doi:10.1006/jsco.2000.

0402
Available online at http://www.idealibrary.com on
J. Symbolic Computation (2001) 31, 367408
Decomposition Plans for Geometric Constraint
Systems, Part I: Performance Measures for CAD
CHRISTOPH M. HOFFMAN

, ANDREW LOMONOSOV

AND MEERA SITHARAM

Computer Science, Purdue University, West Lafayette, IN 47907, U.S.A.

CISE, University of Florida, Gainesville, FL 32611-6120, U.S.A.

A central issue in dealing with geometric constraint systems for CAD/CAM/CAE is the
generation of an optimal decomposition plan that not only aids ecient solution, but also
captures design intent and supports conceptual design. Though complex, this issue has
evolved and crystallized over the past few years, permitting us to take the next important
step: in this paper, we formalize, motivate and explain the decompositionrecombination
(DR)-planning problem as well as several performance measures by which DR-planning
algorithms can be analyzed and compared. These measures include: generality, valid-
ity, completeness, ChurchRosser property, complexity, best- and worst-choice approx-
imation factors, (strict) solvability preservation, ability to deal with underconstrained
systems, and ability to incorporate conceptual design decompositions specied by the
designer. The problem and several of the performance measures are formally dened here
for the rst timethey closely reect specic requirements of CAD/CAM applications.
The clear formulation of the problem and performance measures allow us to precisely
analyze and compare existing DR-planners that use two well-known types of decom-
position methods: SR (constraint shape recognition) and MM (generalized maximum
matching) on constraint graphs. This analysis additionally serves to illustrate and pro-
vide intuitive substance to the newly formalized measures.
In Part II of this article, we use the new performance measures to guide the develop-
ment of a new DR-planning algorithm which excels with respect to these performance
measures.
c 2001 Academic Press
1. Introduction and Motivation
Geometric constraints are at the heart of computer aided engineering applications (see,
e.g. Homann and Rossignac, 1996; Homann, 1997), and also arise in many geometric
modeling contexts such as virtual reality, robotics, molecular modeling, teaching geom-
etry, etc. Both parts of this paper, however, deal with geometric constraint systems
primarily within the context of product design and assembly. Figure 3 illustrates those
(boldface) components within a standard CAD/CAM/CAE master model architecture
(Bronsvoort and Jansen, 1994; Kraker et al., 1997; Homann and Joan-Arinyo, 1998)
that Parts I and II of this paper directly address.

Supported in part by NSF Grants CDA 92-23502 and CCR 95-05745, and by ONR Contract N00014-
96-1-0635.

Supported in part by NSF Grant CCR 94-09809.

Corresponding Author: E-mail: [email protected]

07477171/01/040367 + 42 $35.00/0 c 2001 Academic Press
368 C. M. Homan et al.
Note. Throughout this manuscript, slant is used for emphasis, while italics are used for
introducing new terminology.
Informally, a geometric constraint problem consists of a nite set of geometric objects
and a nite set of constraints between them. The geometric objects are drawn from a xed
set of types such as points, lines, circles and conics in the plane, or points, lines, planes,
cylinders and spheres in three dimensions. The constraints are spatial and include logical
constraints such as incidence, tangency, perpendicularity, etc., and metric constraints
such as distance, angle, radius, etc. The spatial constraints can usually be written as
algebraic equations whose variables are the coordinates of the participating geometric
objects.
A solution of a geometric constraint problem is a real zero of the corresponding alge-
braic system. In other words, a solution is a class of valid instantiations of the geometric
elements such that all constraints are satised. Here, it is understood that such a solution
is in a particular geometry, for example the Euclidean plane, the sphere, or Euclidean
thrree-dimensional space. For recent reviews of the extensive literature on geometric
constraint solving see, e.g. Homann et al. (1998), Kramer (1992) and Fudos (1995).
1.1. the main reason to decompose geometric constraint systems
Currently there is a lack of eective spatial variational constraint solvers and assem-
bly constraint solvers that scale to large problem sizes and can be used interactively
by the designer as conceptual tools throughout the design process. Almost all current
CAD/CAM systems primarily use a non-variational, history-based three-dimensional
constraint mechanism. This basic inadequacy in spatial constraint solving seriously hin-
ders progress in the development of intelligent and agile CAD/CAM/CAE systems.
One governing issue is eciency: computing the solution of the non-linear algebraic
system that arises from geometric constraints is computationally challenging, and except
for very simple geometric constraint systems, this problem is not tractable in practice
without further machinery. The so-called constraint propagation-based solvers (e.g. Gao
and Chou, 1998a; Klein, 1998) generally suer from a drawback that cannot be easily
overcome: they nd it dicult to decompose cyclically constrained systems, an essential
feature of variational problems. Direct approaches to algebraically processing the entire
system include:
(1) standard methods for polynomial ideal membership and locating solutions in alge-
braically closed elds, for example using Gr obner bases (Ruiz and Ferreira, 1996)
or the WuRitt method (Chou et al., 1996; Gao and Chou, 1998b);
(2) numerous algorithms and implementations for solving over the reals based on the
methods of, for example, Canny (1993), Renegar (1992), Collins (1975) and Lazard
(1981), etc.; and
(3) algorithms for decomposing and solving sparse polynomial systems based on Canny
and Emiris (1993), Sturmfels (1993), Khovanskii (1978), Sridhar et al. (1993) and
Sridhar et al. (1996), etc.
These direct algebraic solvers deal with general systems of polynomial equations, i.e.
they do not exploit geometric domain knowledge; partly as a result, they have at least
exponential time complexity and they are slow in practice as well. In addition, they do not
Decomposition of Geometric Constraints I 369
take into account design considerations and as a result, cannot assist in the conceptual
design process. These drawbacks apply to direct numerical solvers as well, including those
that use homotopy continuation methods, see, for example, Durand (1998). The problem
would be further compounded if we allowed constraints that are natural in the design
process, but which must be expressed as inequalities, such as point P is to the left
of the oriented