); translation from Dokl. Akad. Nauk SSSR 271, 1048–1052 (1983; Zbl 0564.14014).
[M-L] L. Makar-Limonov, “A new proof of the Abhyankar-Moh-Suzuki Theorem”, Preprint 2013.
[Mi1] M. Miyanishi, “Analytic irreducibility of certain curves on a nonsingular affine rational surface”, Proc. int. Symp. on algebraic geometry, Kyoto 1977, 575–587 (1977; Zbl 0417.14024).
[Mi2] M. Miyanishi, “Frobenius sandwiches of affine algebraic surfaces”, CRM Proceedings and Lecture Notes 54, 243–260 (2011; Zbl 1247.14065).
[Mo1] T.-T. Moh, “On analytic irreduciblity at \(\infty\) of a pencil of curves”, Proc. Am. Math. Soc. 44, 22–24 (1974; Zbl 0309.14011).
[Mo2] T.-T. Moh, “On the classification problem of embedded lines in characteristic p”, Algebraic geometry and commutative algebra, in Honor of Masayoshi Nagata, Vol. I, 267–279 (1988; Zbl 0701.14010).
[MS] M. Miyanishi and T. Sugie, “Generically rational polynomials”, Osaka J. Math. 17, No. 2, 339–362 (1980; Zbl 0457.13006).
[Na1] M. Nagata, “A theorem of Gutwirth”, J. Math. Kyoto Univ. 11, 149–154 (1971; Zbl 0213.23201).
[Na2] M. Nagata, “On automorphism group of \(k[x,y]\)”, Tokyo: Kinokuniya Book-Store Co., Ltd. (1972; Zbl 0306.14001).
[NN] W. D. Neumann and P. Norbury, “Rational polynomials of simple type”, Pac. J. Math. 204, No. 1, 177–207 (2002; Zbl 1055.14062).
[Ru1] P. Russell, “Field generators in two variables”, J. Math. Kyoto Univ. 15, 555–571 (1975; Zbl 0341.14008).
[Ru2] P. Russell, “Simple birational extensions of two dimensional affine rational domains”, Compos. Math. 33, 197–208 (1976; Zbl 0342.13003).
[Ru3] P. Russell, “Hamburger-Noether expansions and approximate roots of polynomials”, Manuscr. Math. 31, 25–95 (1980; Zbl 0455.14018).
[RS] P. Russell and A. Sathaye, “On finding and cancelling variables in \(k[X,Y,Z]\)”, J. Algebra 57, 151–166 (1979; Zbl 0411.13011).
[San] L. M. Sangines Garcia, “On quadratic planes”, Thesis (Ph.D.) McGill University (Canada) (1983).
[Sa1] A. Sathaye, “On linear planes”, Proc. Am. Math. Soc. 56, 1–7 (1976; Zbl 0345.14013).
[Sa2] A. Sathaye, “On planar curves”, Am. J. Math. 99, 1105–1135 (1977; Zbl 0386.14017).
[Sa3] A. Sathaye, “Generalized Newton-Puiseux expansion and Abhyankar-Moh semigroup theorem”, Invent. Math. 74, 149–157 (1983; Zbl 0549.14001).
[Sa4] A. Sathaye, “Polynomial rings in two variables over a D.V.R.: a criterion”, Invent. Math. 74, 159–168 (1983; Zbl 0538.13006).
[Sa5] A. Sathaye, “An application of generalized Newton Puiseux expansions to a conjecture of D. Daigle and G. Freudenburg”, Algebra, arithmetic and geometry with applications. Papers from Shreeram S. Ahhyankar’s 70th birthday conference, Purdue University, West Lafayette, IN, USA, July 19–26, 2000. Berlin: Springer. 687–701 (2003; Zbl 1154.14324).
[SS] A. Sathaye and J. Stenerson, “Plane polynomial curves”, Algebraic geometry and its applications. Collections of papers from Shreeram S. Abhyankar’s 60th birthday conference held at Purdue University, West Lafayette, IN, USA, June 1-4, 1990. New York: Springer-Verlag. 121–142 (1994; Zbl 0809.14022).
[Se] B. Segre, “Corrispondenze di Möbius e transformazioni cremoniane intere”, Atti Acad. Sci. Torino Cl. Fis. Mat. Nat. 91, 319 (1956/1957).
[Sh] A. R. Shastri, “Polynomial representations of knots”, Tôhoku Math. J., II. Ser. 44, No. 1, 11–17 (1992; Zbl 0743.57006).
[Sr] V. Srinivas, “On the embedding dimension of an affine variety”, Math. Ann. 289, No. 1, 125–132 (1991; Zbl 0725.14003).
[SFY] M. Fujimoto, M. Suzuki and K. Yokoyama, “On polynomial curves in the affine plane”, Osaka J. Math. 43, No. 3, 597–608 (2006; Zbl 1105.14042).
[Su] M. Suzuki, “Propriét’es topologiques des polynômes de deux variables complexes, et automorphismes algébriques de l’espace \(\mathbb C^2\)”, J. Math. Soc. Japan 26, 241–257 (1974; Zbl 0276.14001).
[Ve] S. Vénéreau, “New bad lines in \(R[x,y]\) and optimization of the epimorphism theorem”, J. Algebra 302, No. 2, 729–749 (2006; Zbl 1107.14051).
[VdK] W. van der Kulk, “On polynomial rings in two variables”, Nieuw Arch. Wiskd., III. Ser. 1, 33–41 (1953; Zbl 0050.26002).
[Wr] D. Wright, “Cancellation of variables of the form \(bT^n-a\)”, J. Algebra 52, 94–100 (1978; Zbl 0383.13014).
[Za] M. Zaidenberg, “Isotrivial families of curves on affine surfaces and the characterization of the affine plane”, Math. USSR, Izv. 30, No. 3, 503–532 (1988); translation from Izv. Akad. Nauk SSSR, Ser. Mat. 51, No. 3, 534–567 (1987; Zbl 0666.14018); “Additions and corrections”, Math. USSR, Izv. 38, No. 2, 435–437 (1992); translation from Izv. Akad. Nauk SSSR, Ser. Mat. 55, No. 2, 444–446 (1991; Zbl 0749.14019).


14R10 Affine spaces (automorphisms, embeddings, exotic structures, cancellation problem)
14R15 Jacobian problem
14E25 Embeddings in algebraic geometry
14M07 Low codimension problems in algebraic geometry
14M20 Rational and unirational varieties
14-03 History of algebraic geometry
01A60 History of mathematics in the 20th century
01A61 History of mathematics in the 21st century
Full Text: Crelle EuDML