Ssmart Little Black Boxes

1
Krone ker's smart, little bla k boxes
Mar Giusti
UMS CNRS{Polyte hnique MEDICIS
Laboratoire GAGE

E ole Polyte hnique
F-91128 Palaiseau edex, Fran e
Email: Mar .Giustigage.polyte hnique.fr
Joos Heintz
Depto. de Matem ati as, Est. y Comp.
Fa ultad de Cien ias Universidad de Cantabria
E-39071 Santander, Spain and
Depto. de Matem ati as
Fa ultad de Cien ias Exa tas y Naturales
Universidad de Buenos Aires
Ciudad Universitaria, Pab. I
(1428) Buenos Aires, Argentina
Email: heintzhall.mates o.uni an.es
Abstra t
This paper is devoted to the omplexity analysis of ertain uniformity
properties owned by all known symboli methods of parametri poly-
nomial equation solving (geometri elimination). It is shown that any
parametri elimination pro edure whi h is parsimonious with respe t
to bran hings and divisions must ne essarily have a non-polynomial se-
quential time omplexity, even if highly eÆ ient data stru tures (as e.g.
the arithmeti ir uit en oding of polynomials) are used.
1 Introdu tion
Origins, development and intera tion of modern algebrai geometry and
ommutative algebra may be onsidered as one of the most illustra-
tive examples of histori al diale ti s in mathemati s. Still today, and
more than ever before, timeless idealism (in form of modern ommuta-
Resear h partially supported by the following Argentinian, Spanish, German and
Fren h grants : UBA-CYT TW 80, PIP CONICET 4571/96, ANPCyT PICT 03-
00000-01593, DGICYT PB96{0671{C02{02, ARG 018/98 INF (BMBF), UMS 658
MEDICIS, ECOS A99E06, HF 1999{055
1
2 Giusti and Heintz
tive algebra) is bravely struggling whith se ular materialism (in form of
omplexity issues in omputational algebrai geometry).
Krone ker was doubtless the reator of this eternal battle eld and its
rst war lord. In a similar way as Gauss did for omputational num-
ber theory, Krone ker laid intuitively the mathemati al foundations of
modern omputer algebra. He introdu ed 1882 in [24℄ his famous "elim-
ination method" for polynomial equation systems and his "parametri
representation" of (equidimensional) algebrai varieties. By the way, this
parametri representation was until 10 years ago redis overed again and
again. It entered in modern omputer algebra as "Shape Lemma" (see
e.g. [36, 7, 10, 22℄). Using his elimination method in a highly skillful,
but unfortunately inimitable way, Krone ker was able to state and to
prove a series of fundamental results on arbitrary algebrai varieties. He
was able to dene in a pre ise way the notion of dimension and to prove
a orresponding dimension theorem for arbitrary algebrai varieties over
an algebrai ally losed eld, to estimate the number of equations needed
to dene any algebrai variety in aÆne or proje tive spa e and ertainly
he knew already the spe ial form of \Hilbert's Nullstellensatz".
Not everything that ame to Krone ker's mind was laid down by him
in a expli it and written form. Nevertheless a areful interpretation of
his work suggests his deep understanding of the general stru ture of
algebrai varieties.
A parti ular result, proved by Krone ker, says that any algebrai vari-
ety an be dened by nitely many equations. Later Hilbert generalized
this result to his seminal \Basissatz" introdu ing for its proof a new,
non onstru tive method, far away from the traditional elimination-type
arguments used by Krone ker and other ontemporary mathemati ians.
It took some time to onvin e the mathemati al world that \mysti s"
and \magi ians" are able to produ e ( orre t) mathemati al results, but
nally the new dis ipline of ommutative algebra be ame legitimate.
Hilbert's dis overy of the Basissatz was also the starting point for
a long and huge on i t whi h dominated a onsiderable part of the
history of modern algebrai geometry and whi h did not ome to an end
until today.
Classi al algebrai geometry is motivated by the need { or the wish {
to nd tools whi h allow to \solve" or to \redu e" (whatever this means)
systems of polynomial equations. This leads to the following questions:
are ommutative algebra and its modern derivate, namely todays

Krone ker's bla k boxes 3
s heme-theoreti al algebrai geometry, able to absorb lassi al alge-
brai geometry?
is the solution to Hilbert's \Hauptproblem der Idealtheorie" (the ideal
membership problem) the key to all omputational issues in lassi al
algebrai geometry?
These questions, impli itly raised by the work of Hilbert and Ma au-
lay, look very a ademi . However any attempt to answer them leads to
deep onsequen es.
Let us here outline just one possible way to answer these questions.
If we onsider the \Hauptproblem der Idealtheorie" as a question of
pure theoreti al omputer s ien e, this problem turns out to be ompu-
tationally intra table, at least in worst ase. More pre isely, the \Haupt-
problem der Idealtheorie" turns out to be omplete in exponential (mem-
ory) spa e (see [33, 43, 8℄). On the other hand, almost all of the most
fundamental problems of omputational lassi al algebrai geometry are
proved to be solvable in polynomial spa e (see e.g. [32, 9, 30, 29, 25℄).
Thus omputational omplexity is able to distinguish between geometry
and algebra and supports the viewpoint of Krone ker (geometry) against
the viewpoint of Hilbert and Ma aulay (algebra).
It is well known that Krone ker's personality was highly on i tive
for his time. It is less known how mu h posthumous reje tion Kro-
ne ker's personality was able to produ e. Hilbert's writings are eloquent
in this point (\die Krone kers he Verbotsdiktatur", see [21℄), whereas
Ma aulay's atta k against Krone ker's work on elimination was a rather
well edu ated one (see [28℄, Prefa e). In this ontext let us also remind
Andre Weil's \elimination of the elimination theory". Krone ker's radi-
al spirit did not re ognise limitations. His radi alism was as universal
as his spirit was. Of ourse he was right requiring that any mathemati-
al re e tion has to end up with nitely many and pra ti ally realizable
omputations whi h settle the on rete (e.g. appli ation) problem un-
der onsideration. However de laring the natural numbers as the only
mathemati al obje ts reated by god for mankind and de laring the
\rest"(real, omplex numbers, innite ardinals and ordinals) as devils
work, tempting humans to play with the innite, he demonstrated that
he was not able or willing to distinguish between syntaxis and seman-
ti s. He was right to require that mathemati al expressions (algebra)
have always to move within nitary limits, but he was wrong to ex lude
re e tions about innite mathemati al obje ts (geometry) using a ni-
4 Giusti and Heintz
tarian language. Of ourse his own work ended up to be godless enough
to guarantee him mathemati al re ognition by his worst adversaries.
Krone ker's own formulation of his elimination method in [24℄ was
impre ise and general enough to allow dierent interpretations of it,
omputationally eÆ ient and ineÆ ient ones. Possibly it was in his
mind to leave open the door for future omplexity issues of his method.
Hilbert's and Ma aulay's atta ks against Krone ker's method were
based on the omputationally ineÆ ient interpretation. They noti ed
that under this interpretation Krone ker's elimination method leads to
a hyperexponential swell of intermediate expression size and used this
observation for the promotion of their own, more \simple" and \mathe-
mati al" point of view. Hilbert and Ma aulay's position be ame nally
predominant in the future development of algebrai geometry and om-
mutative algebra (see e.g. [40℄). Their ideas led to the modern, omputer
implemented tool of Grobner basis algorithms for the symboli resolu-
tion (simpli ation) of polynomial equation systems. This tool was in-
trodu ed in the sixties by B. Bu hberger (and his s hool) and represents
today the ore of all urrent omputer algebra software pa kages.
The dis overy of ee tive (aÆne) Nullstellensatze and their appli a-
tion to the omplexity analysis of Grobner basis algorithms for the so-
lution of geometri al problems, represented at the end of the eighties
the turning point for a pro ess whi h led nally ba k to Krone ker's
original ideas. This pro ess was not a ons ious motion with lear goals,
but rather a slow emerging of mathemati al insight and of algorithmi
design within the s ope of Krone ker's intuitions.
The rst step in this dire tion was even made as part of a mathemat-
i al proof and not of an algorithmi design. In its standard interpre-
tation, Krone ker's elimination method relies on an iterated use of re-
sultants of suitable univariate polynomials with parametri oeÆ ients.
In fa t, a resultant is nothing but the onstant term of the hara teris-
ti polynomial of a suitable linear map determined by the polynomials
under onsideration. Repla ing in this version of Krone ker's elimina-
tion algorithm the o uring resultants just by the onstant terms of the
orresponding minimal polynomials, one obtains an enormous redu -
tion of degree, height and also of arithmeti ir uit (straight-line pro-
gram) omplexity for the polynomials produ ed during the pro edure
(see e.g. [4, 5, 16℄). By the way, let us remark that Krone ker applies
in his method the mentioned simpli ation, however he omits to draw
these important on lusions from his argument.
Other important ingredients of the emerging new algorithmi method
were the parametri representation of equidimensional algebrai vari-
eties (a redis overy of Krone ker's old idea) and the arithmeti ir uit
(straight-line program) representation of polynomials (whi h was nei-
ther out of the s ope of Krone ker's intuition nor learly in luded in the
main stream of his thinking). The ombination of these ingredients pro-
du ed a onsiderable ee t upon the worst ase omplexity of symboli
elimination pro edures (see e.g. [12, 23, 14℄).
Nevertheless all this progress did not suÆ e to allow the design of
pra ti ally eÆ ient symboli elimination algorithms and in parti ular
of algorithms whi h were able to ompete in omplexity aspe ts with
their numeri al ounterparts. The traditional huge gap between sym-
boli and numeri polynomial equation solving methods remained open.
Whereas numeri algorithms are very eÆ ient with respe t to the num-
ber of arithmeti al operations they require, they annot be eÆ iently
used in ase of parametri , underdetermined, overdetermined or degen-
erate polynomial equation systems. Symboli algorithms are free from
these restri tions but they are also too ineÆ ient for any reasonable use
in pra ti e. A way out of this dilemma be ame apparent in [13, 11℄ by
a new interpretation of Newton's lassi al method. Interpreting New-
ton's approximation algorithm as a global pro edure instead as a lo al
one, allowed its use as a tool for data ompression in the Krone ker{like
elimination pro edure of [14℄ ( whi h relied on the arithmeti ir uit rep-
resentation of polynomials). It turns out that Newton's method is well
adapted to exa t symboli omputation if the orre t (seminumeri al)
data stru ture is used. In this way a new algorithmi method nally
emerged. This method is based on a ombination of Krone ker's and
Newton's ideas, and is able to distinguish di hotomi ally between \well
behaved" and \badly behaved" polynomial equations systems. Moreover
this algorithmi method is optimal for worst ase (i.e. generi ) systems.
Unlike Grobner basis algorithms this new method avoids any signif-
i ant omputational over ow during its exe ution. Roughly speaking,
the new algorithms are always polynomial in the output size and even
polynomial in the input size, if the given polynomial equation system is
\well behaved".
This view of algorithmi algebrai geometry produ ed also the follow-
ing new insight:
elimination polynomials are always \smart"(i.e. not easy and not
hard) to evaluate. How many variables they ever may ontain, their eval-
uation omplexity is always polynomial in their degree (whereas their
6 Giusti and Heintz
number of monomials may be exponential in the number of their vari-
ables).
Although this insight is beyond of the s ope of Krone ker's way of
thinking, his formulation of the elimination method indi ates that he
might have intended to leave the door open for omplexity issues.
A further development of this new algorithmi method was able to
demonstrate its pra ti al eÆ ien y (see [27℄ and [31℄). Moreover it
turns out that the parameters whi h dominate the omplexity of the
new symboli pro edures determine also the eÆ ien y of their numeri al
ounterparts (at least if aspe ts of diophantine approximation are taken
into a ount; see [34℄). The new symboli algorithms (and their numeri-
al ounterparts) have still a worst ase omplexity whi h is exponential
in the (purely synta ti al) input length. One may ask whether this fa t
is due to the algorithms and data stru tures employed or whether this
is due to the intrinsi nature of geometri elimination.
The new results presented in this paper address this question. In
order to dis uss this point, let us turn ba k to our interpretation of
Krone ker's ideas behind geometri elimination. Krone ker's elimina-
tion method (and theory) behaves well under spe ialisation of the input
equations. In fa t, at any moment of the pro edure one may onsider
the (parametri ) input equations as given by their oeÆ ients and these
oeÆ ients may be onsidered as purely algebrai obje ts, determined
only by their algebrai relations. A given equation system may even be
\generalized", i.e. the oeÆ ients of the given input equations may be
repla ed by indeterminates and the dis ussion of the solvability of the
generalized systems answers all imaginable questions about the solvabil-
ity of the original system. This on ept of spe ialisation{generalisation
reveals an idea of universality (or uniformity) behind Krone ker's elimi-
nation theory. This philosophi al idea of universality be ame one of the
orner stones of modern algebrai geometry and ommutative algebra.
Sin e we are (still) unable to think in a dierent way, we shall onsider
the input equations of a parametri elimination problem as fun tions
whi h may be alled by their values in the variables to be eliminated
(bla k-box representation) or simply as being given by their oeÆ ients
(formal representation).
Any elimination algorithm we are able to imagine today starts from
this kind of data. In other words, the bla k box representation onsti-
tutes today the most general way we may think the equations of our
input system to be given. Of ourse ea h evaluation of the input equa-
tions has its osts and these osts may be measured by the size of a
division-free arithmeti ir uit whi h represents the input equations.
In this paper we shall show that any suÆ iently uniform elimination
pro edure (whi h avoids super uous bran hings) be omes ne essarily
exponential in worst ase, if the input equations are given in bla k-box
representation and if the required output is a anoni al elimination poly-
nomial ( a \resolvent" in the terminology of Krone ker and Ma aulay).
This is a general and provable fa t for any symboli as well as for any
numeri al elimination pro edure. We shall also show, that even in ase
that the input equations are not given by a bla k box, but by an expli it
arithmeti ir uit, the same on lusion holds true for any suÆ iently
universal and uniform elimination pro edure whi h is able to ompute
eÆ iently Zariski losures and (generi ally squarefree) parametri great-
est ommon divisors for ir uit represented algebrai families of polyno-
mials.
Below we shall give a pre ise denition of the meaning of a \suÆ iently
universal and uniform elimination pro edure" for the ase of a at family
of elimination problems. Su h an elimination pro edure will be alled
parametri .
If an elimination pro edure is used in order to assign \ oordinates"
(i.e. in order to parametrise) suitable and easy-to-represent families of
algebrai varieties the same on lusion holds again, without any further
restri tion on the input representation (see [6℄ and [18℄). Summing
up, we may say: the vision of elimination theory initiated by Krone ker
hides a on ept of universality and uniformity whi h obstru ts its general
eÆ ien y. The question what happens with omplexity when we drop
this universality and uniformity requirement, ex eeds the horizon of to-
days mathemati al thinking and is equivalent to the question whether
P 6= NP holds over the omplex numbers (in the sense of the BSS
omplexity model; see [2℄).
2 Parametri elimination
2.1 Parametri elimination pro edures
The pro edures (algorithms) onsidered in this paper operate with essen-
tially division-free arithmeti ir uits (straight-line programs) as basi
data stru tures for the representation of inputs and outputs. Su h a
ir uit depends on ertain input nodes, labelled by indeterminates over
a given ground eld k (in the sequel we shall suppose that k is innite
8 Giusti and Heintz
and perfe t with algebrai losure k). These indeterminates are thought
to be subdivided into two disjoint sets representing the parameters and
variables of the given ir uit. The ir uit nodes of indegree zero whi h
are not inputs are labelled by elements of k, whi h are alled the s alars
of the ir uit (here "indegree" means the number of in oming edges of
the orresponding node). Internal nodes are labelled by arithmeti oper-
ations (addition, subtra tion, multipli ation and division). The internal
nodes of the ir uit represent polynomials in the variables of the ir uit.
The oeÆ ients of these polynomials belong to the parameter eld K ,
generated over the ground eld k by the parameters of the ir uit. This
is a hieved by allowing in a essentially division-free ir uit only divisions
whi h involve elements of K . Thus essentially division-free ir uits do
not ontain divisions involving intermediate results whi h depend on
variables. A ir uit whi h ontains only divisions by nonzero elements
of k is alled totally division-free. The output nodes of an essentially
division-free ir uit may o ur labelled by sign marks of the form \= 0"
or \6= 0" or may remain unlabelled (by sign marks). Thus the given
ir uit represents by means of its labelled output nodes a system of
polynomial equations and inequations whi h determine in their turn a
lo ally losed set (i.e. an embedded aÆne variety) with respe t to the
Zariski topology of the aÆne spa e of parameter and variable instan es.
The unlabelled output nodes of the given ir uit represent a polyno-
mial appli ation (in fa t a morphism of algebrai varieties) whi h maps
this lo ally losed set into a suitable aÆne spa e. We shall interprete
the system of polynomial equations and inequations represented by the
ir uit as a parametri family of systems in the variables of the ir uit.
The orresponding varieties onstitute an parametri family of varieties.
The same point of view is applied to the morphism determined by the
unlabelled output nodes of the ir uit. We shall onsider this morphism
as a parametri family of morphisms.
To a given essentially division-free arithmeti ir uit we may asso-
iate dierent omplexity measures and models. In this paper we shall
be mainly on erned with sequential omputing time, measured by the
size of the ir uit. O asionally we will also refer to parallel time, mea-
sured by the depth of the ir uit. In our main omplexity model is the
total one, where we take into a ount all arithmeti operations (addi-
tions, subtra tions, multipli ations and possibly o uring divisions) at
unit osts. For purely te hni al reasons we shall also onsider two non-
s alar omplexity models, one over the ground eld k and the other one
over the parameter eld K . In the non-s alar omplexity model over
K we ount only the essential multipli ations (i.e. multipli ations be-
tween intermediate results whi h a tually involve variables and not only
parameters). This means that K -linear operations (i.e. additions and
multipli ations by arbitrary elements of K ) are ost free. Similarly, k-
linear operations are not ounted in the non-s alar model over k. For
more details about omplexity measures and models we refer to [3℄.
Given an essentially division-free arithmeti ir uit as input, an elimi-
nation problem onsists in the task of nding an essentially division-free
output ir uit whi h des ribes the Zariski losure of the image of the
morphism determined by the input ir uit. The output ir uit and the
orresponding algebrai variety are also alled a solution of the given
elimination problem. We say that a given parameter point xes an in-
stan e of the elimination problem under onsideration. In this sense
a problem instan e is des ribed by an input and an output (solution)
instan e.
In this paper we restri t our attention to input ir uits whi h are
totally division-free and ontain only output nodes labelled by \=0" and
unlabelled output nodes. Mostly our output ir uits will also be totally
division-free and will ontain only one output node, labelled by the mark
\=0". This output node will always represent a anoni al elimination
polynomial asso iated to the elimination problem under onsideration
(see Se tion 2.2 for more details).
In ase that our output ir uit ontains divisions (depending only on
parameters but not on variables), we require to be able to perform these
divisions for any problem instan e. In order to make this requirement
sound, we admit in our algorithmi model ertain limit pro esses in the
spirit of de l'Hôpital's rule (below we shall modelise these limit pro-
esses algebrai ally, in terms of pla es and valuations). The restri tion
we impose on the possible divisions in an output ir uit represents a
rst fundamental geometri uniformity requirement for our algorithmi
model.
An algorithm whi h solves a given elimination problem may be on-
sidered as a (geometri ) elimination pro edure. However this simple
minded notion is too restri tive for our purpose of showing lower om-
plexity bounds for elimination problems. It is thinkable that there exists
for every individual elimination problem an eÆ ient ad ho algorithm,
but that there is no universal way to nd and to represent all these ad ho
pro edures. Therefore, a geometri elimination pro edure in the sense of
this paper will satisfy ertain uniformity and universality requirements
whi h we are going to explain now.
10 Giusti and Heintz
We modelise our elimination pro edures by families of arithmeti net-
works (also alled arithmeti -boolean ir uits) whi h solve entire
lasses of elimination problems of arbitrary input size (see [41℄, [42℄).
In this sense we shall require the universality of our geometri elimina-
tion pro edures. Moreover, we require that our elimination pro edures
should be essentially division-free.
In a universal geometri elimination pro edure, bran hings and di-
visions by intermediate results (that involve only parameters, but not
variables) annot be avoided. From our elimination pro edures we shall
require to be parsimonious with respe t to bran hings (and divisions).
In parti ular we shall require that our elimination pro edures do not in-
trodu e bran hings and divisions for the solution of a given elimination
problem when traditional algorithms do not demand this (an example of
su h a situation is given by the at families of elimination problems we
are going to onsider in the sequel). This restri tion represents a se ond
fundamental uniformity requirement for our algorithmi model.
We all a universal elimination pro edure parametri if it satises our
rst and se ond uniformity requirement, i.e. if the pro edure does not
ontain bran hings whi h otherwise ould be avoided and if all possibly
o urring divisions an be performed on all problem instan es, in the
way we have explained before. In this paper we shall only onsider
parametri elimination pro edures.
We all a parametri elimination pro edure geometri ally robust if it
produ es for any input instan e an output ir uit whi h depends only
on the mathemati al obje ts \input equation system" and \input mor-
phism" but not on their ir uit representation. We shall apply this
notion only to elimination problems given by (geometri ally or s heme-
theoreti ally) at families of algebrai varieties. This means informally
that a parametri elimination pro edure is geometri ally robust if it pro-
du es for at families of problem instan es \ ontinuous" solutions.
Of ourse, our notion of geometri robustness depends on the (geo-
metri or s heme-theoreti al) ontext, i.e. it is not the same for s hemes
or varieties. In Se tion 2.2 we shall explain our idea of geometri robust-
ness in the typi al situation of at families of algebrai varieties given
by redu ed omplete interse tions.
Traditionally, the size of a system of polynomial equations (and in-
equations) is measured in purely extrinsi , synta ti terms (e.g. number
of parameters and variables, degree of the input polynomials, size and
depth of the input ir uit et ). However, there exists a new generation of
symboli and numeri algorithms whi h take also into a ount intrinsi ,
semanti (e.g. geometri or arithmeti ) invariants of the input equa-
tion system in order to measure the omplexity of elimination pro edure
under onsideration more a urately (see e.g. [13, 11, 15℄ and [38, 39℄).
In this paper we shall turn ba k to the traditional point of view. In
Theorem 5 we shall show that, under our universality and uniformity re-
stri tions, no parametri elimination pro edure whi h in ludes eÆ ient
omputation of Zariski losures and of generi ally squarefree paramet-
ri greatest ommon divisors for ir uit represented algebrai families of
polynomials, is able to solve an arbitrary elimination problem in polyno-
mial (sequential) time, if time is measured in terms of ir uit size and
input length is measured in synta ti al terms only.
Finally let us refer to the books [3℄, [35℄ and [37℄ as a general ba k-
ground for notions of algebrai omplexity theory and algebrai geome-
try we are going to use in this paper.
2.2 Flat families of elimination problems

Let, as before, k be an innite and perfe t eld with algebrai losure
k and let U1 ; : : : ; Ur ; X1 ; : : : ; Xn; Y be indeterminates over k. In the se-
quel we shall onsider X1 ; : : : ; Xn and Y as variables and U1 ; : : : ; Ur
as parameters. Let G1 ; : : : ; Gn and F be polynomials belonging to
the k-algebra k[U1; : : : ; Ur ; X1 ; : : : ; Xn ℄. Suppose that the polynomials
G1 ; : : : ; Gn form a regular sequen e in k[U1 ; : : : ; Ur ; X1 ; : : : ; Xn ℄ den-
ing thus an equidimensional subvariety V := fG1 = 0; : : : ; Gn = 0g
of the (r + n)-dimensional aÆne spa e A r A n over the eld k. The
algebrai variety V has dimension r. Let Æ be the (geometri ) degree
of V as dened in [17℄ (this degree does not take into a ount mul-
tipli ities or omponents at innity). Suppose furthermore that the
morphism of aÆne varieties : V ! A r , indu ed by the anoni al
proje tion of A r A n onto A r , is nite and generi ally unramied (this
implies that is at and that the ideal generated by G1 ; : : : ; Gn in
k[U1; : : : ; Ur ; X1 ; : : : ; Xn ℄ is radi al). Let ~ : V ! A r+1 be the mor-
phism dened by ~ (z ) := ((z ); F (z )) for any point z of the variety V .
The image of ~ is a hypersurfa e of A r+1 whose minimal equation is
a polynomial of k[U1 ; : : : ; Ur ; Y ℄ whi h we denote by P . Let us write
deg P for the total degree of the polynomial P and degY P for its par-
tial degree in the variable Y . Observe that P is moni in Y and that
deg P Æ deg F holds. Furthermore, for a Zariski dense set of points
u of A r , we have that degY P is the ardinality of the image of the re-
stri tion of F to the nite set 1 (u). The polynomial P (U1 ; : : : ; Ur ; F )
vanishes on the variety V .
Let us onsider an arbitrary point u = (u1 ; : : : ; ur ) of A r . For arbi-
trary polynomials A 2 k[U1 ; : : : ; Ur ; X1 ; : : : ; Xn℄ and B 2 k[U1; : : : ; Ur ; Y ℄
we denote by A(u) and B (u) the polynomials A(u1 ; : : : ; ur ; X1 ; : : : ; Xn )
and B (u1 ; : : : ; ur ; Y ) whi h belong to k(u1 ; : : : ; ur )[X1 ; : : : ; Xn ℄ and
k(u1; : : : ; ur )[Y ℄ respe tively. Similarly we denote for an arbitrary poly-
nomial C 2 k[U1 ; : : : ; Ur ℄ by C (u) the value C (u1 ; : : : ; ur ) whi h belongs
to the eld k(u1 ; : : : ; ur ). The polynomials G(1u) ; : : : ; G(nu) dene a zero
dimensional subvariety V (u) := fG(1u) = 0; : : : ; G(nu) = 0g = 1 (u) of
the aÆne spa e A n . The degree ( ardinality) of V (u) is bounded by Æ.
Denote by ~(u) : V (u) ! A 1 the morphisms indu ed by the polynomial
F (u) on the variety V (u) . Observe that the polynomial P (u) vanishes on
the (nite) image of the morphism ~ (u) . Observe also that the polyno-
mial P (u) is not ne essarily the minimal equation of the image of ~ (u) ).
We all the equation system G1 = 0; : : : ; Gn = 0 and the polynomial
F a at family of elimination problems depending on the parameters
U1; : : : ; Ur and we all P the asso iated elimination polynomial. An
element u 2 A r is onsidered as a parameter point whi h determines
a parti ular problem instan e (see Se tion 2.1). The equation system
G1 = 0; : : : ; Gn = 0 together with the polynomial F is alled the general
instan e of the given at family of elimination problems and the elimi-
nation polynomial P is alled the general solution of this at family.
The problem instan e determined by the parameter point u 2 A r
is given by the equations G(1u) = 0; : : : ; G(nu) = 0 and the polynomial
F (u) . The polynomial P (u) is alled a solution of this parti ular problem
instan e. We all two parameter points u; u0 2 A r equivalent (in symbols:
0 0 0
u u0 ) if G(1u) = G1(u ) ; : : : ; G(nu) = Gn(u ) and F (u) = F (u ) holds.
0
Observe that u u0 implies P (u) = P (u ) . We all polynomials A 2
k[U1; : : : ; Ur ; X1 ; : : : ; Xn ℄, B 2 k[U1 ; : : : ; Ur ; Y ℄ and C 2 k[U1 ; : : : ; Ur ℄
invariant (with respe t to ) if for any two parameter points u; u0 of
0 0
A with u u0 the respe tive identities A(u) = A(u ) , B (u) = B (u ) and
r
0
C (u) = C (u ) hold.
An arithmeti ir uit in k[U1 ; : : : ; Ur ; Y ℄ with s alars in k[U1 ; : : : ; Ur ℄
is a totally division-free arithmeti ir uit in k[U1 ; : : : ; Ur ; Y ℄, say ,
modelised in the following way: is given by a dire ted a y li graph
whose internal nodes are labelled as before by arithmeti operations.
There is only one input node of , labelled by the variable Y . The other
nodes of indegree zero the ir uit may ontain, are labelled by arbitrary
elements of k[U1 ; : : : ; Ur ℄. These elements are onsidered as the s alars
of . We all su h an arithmeti ir uit invariant (with respe t to
the equivalen e relation ) if all its s alars are invariant polynomials
of k[U1 ; : : : ; Ur ℄. Considering instead of Y the variables X1 ; : : : ; Xn as
inputs, one may analogously dene the notion of an arithmeti ir uit in
k[U1; : : : ; Ur ; X1 ; : : : ; Xn ℄ with s alars in k[U1 ; : : : ; Ur ℄ and the meaning
of its invarian e. However, typi ally we shall limit ourselves to ir uits
in k[U1 ; : : : ; Ur ; Y ℄ with s alars in k[U1 ; : : : ; Ur ℄.
We are now ready to hara terise in the given situation what we mean
by a geometri ally robust parametri elimination pro edure. Suppose
that the polynomials G1 ; : : : ; Gn and F are given by a totally division-
free arithmeti ir uit in k[U1 ; : : : ; Ur ; X1 ; : : : ; Xn ℄. A geometri ally
robust parametri elimination pro edure a epts the ir uit as input
and produ es as output an invariant ir uit in k[U1 ; : : : ; Ur ; Y ℄ with
s alars in k[U1 ; : : : ; Ur ℄, su h that represents the polynomial P . Ob-
serve that in our denition of geometri robustness we did not require
that is an invariant ir uit be ause this would be too restri tive for the
modelling of on rete situations in omputational elimination theory.
The invarian e property required for the output ir uit means the
following: let u = (u1 ; : : : ; ur ) be a parameter point of A r and let (u)
be the arithmeti ir uit in k(u1; : : : ; ur )[Y ℄ obtained from the ir uit
evaluating in the point u the elements of k[U1 ; : : : ; Ur ℄ whi h o ur
as s alars of . Then the invarian e of means that the ir uit (u)
depends only on the parti ular problem instan e determined by the pa-
rameter point u but not on u itself. Said otherwise, a geometri ally
robust elimination pro edure produ es the solution of a parti ular prob-
lem instan e in a way whi h is independent of the possibly dierent
representations of the given problem instan e.
By denition, a geometri ally robust parametri elimination pro e-
dure produ es always the general solution of the at family of elimi-
nation problems under onsideration. This means that for at families,
geometri ally robust parametri elimination pro edures do not introdu e
bran hings in the output ir uits. In Se tion 3.1 we shall exhibit a om-
plexity result whi h may be paraphrased as follows: within the standard
philosophy of ommutative algebra, none of the known (exponential time)
parametri elimination pro edures an be improved to a polynomial time
algorithm. For this purpose it is important to remark that the known
parametri elimination pro edures (whi h are without ex eption based
on linear algebra as well as on omprehensive Grobner basis te hniques)
are all geometri ally robust.
The invarian e property of these pro edures is easily veried in the
situation of a at family of elimination problems. One has only to ob-
serve that all known elimination pro edures a ept the input polynomi-
als G1 ; : : : ; Gn and F in their dense or sparse oeÆ ient representation
or as evaluation bla k box with respe t to the variables X1 ; : : : ; Xn .
Finally we are going to explain what we mean by a (generi ally square-
free) parametri greatest ommon divisor of an algebrai family of poly-
nomials and by the problem of the omputation of this greatest ommon
divisor (in ase of a ir uit represented family).
Suppose that there is given a positive number s of nonzero polynomi-
als, say B1 ; : : : ; Bs 2 k[U1 ; : : : ; Ur ; Y ℄. Let V := fB1 = 0; : : : ; Bs = 0g.
Suppose that V is nonempty. We onsider now the morphism of aÆne
varieties : V ! A r , indu ed by the anoni al proje tion of A r A 1
onto A r . Let S be the Zariski losure of (V ) and suppose that S is an
irredu ible losed subvariety of A r . Let us denote by k [S ℄ the oordinate
ring of S . Sin e S is irredu ible we on lude that k[S ℄ is a domain with
a well dened fun tion eld whi h we denote by k(S ).
Let b1 ; : : : ; bs 2 k[S ℄[Y ℄ be the polynomials in the variable Y with
oeÆ ients in k[S ℄, indu ed by B1 ; : : : ; Bs . Suppose thet there exists an
index 1 k s with bk 6= 0. Without loss of generality we may suppose
that for some index 1 m s the polynomials b1 ; : : : ; bm are exa tly the
non{zero elements of b1 ; : : : ; bs . Observe that ea h polynomial b1 ; : : : ; bs
has positive degree (in the variable Y ).
We onsider b1 ; : : : ; bm as an algebrai family of polynomials (in the
variable Y ) and B1 ; : : : ; Bm as their representatives. The polynomials
b1; : : : ; bm have in k(S )[Y ℄ a well dened normalised (i.e. moni ) greatest
ommon divisor, whi h we denote by h. Let D be the degree of h (with
respe t to the variable Y ).
We are now going to des ribe ertain geometri requirements whi h
will allow us to onsider h as a parametri greatest ommon divisor of
the algebrai family of polynomials b1 ; : : : ; bm.
Our rst requirement is D 1.
Let us x for the moment an arbitrary point u 2 S . The evaluation in
u determines a anoni al k{algebra homomorphism u : k[S ℄ ! k. Let
: k(S ) ! k [ f1g be any pla e whi h extends the homomorphism
u (this means that the valuation ring of ontains the lo al ring of the
point u in the variety S and that the residue lass of is ontained in
k).
We require now that the pla e takes only nite values (i.e. values of
k) in the oeÆ ients of the polynomial h (re all that these oeÆ ients
belong to the eld k(S )).
Moreover we require that the values of the pla e in these oeÆ ients
depend only on the point u and not on the parti ular hoi e of the pla e
extending the homomorphism u .
In this way the pla e maps the polynomial h to a moni polynomial
of degree D in Y with oeÆ ients in k. This polynomial depends only
on the point u 2 S and we denote it therefore by h(u)(Y ). In analogy
with this notation we write bk (u)(Y ) := Bk (u)(Y ) for 1 k m.
Finally we require that the polynomials b1 (u)(Y ); : : : ; bm(u)(Y ) of
k(Y ) are not all zero and that their (normalized) greatest ommon di-
visor is divisible by h(u)(Y ).
We say that a polynomial H of k(U1 ; : : : ; Ur )[Y ℄ with degY H = D
represents the greatest ommon divisor h 2 k (S )[Y ℄ if the oeÆ ients of
H with respe t to the variable Y indu e well{dened rational fun tions
of the variety S and if these rational fun tions are exa tly the oeÆ ienst
of h (with respe t to the varable Y ).
Suppose now that the polynomials B1 ; : : : ; Bs 2 k[U1 ; : : : ; Ur ; Y ℄ sat-
isfy all our requirements for any point u 2 S . Then we all h a para-
metri greatest ommon divisor of the algebrai family of polynomials
b1; : : : ; bm 2 k[S ℄[Y ℄. Any polynomial H 2 k(U1 ; : : : ; Ur )[Y ℄ whi h rep-
resents h is said to represent the parametri greatest ommon divisor
asso iated to the polynomials B1 ; : : : ; Bs .
A moni squarefree polynomial h^ 2 k(S )[Y ℄ with the same zeroes
as h in an algebrai losure of k(S ), is alled a generi ally squarefree
parametri greatest ommon divisor of the algebrai family b1 ; : : : ; bm 2
k[S ℄[Y ℄.
If the hara teristi of the ground eld k is zero, su h a generi ally
squarefree parametri greatest ommon divisor h^ always exists and has
the same properties as h with respe t to the pla es : k(S ) ! k [f1g
onsidered before.
If the hara teristi of k is positive, this general on lusion is not true
anymore. However, in the purely geometri ontext of the present paper,
we may always arrange the polynomials B1 ; : : : ; Bs and their arithmeti
ir uits in order to assure the existen e of a generi ally squarefree great-
est ommon divisor. For this purpose we need that k is a perfe t eld.
In ase that a generi ally squarefree greatest ommon divisor h^ exists,
the notion of a representative of h^ is dened in the same way as for h.
Suppose now that the polynomials B1 ; : : : ; Bs are given by a totally
division{free arithmeti ir uit in k[U1; : : : ; Ur ; Y ℄.
We are now going to formulate the algorithmi problem of omputing
the (generi ally square{free) parametri greatest ommon divisor h (or
^h) of the algebrai family of polynomials b1; : : : ; bm 2 k[S ℄[Y ℄.
We onsider Y as variable and U1 ; : : : ; Ur as parameters. We are look-
ing for an essentially division{free arithmeti ir uit in
k(U1 ; : : : ; Ur )[Y ℄ whi h omputes a representative of the (generi ally
square{free) parametri greatest ommon divisor h (or h^ ), asso iated
to the polynomials B1 ; : : : ; Bs .
We require that the s alars of indu e well{dened rational fun tions
of the variety S and that for any point u 2 S and any pla e : k(S ) !
k [ f1g extending the homomorphism u : k[S ℄ ! k the following
ondition is satisied:
any s alar of the arithmeti ir uit indu es a rational fun tion of
the variety S , whi h is mapped by into a nite value of k. This value
is uniquely determined by the point u 2 S .
The problem of omputing the (generi ally square{free) parametri
greatest ommon divisor of the algebrai family of polynomials
b1; : : : ; bm 2 k[S ℄[Y ℄ onsists now in produ ing from the input ir uit
a (smallest possible) ir uit whi h satises the requirement above.
Observe that the s alars of su h a ir uit , as well as the oeÆ ients
of h, are rational fun tions of the variety S whi h belong to the integral
losure of k[S ℄ in the fun tion eld k(S ). This property of is onserved
under spe ialisations of the k{algebra k[S ℄.
In the proof of Theorem 5 we shall make substantial use of this ob-
servation.
For general ba kground and details about pla es and valuation rings
we refer to [26℄.
3 The intrinsi omplexity of parametri elimination

pro edures
3.1 A parti ular at family of elimination problems
Let n be a xed natural number and let T , U1 ; : : : ; Un , X1 ; : : : ; Xn and Y
be indeterminates over Q . In the sequel we are going to use the following
notation: for arbitrary natural numbers i and j we shall denote by [j ℄i
the ith digit of the binary representation of j . Let P be the following
polynomial of Q [T; U1; : : : ; Un ; Y ℄:
n
2Y1 n
Y
P (T; U1; : : : ; Un; Y ) := Y (j + T Ui[j℄i ) : (3.1)
j =0 i=1
We observe that the dense representation of P with respe t to the
variable Y takes the form
n n
P (T; U1 ; : : : ; Un ; Y ) = Y 2 + A1 Y 2 1
+ + A 2n ;
where A1 ; : : : ; A2n are suitably dened polynomials of Q [T; U1; : : : ; Un ℄.
Let 1 k 2n. In order to determine the polynomial Ak , we ob-
serve, by expanding the right hand side of (3.1), that Ak olle ts the
ontribution of all terms of the form
k
Y n
Y
(jh + T Ui[jh ℄i )
h=1 i=1
with 0 j1 < < jk 2n 1. Therefore the polynomial Ak an be
expressed as follows:
X k
Y n
Y
Ak = (jh + T Ui[jh ℄i )
0j1 <<jk 2n 1 h=1 i=1
X k
Y n
Y
= ( 1)k jh + T Ui[jh ℄i :
0 j1 <<jk 2n 1 h=1 i=1
Observe that for 0 j1 < < jk 2n 1 the expression
k
Y n
Y
jh + T Ui[jh ℄i
h=1 i=1
an be rewritten as:
k
X n
Y
j1 j k + T j1 jbh jk Ui[jh ℄i + terms of higher degree in T:
h=1 i=1
Therefore, we on lude that Ak has the form:
X
Ak = j1 j k
j1 <<jk 2n
0 1
0 1
X k
X n
Y
+T j1 jbh jk Ui[jh ℄i A (3.2)
0j1 <<jk 2n 1 h=1 i=1
+ terms of higher degree in T:
Let us denote by Lk the oeÆ ient of T in the representation (3.2),

namely:
X k
X n
Y
Lk := j1 jbh jk Ui[jh ℄i :
0 j1 <<jk 2n 1 h=1 i=1
For later use we are now going to show the following result, for whose
proof we are indebted to G. Matera.
Lemma 1 The polynomials L1; : : : ; L2n are Q -linearly independent in

Q [U1 ; : : : ; Un ℄.
Proof Let us abbreviate N := 2n 1. We observe thatQfor 1 k N + 1

and 0 j N the oeÆ ient ` of the monomial n U [j℄i o uring
k;j i=1 i
in the polynomial Lk an be represented as
X
`k;j = j1 j k 1 :
j1 <<jk 1 N
0
jr 6=j for r=1;:::;k 1
Claim: For xed N and k, the oeÆ ient `k;j an be written as a poly-
nomial expression of degree exa tly k 1 in the index j . Moreover, this
polynomial expression for `k;j has integer oeÆ ients.
Proof of the Claim. We pro eed by indu tion on the index parameter k .
For k = 1 we have `1;j = 1 for any 0 j N and therefore `1;j is a
polynomial of degree k 1 = 0 in the index j .
Let 1 k N + 1. Assume indu tively that `k;j is a polynomial
of degree exa tly k 1 in the index j and that the oeÆ ients of this
polynomial are integers. We are now going to show that `k+1;j is a
polynomial of degree exa tly k in j and that the oeÆ ients of this
polynomial are integers too. Observe that
X
`k+1;j = j1 j k
0 j1 <<jk N
jr 6=j for r=1;:::;k
!
X X
= j1 j k j j1 j k 1 :
0j1 <<jk N 0 j1 <<jk 1 N
jr 6=j for r=1;:::;k 1
holds. Sin e the term
X
j1 jk
0 j1 <<jk N
does not depend on j and sin e by indu tion hypothesis
X
`k;j = j1 j k 1
j1 <<jk 1 N
0
jr 6=j for r=1;:::;k 1
is a polynomial of degree exa tly k 1 in j , we on lude that `k+1;j is
a polynomial of degree exa tly k in j . Moreover, the oeÆ ients of this
polynomial are integers. This proves our laim.
It is now easy to nish the proof of Lemma 1. By our laim there
exist for arbitrary 1 k N + 1 integers (0k) ; ; (kk)1 with (kk)1 6= 0
su h that for any 0 j N the identity `k;j = (0k) + + (kk)1 j k 1
holds. Hen e for arbitrary 0 k N there exist rational numbers
1(k) ; : : : ; (kk+1
)
(not depending on j ) su h for any 0 j N the ondition
j k = (1k) `1;j + + k(k+1
)
`k+1;j
is satised (here we use the onvention 00 := 1). This implies for any
index 0 k N the polynomial identity
X n
Y
(1k) L1 + + (kk+1
)
Lk+1 = jk Ui[j℄i :
jN
0 i=1
X n
Y
Hen e for any 0 kN the polynomial Pk := jk Ui[j℄i
0j N i=1
belongs to the Q -ve tor spa e generated by L1 ; : : : ; LN +1 .
On the other hand, we dedu e from the nonsingularity of the
Vandermonde matrix j k 0k;jN that the polynomials P0 ; : : : ; PN are
Q -linearly independent. Therefore the Q -ve tor spa e generated by
L1; : : : ; LN +1 in Q [U1 ; : : : ; Un ℄ has dimension N + 1 = 2n . This implies
that L1 ; : : : ; L2n are Q -linearly independent.
With the notations of Se tion 2.2, put now r := n + 1, T := Un+1 and
let us onsider the following polynomials of Q [T; U1; : : : ; Un; X1 ; : : : ; Xn ℄:
G1 := X1 2 X1 ; : : : ; Gn := Xn 2 Xn and
n
X n
Y
F := 2i 1
Xi + T (1 + (Ui 1)Xi ): (3.3)
i=1 i=1
It is lear from their denition that the polynomials G1 ; : : : ; Gn and
F an be evaluated by a (non-invariant) totally division-free arithmeti
ir uit of size O(n) in Q [T; U1 ; : : : ; Un ; X1 ; : : : ; Xn ℄. Observe that the
polynomials G1 ; : : : ; Gn do not depend on the T; U1; : : : ; Un and that
their degree is two. The polynomial F is of degree 2n + 1. More pre-
isely, we have degX1 ;:::;Xn F = n, degU1 ;:::;Un F = n, and degT F = 1.
Although the polynomial F an be evaluated using O(n) arithmeti op-
erations, the sparse representation of F , as a polynomial in the variables
T; U1; : : : ; Un ; X1 ; : : : ; Xn , ontains asymptoti ally 2n non-zero mono-
mial terms.
Let us now verify that the polynomials G1 ; : : : ; Gn and F form a at
family of elimination problems depending on the parameters T; U1 ; : : : ; Un .
The variety V := fG1 = 0; : : : ; Gn = 0g is nothing but the union
of 2n aÆne linear subspa es of A n+1 A n , ea h of them of the form
A n+1 f g, where is any point of the hyper ube f0; 1gn. The anoni al
proje tion A n+1 A n ! A n+1 indu es a morphism : V ! A n+1
whi h glues together the anoni al proje tions A n+1 f g ! A n+1 for
any in f0; 1gn. Obviously the morphism is nite and generi ally
unramied. In parti ular has onstant bres whi h are all anoni ally
isomorphi to the hyper ube P f0; 1gn. Let (j1 ; : : : ; jn ) be an arbitrary
point of f0; 1g and let j := 1in ji 2i 1 be the integer 0 j < 2n
n
whose bit representation is jn jn 1 : : : j1 . One veries immediately the
identity
n
Y
F (T; U1; : : : ; Un ; j1 ; : : : ; jn ) = j + T Uiji :
i=1
Pn
Therefore for any point (t; u1 ; : : : ; un ; j1 ; : : : ; jn ) 2 V with j := 1 ji 2i 1
we have
n
Y
F (t; u1 ; : : : ; un ; j1 ; : : : ; jn ) = j + T uji i :
i=1
>From this observation we dedu e easily that the polynomial
P 2 Q [T; U1; : : : ; Un ; Y ℄
we are looking for (as the elimination polynomial asso iated to the at
family G1 ; : : : ; Gn ; F ) is in fa t
n
2Y1 n
Y
P= (Y (j + T Ui [j℄i )):
j =0 i=1
With the notations of the beginning of this se tion, this polynomial has
the form
n X n n X n
P =Y2 + Ak Y 2 k Y 2 + (ak +T Lk )Y 2 k modulo T 2 ;
1k2n 1k2n
(3.4)
P
with ak := 1j1 <:::<jk 2n 1 j1 : : : jk for 1 k 2n .
Suppose now that there is given a geometri ally robust parametri
elimination pro edure. This pro edure produ es from the input ir uit
an invariant arithmeti ir uit in Q [T; U1; : : : ; Un ; Y ℄, with s alars
in Q [T; U1; : : : ; Un ℄, whi h evaluates the polynomial P . Re all that the
invarian e of means that the s alars of are invariant polynomials of
Q [T; U1 ; : : : ; Un ℄, say
1 ; : : : ;
N .
Let L( ) be the total and L( ) the non-s alar size of the arithmeti
ir uit over Q [T; U1; : : : ; Un℄. Without loss of generality we have
L( ) L( ) and N (L( ) + 3)2 .
Let Z1 ; : : : ; ZN be new indeterminates. From the graph stru ture
of the ir uit we dedu e that there exists for ea h 1 k 2n a
polynomial Qk 2 Q [Z1 ; : : : ; ZN ℄ satisfying the ondition
Qk (
1 ; : : : ;
N ) = Ak (3.5)
(see [3℄ for details). Let us now onsider two arbitrary elements
u; u0 2 A n . Observe that the parameter points (0; u) and (0; u0 ) of
A n+1 are equivalent (in symbols: (0; u) (0; u0 )). From the invarian e
of
1 ; : : : ;
N we dedu e therefore that
j (0; u) =
j (0; u0 ) holds for
any 1 j N . This means that the polynomials
!1 =
1 (0; U1; : : : ; Un ); : : : ; !N :=
N (0; U1 ; : : : ; Un )
are independent from the variables U1 ; : : : ; Un and therefore elements
of Q . From identity (3.4) we on lude that the same is true for 1 :=
A1 (0; U1 ; : : : ; Un ); : : : ; 2n := A2n (0; U1; : : : ; Un ). We shall abbreviate
! := (!1 ; : : : ; !N ) and := (1 ; : : : ; 2n ).
Let us onsider the morphisms of aÆne spa es : A n+1 ! A N and
2n
: A N ! A given by := (
1 ; : : : ;
N ) and := (Qk )1k2n .
Observe that
Æ = (Qk (
; : : : ;
N )) k n = (Ak ) k n
1 1 2 1 2
holds. From our previous onsiderations we dedu e the identities

( Æ )(0; U1 ; : : : ; Un )
= (Qk (
1 (0; U1 ; : : : ; Un ); : : : ;
N (0; U1 ; : : : ; Un )))1k2n
= (Qk (!))1k2n :
In parti ular we have (!) = .
We analyse now the lo al behaviour of the morphism in the point
! 2 A N . Let E! and E be the tangent spa es of the points ! 2 A N
2n
and 2 A . Let us denote the dierential of the map in the point
! by
n
(D )! : E! ! E . Taking the anoni al proje tions of A N and
A 2 as lo al oordinates in the points ! and respe tively, we identify
n
E! with A N and E with A 2 .
For any point u 2 Q n we onsidern
the parametri algebrai urves
u : A 1 ! A N and Æu : A 1 ! A 2 dened as
u := (
1 (T; u); : : : ;
N (T; u)) and Æu := (Ak (T; u))1k2n :
Observe that Æ u = Æu and that u (0) = !, Æu (0) = holds
(independently of the point u).
Now x u 2 Q n and onsider

A
u0 (0) = ( 1 (0; u); : : : ; N (0; u)) and Æu0 (0) = ( k (0; u))1k2n :
T T T
0 0
We have u (0) 2 E! and Æ (0) 2 E . >From (3.4) we dedu e that for
any 1 k 2n the identity A k
T (0; u)) = Lk (u) holds. Therefore we
0
have Æu (0) = (Lk (u))1k2n . This implies
(D )! ( 0 (0)) = Æ0 (0) = (Lk (u))1k2n :
u u (3.6)
holds. From Lemma 1 we dedu e that there exist for 1 l 2n rational
points ul 2 Q n su h that the matrix (Lk (ul ))1k;l2n is regular. From
(3.6) we dedu e now that the 2n N -matrix built by the row ve tors
u0 1 (0); : : : ; u0 2n (0) has rank at least 2n. This implies the lower bound
N 2n .
Therefore we have 2n N (L( ) + 3)2 and hen e the estimate
n2
2 3 L( ) L( ). We have therefore shown that any geometri-
ally robust parametri elimination pro edure applied to the at family
of elimination problems (3.3) produ es a solution ir uit of size at least
n
2 2 3, i.e. a ir uit of exponential size in the length O(n) of the input.
>From the previous example we dedu e that the goal of a polyno-
mial time pro edure for geometri (or algebrai ) elimination an not
be rea hed just by means of an improvement of known, ommutative
algebra-based elimination methods.
On the other hand, our proof method is not very spe i for elimina-
tion problems. This an be visualised by the following example:
let A be the Q {subalgebra of Q [T; U1; : : : ; Un℄ generated by the o-
eÆ ients of the polynomial F of (3.3) with respe t to the variables
X1 ; : : : ; Xn . In the same manner as above, one may show that in the
non{s alar omplexity model with respe t to A, any totally division{free
arithmeti ir uit whi h evaluates the polynomial F using only s alars
from the Q {algebra A, has ne essarily exponential size in n.
On the other hand, the polynomial F an be evaluated by a totally
division{free arithmeti ir uit in Q [T; U1 ; : : : ; Un ; X1 ; : : : ; Xn℄ of size
O(n).
Therefore we see that in the non-s alar model the (sequential time)
omplexity of a polynomial depends strongly on the stru ture of the
algebra of s alars.
It was fundamental for our argumentation above that our notion of a
geometri ally robust parametri elimination pro edure ex ludes bran h-
ings and divisions in the output program. We resume the on lusions
from the omplexity dis ussion of our example in the following form.
Theorem 2 ([18℄) For any n 2 N there exists a at family of elimina-

tion problems depending on n +1 parameters and n variables over Q and
having input length O(n), su h that the following holds: any geometri-
ally robust parametri elimination pro edure whi h solves this problem
n
produ es an output ir uit of size at least 2 2 3 (i.e. of exponential size
in the input length).
3.2 Cir uit en oding of polynomials

We are now going to explain how we may en ode polynomials of a ertain
omplexity lass by their values in suitable test sequen es.
Let L and n be given natural numbers whi h we think xed for the
moment and let X1 ; : : : ; Xn be indeterminates over k. In this se -
tion we shall only onsider totally division-free arithmeti ir uits in
k[X1 ; : : : ; Xn ℄ and we shall only onsider the non-s alar sequential om-
plexity model over k.
By H := HL;n we denote the omplexity lass of all polynomials
H 2 k[X1 ; : : : ; Xn ℄ whi h an be evaluated by a totally division-free
arithmeti ir uit in k[X1 ; : : : ; Xn ℄ of size at most L.
For any polynomial H 2 H we have deg H 2L. On the other hand we
L
have e.g. X12 2 H. This implies that 2L is an exa t degree bound for the
elements of H. Let D := 2L + 1. Sin e the elements of H are ontained
in the D-dimensional k-linear subspa e of k[X1 ; : : : ; Xn ℄ onsisting of
the polynomials of degree at most 2L = D 1, we may onsider H as a
subset of A D . Observe that for any H 2 H and any 2 k the element
H belongs to H. Let W := WL;n be the Zariski losure of H := HL;n in
A D . Sin e H is a one over k , the same is true for W . On the other hand,
H is the image of a k-denable morphism of aÆne spa es A L
( +3)2
!
A D. Thus, in on lusion, W is an irredu ible algebrai variety of A D ,
denable by homogeneous polynomials belonging to k[X1 ; : : : ; Xn ℄ (see
[3℄, Chapter 9 for details).
In the sequel we shall interprete the points of A D as polynomials of
k[X1 ; : : : ; Xn ℄ having degree at most D 1 and vi eversa. In parti ular
we shall interprete the points of W as polynomials.
Denition 1 Let be given a sequen e of points 1 ; : : : ; m 2 kn and let

:= ( 1 ; : : : ; m ). Let us all m the length of .
(i) We all a orre t test sequen e for the polynomials of
k[X1 ; : : : ; Xn ℄
of non-s alar omplexity at most L (i.e. for HL;n ), if for any
H 2 WL;n the following impli ation holds:
H ( 1 ) = 0; : : : ; H ( m ) = 0 =) H = 0:
(ii) We all an identi ation sequen e for the polynomials of
k[X1 ; : : : ; Xn ℄
of non-s alar omplexity at most L (i.e. for HL;n ), if for any two
elements H1 ; H2 of WL;n the following impli ation holds:
H1 ( 1 ) = H2 ( 1 ); : : : ; H1 ( m ) = H2 ( m ) =) H1 = H2 :
Although the polynomials of the omplexity lass HL;n may have ex-
ponential degree 2L , there exist short identi ation sequen es for HL;n .
This is the ontent of the next result.
Lemma 3 ([20, 23℄) Let the notation be as before. Let M k be

a nite set of ardinality at least 4LD2 and let m := 6(2L + n + 1)2 .
The there exist points 1 ; : : : ; m 2 M n su h that := ( 1 ; : : : ; m ) is
an identi ation sequen e for HL;n . Suppose that the points of M n are
equidistributed. The probability of nding by a random2 hoi e su h an
identi ation sequen e is at least (1 4LD 12(2L+n+1) ) > 1=2.
Proof From [20, 23℄ we dedu e that there exist in M n orre t test
sequen es of length m := 6(2L + n + 1)2 for H2L;n and that su h a
orre t test sequen e 2 an be found with probability of su ess
(1 4LD 12(2L+n+1) ) by a random hoi e in M n .
Let := ( 1 ; : : : ; m ) 2 kmn be a orre t test sequen e for H L;n .
2
Let H1 ; H2 2 WL;n and suppose that
H1 ( 1 ) = H2 ( 1 ); : : : ; H1 ( m ) = H2 ( m )
holds. Observe that H := H2 H1 belongs to W2L;n . Therefore we
have H ( 1 ) = 0; : : : ; H ( m ) = 0, and, sin e is a orre t test sequen e
for H2L;n, we may on lude H = H2 H1 = 0.
Let now m := 6(2L + n + 1)2 . Then, by Lemma 3, we may x an
identi ation sequen e := ( 1 ; : : : ; m ) 2 kmn for the omplexity
lass H = HL;n . Let S1 ; : : : ; Sm be new indeterminates denoting the
anoni al oordinate fun tions of A m . We are going to onsider the
( ) ( )
morphism L;n : WL;n ! A m , dened for H 2 WL;n by L;n (H ) :=
( )
(H ( 1 ); : : : ; H ( m )). Let SL;n be the Zariski losure of the image of the
( ) ( ) ( )
morphism L;n and let us abbreviate := L;n and S := SL;n .
Sin e the variety W and the morphism : W ! S are k-denable,
we on lude that S is k-denable too. Moreover for any polynomial
H 2 W and any value 2 k we have
(H ) = (H ( 1 ); : : : ; H ( m )) = (H )
and from this homogeneity property of we on lude that S is a one
over the eld k. Therefore S is denable by homogeneous polynomials
belonging to k[S1 ; : : : ; Sm ℄. Sin e the variety W is irredu ible and S is
the losure of the image of , we on lude that S is irredu ible too.
By assumption := ( 1 ; : : : ; m ) is an identi ation sequen e for the
omplexity lass H = HL;n. Therefore we on lude that : W ! S is
an inje tive, dominant morphism. Hen e is birational.
Let = (1 ; : : : ; m ) where 1 ; : : : ; m are suitable oordinate fun -
tions of the aÆne variety W . Let us onsider W as a losed subvariety
of the aÆne spa e A D . We re all that the points of A D orrespond to
the polynomials of k[X1 ; : : : ; Xn ℄ of degree at most D and that the mor-
phism is dened by means of the evaluation of the polynomials of W
in the points 1 ; : : : ; m . This implies that there exist linear forms in
the oordinate ring of A D su h that 1 ; : : : ; m are the restri tions of
these linear forms to the variety W . From the inje tivity of we dedu e
that W \ f1 = 0; : : : ; m = 0g ontains only the origin of A D . This
implies that the homogeneous map indu es a nite morphism between
the proje tive varieties asso iated to the ones W and S . In fa t, the
standard proof of this lassi al result implies something more, namely
that also the morphism : W ! S is nite (see [37℄ I.5.3, Theorem
8 and proof of Theorem 7). In parti ular, is a surje tive, losed map
and hen e a homeomorphism with respe t to the Zariski topologies of
W and S .
Thus we have shown the following result:
Lemma 4 Let notations be as before and let m := 6(2L + n + 1)2 . Sup-

pose that there is given an identi ation sequen e := ( 1 ; : : : ; m ) 2
kmn for the omplexity lass HL;n and let L;n( )
: WL;n ! SL;n ( )
be
the morphism of aÆne varieties asso iated to the identi ation sequen e
( )
. Then L;n is a nite, bije tive and birational morphism of algebrai
( )
varieties and in parti ular L;n is a homeomorphism with respe t to the
( )
Zariski topologies of WL;n and SL;n
( )
Sin e the varieties WL;n and SL;n are irredu ible, we on lude that
( )
their oordinate rings k[WL;n ℄ and k[SL;n ℄ are domains with fun tion
( ) ( )
elds k(WL;n) and k(SL;n ). The morphism L;n indu es an embedding
( )
of the oordinate ring k[SL;n ℄ into k[WL;n℄ (and the same is true for the
orresponding fun tion elds). Moreover, the niteness of the morphism
( ) ( )
L;n means that k[WL;n℄ is an integral ring extension of k[SL;n ℄.
Disregarding the omplexity aspe t, Lemma 4 says that it is possible
to re onstru t the oeÆ ients of a polynomial H 2 HL;n from the values
of H in a given identi ation sequen e , even if the sequen e is
short in omparison with the degree of H . Lemma 4 says further that
this re onstru tion is rational (i.e. it uses only arithmeti al operations)
and that possibly o uring divisions may always be performed by limit
pro esses in the spirit of de l'Hôpital's rule. These pro esses produ e
only nite limits, be ause the oeÆ ients of H are integrally dependent
on the values of H in the given identi ation sequen e . In algebrai
terms, we may modelise these limit pro esses by pla es ( orresponding to
( ) ( )
valuation rings) whi h map the fun tion eld k(SL;n ) of the variety SL;n

into the set of values k [f1g and take only nite values on k[SL;n℄. The ( )
niteness of the limits mentioned before is modelised by the requirement

that any extension of su h a pla e to the fun tion eld k(WL;n ) takes
nite values on k[WL;n℄. This requirement is satised, be ause k[WL;n ℄
( )
is integral over k[SL;n ℄ (see [26℄ for details about pla es and integral
extensions).
3.3 The omplexity of parametri elimination pro edures
In this se tion we suppose that the ground eld k is of hara teristi zero.
Let n be a xed natural number, m(n) := 6(3n + 1)2 ; N (n) := (n + 3)2
and X1 ; : : : ; Xn ; Z1 ; : : : ; ZN (n) and Y; S1; : : : ; Sm(n) indeterminates over
k.
Let us x an identi ation sequen e := ( 1 ; : : : ; m(n) ) 2 km(n)n
for the polynomials of k[X1 ; : : : ; Xn ℄ having non-s alar omplexity at
most n (i.e. for the omplexity lass Hn;n ).
>From [3℄, Chapter 9, Theorem 9.9 we dedu e that there exists a
polynomial
Rn 2 k[Z1 ; : : : ; ZN (n); X1 ; : : : ; Xn ℄
satisfying the following onditions,
Rn an be evaluated by a totally division-free arithmeti ir uit in
k[Z1; : : : ; ZN (n); X1 ; : : : ; Xn ℄ of non-s alar size N (n)
for any H 2 Hn;n and any totally division-free arithmeti ir uit in
k[X1 ; : : : ; Xn ℄, su h that has non-s alar size n, s alars 1 ; : : : ; N (n) 2
k and su h that evaluates the polynomial H , we have
H = Rn (1 ; : : : ; N (n) ; X1 ; : : : ; Xn ):
Let us onsider the following existential formula n (S1 ; : : : ; Sm(n) ; Y ) in
the free indeterminates ("free variables" in the terminology of mathe-
mati al logi ) S1 ; : : : ; Sm(n) ; Y ,
^
(9X1 ); : : : ; (9Xn )(9Z1 ); : : : ; (9ZN (n) )( Xi 2 X i = 0 ^
in
1
^
Sk = Rn (Z1 ; : : : ; ZN (n); k ) ^ Y = Rn (Z1 ; : : : ; ZN (n); X1 ; : : : ; Xn )):
1km(n)
Observe that the existential formula (9Y )n (S1 ; : : : ; Sm(n) ; Y ) des ribes
( )
the set n;n (Hn;n ). Thus the formula n (S1 ; : : : ; Sm(n) ; Y ) introdu es an
impli it semanti al dependen e between the indeterminates S1 ; : : : ; Sm(n) .
In the sequel we shall onsider the inteterminates S1 ; : : : ; Sm(n) as pa-
rameters and Y as variable.
Let us nally remark that the formula n (S1 ; : : : ; Sm(n); Y ) may be
represented by a totally division-free ir uit in
k[X1 ; : : : ; Xn ; Z1 ; : : : ; ZN (n) ; S1 ; : : : ; Sm(n) ; Y ℄
of size O(n4 ).
Let us now onsider an arbitrary (universal) parametri elimination
pro edure with asso iated sequential time omplexity measure T and
suppose that and T satisfy the following onditions:
(i) For any totally division-free arithmeti input network of total
size L( ) su h that represents an existential input formula
in the elementary language of algebrai geometry over the eld k,
the elimination pro edure is able to produ e a totally division-
free arithmeti output network with L( ) T (L( )), su h
that represents a quantier-free formula whi h is semanti ally
equivalent to .
(ii) For any totally division-free arithmeti input network 1 repre-
senting a onstru tible subset X of an appropriate aÆne spa e,
the pro edure is able to produ e a totally division-free arith-
meti output ir uit 1 with L( 1 ) T (L(1 )), su h that 1
represents a suitable polynomial equation system for the Zariski
losure of the onstru tible set X .
(iii) Let U1 ; : : : ; Ur and Y be indeterminates and let be given a positive
number s of nonzero polynomials, say
B1 ; : : : ; Bs 2 k[U1 ; : : : ; Ur ; Y ℄. Let V := fB1 = 0; : : : ; Bs = 0g.
Suppose that V is nonempty. Consider the morphism of aÆne
varieties : V ! A r , indu ed by the anoni al proje tion of
A r A 1 onto A r . Let S be the Zariski losure of (V ) and
assume that S is an irredu ible losed subvariety of A r . Suppose
that the polynomials B1 ; : : : ; Bs are given by a totally division{
free arithmeti ir uit 2 in k[U1 ; : : : ; Ur ; Y ℄ and suppose that
they satisfy all our requirements at the end of Se tion 2.2. Hen e
there exists a well{dened parametri greatest ommon divisor
h 2 k(S )[Y ℄ asso iated to the polynomials B1 ; : : : ; Bs .
Then the pro edure is able to produ e an essentially division-
free arithmeti ir uit 2 in k(U1 ; : : : ; Ur )[Y ℄ with
L( 2 ) T (L(2 )) su h that 2 omputes a representative H 2
k(U1 ; : : : ; Ur )[Y ℄ of the (generi ally squarefree) parametri great-
est ommon divisor h 2 k(S )[Y ℄ and su h that 2 satises the
requirements formulated at the end of Se tion 2.2 for su h a ir-
uit. Moreover, the same holds true for the generi ally squarefree
parametri greatest ommon divisor asso iated to the polynomi-
als B1 ; : : : ; Bs .
In view of onditions (ii) and (iii) above, we say that the paramet-
ri elimination pro edure omputes eÆ iently Zariski losures and
(generi ally squarefree) greatest ommon divisors.
The requirement that 1 and 2 are ir uits (and not arithmeti net-
works) ontains impli itly the meaning that the pro edure is bran hing
parsimonious (see Se tion 2). Similarly, the behaviour of the ir uit 2
under spe ialisation by pla es expresses the requirement that the pro e-
dure is division parsimonious.
Let us nally observe that, after a suitable adaption of the data stru -
tures, all known universal elimination pro edures satisfy with respe t to
appropriate sequential time omplexity measures our onditions (i); (iii)
and (iii) (see Se tion 3.4 for more details).
We are now ready to state the main result of this paper.
Theorem 5 Assume that the ground eld k is of hara teristi zero.

Let be an arbitrary parametri elimination pro edure with asso iated
sequential time omplexity measure T and suppose that and T sat-
isfy onditions (i); (iii) and (iii) above. There exists a family of totally
division-free arithmeti input ir uits = (n )n2N su h that ea h n
has total size nO(1) and represents a parametri family of polynomial
equation systems with a well dened, anoni al elimination polynomial
Pn . The elimination pro edure produ es for ea h n 2 N from the input
ir uit n an essentially division-free output ir uit n whi h represents
the elimination polynomial Pn . The total size of the output ir uit n
annot be polynomial in the parameter n. Therefore T is not a polyno-
mial fun tion.
Proof Let be given a parametri elimination pro edure with asso-

iated sequential time omplexity measure T , as in the statement of
the theorem. Let n 2 N . We apply rst the pro edure to any to-
tally division-free arithmeti ir uit n of size O(n4 ) whi h represents
the formula n (S1 ; : : : ; Sm(n) ; Y ). In virtue of ondition (i) above, we
obtain as output a totally division-free arithmeti network 1(n) of size
T (O(n4 )) representing a quantier-free formula whi h is semanti ally
equivalent to n (S1 ; : : : ; Sm(n) ; Y ). Applying now the pro edure to
the network 1(n) , we obtain in virtue of ondition (ii) above a totally
division-free ir uit 2(n) of size T 2 (O(n4 )) whi h represents a positive
number s of polynomials, say B1 ; : : : ; Bs 2 k[S1 ; : : : ; Sm(n) ; Y ℄, su h that
B1 = 0; : : : ; Bs = 0 forms a polynomial equation system for the Zariski
losure of the onstru tible subset of the aÆne spa e A m(n) A 1 , de-
ned by the formula n (S1 ; : : : ; Sm(n) ; Y ). This equation system on-
tains polynomials of k[S1 ; : : : ; Sm(n) ℄ whi h determine the Zariski losed
( )
subset S := Sn;n of A m(n) (re all that the existential formula
(9Y )n (S1 ; : : : ; Sm(n) ; Y )
( ) ( )
des ribes the set n;n (Hn;n ) and that Sn;n is its Zariski losure).
Let b1 ; : : : ; bs 2 k[S ℄[Y ℄ be the polynomials in Y with oeÆ ients
in k[S ℄ indu ed by B1 ; : : : ; Bs and observe that not all polynomials
b1; : : : ; bs are zero. We onsider now an arbirary point
u = (u1 ; : : : ; um(n) ) 2 n;n
( )
(Hn;n ):
There exists a point = (1 ; : : : ; N (n) ) 2 A N (n) su h that
E := Rn (; X1 ; : : : ; Xn) := Rn (1 ; : : : ; N (n) ; X1 ; : : : ; Xn )
( )
satises the ondition u = n;n (E ) = (Rn (; 1 ); : : : ; Rn (; m(n) )).
>From Lemma 4 we dedu e that the polynomial E 2 k[X1 ; : : : ; Xn ℄
depends only on the point u and not on the parti ular hoi e of 2
N (n) . Let us therefore write E := E . Let P := Q
A u u (1 ;:::;n )2f0;1gn (Y
Eu (1 ; : : : ; n )).
Interpreting now the formula n (S1 ; : : : ; Sm(n) ; Y ) semanti ally, we
see that n (u1 ; : : : ; um(n) ; Y ) is equivalent to Pu (Y ) = 0.
( )
Pn
For a suitable hoi e of u 2 n;n (Hn;n ) (e.g. hoosing u with Hu =
i
i=1 2 Xi ) we obtain a separable polynomial Pu .
1
There exists therefore a nonempty Zariski open subset U S , on-

( )
tained in n;n (Hn;n ), su h that for u 2 U the greatest ommon divisor of
the non{zero elements of b1 (u; Y ) := B1 (u; Y ); : : : ; bs (u; Y ) := Bs (u; Y )
has the same zeroes in k as the polynomial Pu (Y ) and su h that Pu (Y )
is separable.
Let h 2 k(S )[Y ℄ be the (normalised) greatest ommon divisor of
the non{zero elements between the polynomials b1 ; : : : ; bs and let h^ 2
k(S )[Y ℄ be the unique moni sparable polynomial with the same roots
as h in an algebrai losure of k(S ). Without loss of generality we may
assume that for any point u 2 U the spe ialised polynomial h^ (u; Y ) is a
well{dened element of k[Y ℄ with h^ (u; Y ) = Pu (Y ).
Let A be the integral losure of the oordinate ring k[S ℄ in its fra tion
eld k(S ). From Lemma 4 we dedu e now that ^h and h belong to the
polynomial ring A[Y ℄ and that the polynomials B1 ; : : : ; Bs satisfy the
requirements of the end of Se tion 2.2. In other words,the polynomial h
is the parametri greatest ommon divisor asso iated to B1 ; : : : ; Bs and
^h is its generi ally squarefree ounterpart.
Let us now onsider S1 ; : : : ; Sm(n) as parameters and Y as variable.
Applying nally the pro edure to the ir uit 2(n) we obtain by virtue
of ondition (iii) above an essentially division-free arithmeti ir uit n
in k(S1 ; : : : ; Sm(n) )[Y ℄ whi h omputes a representative of the generi-
ally squarefree greatest ommon divisor h^ 2 A[Y ℄. The size of n is
T 3 (O(n4 )) and the ir uit n satises the requirements of the end of
Se tion 2.2. From ondition (iii) above we dedu e that all s alars of

n are rational fun tions of k (S1 ; : : : ; Sm(n) ) whi h represent elements
of the ring A.
P Q
We onsider now the polynomial F := ni=1 2i 1 Xi + T ni=1 (1+(Ui
1)Xi ) introdu ed in Se tion 3.1. This polynomial an be represented by
a totally division-free ir uit of non-s alar size n over k[U1 ; : : : ; Un ; T ℄.
Let A1 := F (U1 ; : : : ; Un ; T; 1); : : : ; Am(n) := F (U1 ; : : : ; Un ; T; m(n)).
Then A1 ; : : : ; Am(n) are invariant polynomials of k[U1; : : : ; Un ; T ℄. Spe-
ialising now the parameters S1 ; : : : ; Sm(n) into the polynomials
A1 ; : : : ; Am(n) , we obtain a k{algebra homomorphism
k[S ℄ ! k[A ; : : : ; Am n ℄:
1 ( )
We onsider this homomorphism as a spe ialization of the oordinate

ring k[S ℄.
Sin e the ir uit n satises the requirements of the end of Se tion
2.2, the s alars of n be ome now spe ialised into elements of the eld
k(A1 ; : : : ; Am(n) ). These elements belong to the integral losure of the
k{algebra k[A1 ; : : : ; Am(n) ℄ in its fra tion eld k(A1 ; : : : ; Am(n) ). Sin e
k[A1 ; : : : ; Am(n) ℄ is a k{subalgebra of the (integrally losed) polynomial
ring k[U1 ; : : : ; Ur ; T ℄, we on lude that the above spe ialisation maps
the s alars of the ir uit n into elements of k[U1 ; : : : ; Ur ; T ℄ whi h are
integral over k[A1 ; : : : ; Am(n) ℄. This implies that the s alars of n are
spe ialised into invariant polynomials of k[U1 ; : : : ; Ur ; T ℄.
Denote now by n the totally division{free invariant ir uit of
k[U1; : : : ; Ur ; T; Y ℄ obtained by spe ializing in n the s alars as ex-
plained before. Let L( n ) be the non{s alar size of the ir uit n over
k[U1; : : : ; Ur ; T ℄. Then we have L( n ) L( n ) T 3 (O(n4 )) and n
omputes the image of h^ under the above spe ialisation,
Q2n 1 Qn
namely the
[j ℄i
elimination polynomial P := j=0 (Y (j + T i=1 Ui )) of Se tion
3.1.
>From the invarian e of the ir uit n and Theorem 2 and its proof
we dedu e now the estimate L( n ) 2 n2 3. On the other hand we
have T 3 (O(n4 )) L( n ). This implies that T annot be a polynomial
fun tion.
3.4 State of the art in ir uit-based elimination
Let us now analyse from a general non-uniform point of view how the
seminumeri al elimination pro edure designed in [13℄ and [11℄ works on
a given at family of zero-dimensional elimination problems.
Let U1 ; : : : ; Um; X1 ; : : : ; Xn ; Y be indeterminates over the ground eld
k and let G1 ; : : : ; Gn ; F be polynomials belonging to the k-algebra
k[U1 ; : : : ; Um; X1 ; : : : ; Xn ℄:
Let d := maxfdeg G1 ; : : : ; deg Gn g and suppose that G1 ; : : : ; Gn and F
are given by straight-line programs in k[U1 ; : : : ; UmX1 ; : : : ; Xn ℄ of length
L and K respe tively. Suppose that the polynomials G1 ; : : : ; Gn form
a regular sequen e in k[U1 ; : : : ; Um ; X1 ; : : : ; Xn ℄ dening thus an equidi-
mensional subvariety V = fG1 = 0; : : : ; Gn = 0g of A m+n of dimension
m.
Assume that the morphism : V ! A m , indu ed by the anoni al
proje tion of A m+n onto A m is nite and generi ally unramied. Let
Æ be the degree of the variety V and let D Æ be the degree of the
morphism . Furthermore let ~ : V ! A m+1 be the morphism of
aÆne varieties dened by ~ (z ) := ((z ); F (z )) for any point z of V . Let
P 2 k[U1 ; : : : ; Um ; Y ℄ be the minimal polynomial of the image of ~ . The
polynomial P is moni in Y and one sees immediately that deg P
Æ deg F and degY P D holds. Let us write Æ := degU1 ;:::;Um P .
Let us onsider as Algorithm 1 and Algorithm 2 two non-uniform
variants of the basi elimination method designed in [13℄ and [11℄.
Algorithm 1 is represented by an arithmeti network of size KÆO(1) +
L(nd)O(1) where is the degree of the equation system G1 =
0; : : : ; Gn = 0 (observe that always Æ deg G1 deg Gn holds).
The output is a straight-line program 1 in k[U1 ; : : : ; Um ; Y ℄ of length
(K + L)(nÆ)O(1) whi h represents the polynomial P .
Algorithm 2 starts from the geometri des ription of a unramied
parameter (and lifting) point u = (u1 ; : : : ; um) of km whi h has the
additional property that the image of F restri ted to the set fug
1 (u) has ardinality D = degY P . The algorithm produ es then
an arithmeti ir uit 2 in k[U1; : : : ; Um ; Y ℄ of length
O(KDO(1) log Æ ) + ÆO(1) = K (Æ deg F )O(1)
whi h represents the polynomial P .
We observe that KÆO(1) is a hara teristi quantity whi h appears in
the length of both ir uits 1 and 2 . One may ask whether a om-
plexity of type KÆO(1) is intrinsi for the elimination problem under
onsideration.
In view of Bezout's Theorem and the lower omplexity bound of The-
orem 5, one may guess that a quantity of asymptoti order KÆ may be
hara teristi for the intrinsi sequential time omplexity of universal
parametri elimination pro edures.
4 Con lusions
It was fundamental for our argumentation in Se tion 3 that our notion
of parametri elimination pro edure restri ts severely the possibility of
bran hings in the output ir uit or network. This suggests that any
polynomial time elimination algorithm (if there exists one) must have a
huge topologi al omplexity. Thus hypotheti al eÆ ien y in geometri
elimination seems to imply ompli ated asuisti s.
Let us also mention that the proof method of Se tion 3 ontributes
absolutely nothing to the elu idation of the fundamental thesis of al-
gebrai omplexity theory, whi h says that geometri elimination pro-
du es elimination polynomials whi h are intrinsi ally hard to evaluate
(i.e. these polynomials need asymptoti ally degree{mu h sequential time
for their evaluation). A good example for this thesis is the Po hhammer{
Wilkinson polynomial whi h an be obtained easily as the solution of a
suitable elimination problem but whi h is onje turally hard to evaluate
(see e.g. [19, 1℄) Similarly no advan e is obtained by our method with
respe t to the question whether PC 6= NPC holds in the BSS omplexity
model, see [2℄, Chapter7.
In fa t our ontribution onsists only in the dis overy of a very lim-
iting uniformity property present in all known elimination pro edures.
This uniformity property inhibits the transformation of these elimina-
tion pro edures into polynomial time algorithms.
In on lusion, when treating with algorithmi elimination problems,
one should not expe t too mu h from the output: as observed already
by Hilbert, a strong limitation to spe i elimination problems (with ad-
ditional a priori information or additional semanti al stru ture) is ne -
essary in order to avoid huge diÆ ulties and one should also renoun e to
ompute the whole elimination obje t, if this is not required in advan e.
In parti ular, anoni al elimination polynomials ontain the omplete in-
formation about the elimination problem under onsideration, they are
" o{versal", and this makes ne essarily intri ate their representation in
any reasonable data{stru ture (for more details see [6℄). A way out
of this dilemma should be found in analysing whi h information about
elimination obje ts is really relevant, avoiding in this way to struggle
with enormous obje ts whi h en ode a lot of spurious knowledge.
A knowlegment.
The authors wish to express their gratitude to Guillermo Matera and
Rosa Wa hen hauzer (Buenos Aires, Argentina) and to David Castro
and Luis Miguel Pardo (Santander, Spain) for many fruitful dis ussions
and ideas. They are deeply indepted to Ja ques Morgenstern (Ni e,
Fran e), who suggested this resear h and made together with the authors
the rst and most fundamental steps toward the omplexity model and
results of this paper.
Bibliography
[1℄ Blum, Cu ker, Shub, and Smale. Algebrai settings for the problem
\P = NP ?". In Renegar, Shub, and Smale, editors, The Mathemati s of
6
Numeri al Analysis: 1995 (25th) AMS-SIAM Summer Seminar in Applied

Mathemati s (Le tures in Applied Mathemati s, Volume 32), Ameri an
Mathemati al So iety. 1996.
[2℄ Lenore Blum, Felipe Cu ker, Mi hael Shub, and Steve Smale.
Complexity and real omputation. Springer-Verlag, New York, 1998. With
a foreword by Ri hard M. Karp.
[3℄ P. Burgisser, M. Clausen, and M. A. Shokrollahi. Algebrai Complexity
Theory. Springer, 1997.
[4℄ L. Caniglia, A. Galligo, and J. Heintz. Borne simple exponentielle pour
les degres dans le theoreme des zeros sur un orps de hara teristique
quel onque. C. R. A ad. S i. Paris, 307:255{258, 1988.
[5℄ L. Caniglia, A. Galligo, and J. Heintz. Some new ee tivity bounds in
omputational geometry. In T. Mora, editor, Applied Algebra, Algebrai
Algorithms and Error Corre ting Codes. Pro eedings of AAECC-6, volume
357 of LNCS, pages 131{152. Springer, 1989.
[6℄ D. Castro, M. Giusti, J. Heintz, G. Matera, and L. M. Pardo. Data
stru tures and smooth interpolation pro edures in elimination theory.
Manus ript, 1999.
[7℄ A. L. Chistov and D. Y. Grigoriev. Subexponential time solving systems
of algebrai equations. LOMI preprint E-9-83, E-10-83, Steklov Institute,
Leningrad, 1983.
[8℄ M. Demazure. Le theoreme de omplexite de mayr{meyer. Manus ript
E ole Polyte hnique, 1985.
[9℄ A. Di kenstein, N. Fit has, M. Giusti, and C. Sessa. The membership
problme for unmixed polynomial ideals is solvable in single exponential
time. Dis rete Applied Mathemati s, 33:73{94, 1991.
[10℄ P. Gianni and T. Mora. Algebrai solution of systems of polynomial
equations using grobner bases. In Applied Algebra, Algebrai Algorithms
and Error Corre ting Codes, Pro eedings of AAECC-5, volume 356 of
LNCS, pages 247{257. Springer, 1989.
[11℄ M. Giusti, K. Hagele, J. Heintz, J. E. Morais, J. L. Monta~na, and
L. M. Pardo. Lower bounds for diophantine approximation. In Pro eedings
of MEGA'96, volume 117,118, pages 277{317. Journal of Pure and Applied
Algebra, 1997.
[12℄ M. Giusti and J. Heintz. La determination des points isoles et de la
dimension d'une variete algebrique peut se faire en temps polynomial. In
D. Eisenbud and L. Robbiano, editors, Computational Algebrai Geometry
and Commutative Algebra, volume XXXIV of Symposia Matemati a, pages
216{256. Cambridge University Press, 1993.
[13℄ M. Giusti, J. Heintz, J. E. Morais, J. Morgenstern, and L. M. Pardo.
Straight{line programs in geometri elimination theory. J. of Pure and
App. Algebra, 124:101{146, 1998.
[14℄ M. Giusti, J. Heintz, J. E. Morais, and L. M. Pardo. When polynomial
equation systems an be solved fast ? In G. Cohen, H. Giusti, and
T. Mora, editors, Applied Algebra, Algebrai Algorithms and Error
Corre ting Codes, Pro eedings AAECC-11, volume 948 of LNCS, pages
205{231. Springer, 1995.
[15℄ M. Giusti, J. Heintz, J. E. Morais, and L. M. Pardo. Le rôle des
stru tures de donnees dans les problemes d'elimination. C. R. A ad. S i.
Paris, 325:1223{1228, 1997.
[16℄ M. Giusti, J. Heintz, and J. Sabia. On the eÆ ien y of ee tive
nullstellensatze. Computational Complexity, 3:56{95, 1993.
[17℄ J. Heintz. Denability and fast quantier elimination in algebrai ally
losed elds. Theor. Comput. S i., 24(3):239{277, 1983.
[18℄ J. Heintz, G. Matera, L. M. Pardo, and R. Wa hen hauzer. The
intrinsi omplexity of parametri elimination methods. Ele tron. J.
SADIO, 1(1):37{51 (ele troni ), 1998.
[19℄ J. Heintz and J. Morgenstern. On the intrinsi omplexity of
elimination theory. J. of Complexity, 9:471{498, 1993.
[20℄ J. Heintz and C. P. S hnorr. Testing polynomials whi h are easy to
ompute. In Logi and Algorithmi , volume 30 of Monographie de
l'Enseignement Mathematique, pages 237{254, 1982.
[21℄ D. Hilbert. Neubegrundung der mathematik. erste mitteilung.
Abhandl. aus dem Math. Seminar d. Hamb. Univ., 1:157{177, 1922.
[22℄ Hidetsune Kobayashi, Shui hi Moritsugu, and Robert W. Hogan.
Solving systems of algebrai equations. In Symboli and algebrai
omputation (Rome, 1988), pages 139{149. Springer, Berlin, 1989.
[23℄ T. Kri k and L. M. Pardo. A omputational method for diophantine
approximation. In L. Gonzalez-Vega and T. Re io, editors, Algorithms in
Algebrai Geometry and Appli ations. Pro eedings of MEGA'94, volume
143 of Progress in Mathemati s, pages 193{254. Birkhauser Verlag, 1996.
[24℄ L. Krone ker. Grundzuge einer arithmetis hen theorie der
algebrais hen grossen. J. reine angew. Math., 92:1{122, 1882.
[25℄ Klaus Kuhnle. Spa e Optimal Computation of Normal Forms of
Polynomials. PhD thesis, Universitat M un hen, 1998.
[26℄ Serge Lang. Algebra. Addison-Wesley Publishing Co., Reading, Mass.,
se ond edition, 1984.
[27℄ M. Giusti G. Le erf and B. Salvy. A grobner free alternative for
polynomial systems solving. to appear in J. Complexity, 2000.
[28℄ F. S. Ma aulay. The Algebrai Theory of Modular Systems. Cambridge
University Press, 1916.
[29℄ Guillermo Matera. Probabilisti algorithms for geometri elimination.
Appl. Algebra Engrg. Comm. Comput., 9(6):463{520, 1999.
[30℄ Guillermo Matera and Jose Maria Turull Torres. The spa e omplexity
of elimination theory: upper bounds. In Foundations of omputational
mathemati s (Rio de Janeiro, 1997), pages 267{276. Springer, Berlin,
1997.
[31℄ J. Heintz G. Matera and A. Waissbein. On the time{spa e omplexity
of geometri elimination pro edures. to appear in Appl. Algebra Engrg.
Comm. Comput., 2000.
[32℄ E. Mayr. Membership in polynomial ideals over Q is exponential spa e
omplete. In B. Monien et al, editor, Pro eedings of the 6th Annual
Symposium on Theoreti al Aspe ts of Computer S ien e (STACS'89),
Paderborn (FRG) 1989, number 349 in Le ture Notes in Computer
S ien e, pages 400{406. Springer, 1989.
[33℄ E. Mayr and A. Meyer. The omplexity of the word problem for
ommutative semigroups. Adv. in Math., 46:305{329, 1982.
[34℄ D. Castro K. Hagele J.E. Morais and L.M. Pardo. Krone ker's and
newton's approa hes to solving : a rst omparison. to appear in J. of
Complexity, 2000.
[35℄ David Mumford. The Red Book of Varieties and S hemes, volume 1358
of LNM. Springer, Berlin, 1 edition, 1988.
[36℄ Raghavan Narasimhan. Introdu tion to the theory of analyti spa es.
Springer-Verlag, Berlin, 1966. Le ture Notes in Mathemati s, No. 25.
[37℄ I. R. Shavarevi h. Basi Algebrai Geometry : Varieties in Proje tive
Spa e. Springer, 1994.
[38℄ M. Shub and S. Smale. Complexity of Bezout's theorem I: Geometri
aspe ts. J. of the AMS, 6(2):459{501, 1993.
[39℄ M. Shub and S. Smale. Complexity of Bezout's theorem V: Polynomial
time. Theor. Comp. S i., 133:141{164, 1994.
[40℄ B. L. van der Waerden. Moderne Algebra II. Springer Verlag,Berlin,
1931.
[41℄ J. von zur Gathen. Parallel arithmeti omputations: a survey. In
B. Rovan J. Gruska and J. Wiedermann, editors, Pro eedings of the 12th
Symposium on Mathemati al Foundations of Computer S ien e, volume
233 of LNCS, pages 93{112, Bratislava, Cze hoslovakia, August 1986.
Springer.
[42℄ J. von zur Gathen. Parallel linear algebra. In John H. Reif, editor,
Synthesis of Parallel Algorithms. Morgan Kaufmann, 1993.
[43℄ Chee-K. Yap. A new lower bound onstru tion for ommutative Thue
systems with appli ations. J. Symboli Comput., 12(1):1{27, 1991.

Ssmart Little Black Boxes

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Ssmart Little Black Boxes

Caricato da

Copyright:

Formati disponibili

1

Krone ker's smart, little bla k boxes

are ommutative algebra and its modern derivate, namely todays

2.2 Flat families of elimination problems

3 The intrinsi omplexity of parametri elimination

Let us denote by Lk the oeÆ ient of T in the representation (3.2),

Lemma 1 The polynomials L1; : : : ; L2n are Q -linearly independent in

Proof Let us abbreviate N := 2n 1. We observe thatQfor 1 k N + 1

holds. From our previous onsiderations we dedu e the identities

Theorem 2 ([18℄) For any n 2 N there exists a at family of elimina-

3.2 Cir uit en oding of polynomials

Denition 1 Let be given a sequen e of points 1 ; : : : ; m 2 kn and let

Lemma 3 ([20, 23℄) Let the notation be as before. Let M k be

Lemma 4 Let notations be as before and let m := 6(2L + n + 1)2 . Sup-

niteness of the limits mentioned before is modelised by the requirement

Theorem 5 Assume that the ground eld k is of hara teristi zero.

Proof Let be given a parametri elimination pro edure with asso-

There exists therefore a nonempty Zariski open subset U S , on-

We onsider this homomorphism as a spe ialization of the oordinate

Numeri al Analysis: 1995 (25th) AMS-SIAM Summer Seminar in Applied

Potrebbero piacerti anche

Ssmart Little Black Boxes

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Ssmart Little Black Boxes

Caricato da

Copyright:

Formati disponibili

1

Krone ker's smart, little bla k boxes

 are ommutative algebra and its modern derivate, namely todays

2.2 Flat families of elimination problems

3 The intrinsi omplexity of parametri elimination

Let us denote by Lk the oeÆ ient of T in the representation (3.2),

Lemma 1 The polynomials L1; : : : ; L2n are Q -linearly independent in

Proof Let us abbreviate N := 2n 1. We observe thatQfor 1  k  N + 1

holds. From our previous onsiderations we dedu e the identities

Theorem 2 ([18℄) For any n 2 N there exists a at family of elimina-

3.2 Cir uit en oding of polynomials

De nition 1 Let be given a sequen e of points 1 ; : : : ; m 2 kn and let

Lemma 3 ([20, 23℄) Let the notation be as before. Let M  k be

Lemma 4 Let notations be as before and let m := 6(2L + n + 1)2 . Sup-

niteness of the limits mentioned before is modelised by the requirement

Theorem 5 Assume that the ground eld k is of hara teristi zero.

Proof Let be given a parametri elimination pro edure  with asso-

There exists therefore a nonempty Zariski open subset U  S , on-

We onsider this homomorphism as a spe ialization of the oordinate

Numeri al Analysis: 1995 (25th) AMS-SIAM Summer Seminar in Applied

Potrebbero piacerti anche

are ommutative algebra and its modern derivate, namely todays

Proof Let us abbreviate N := 2n 1. We observe thatQfor 1 k N + 1

Denition 1 Let be given a sequen e of points 1 ; : : : ; m 2 kn and let

Lemma 3 ([20, 23℄) Let the notation be as before. Let M k be

Proof Let be given a parametri elimination pro edure with asso-

There exists therefore a nonempty Zariski open subset U S , on-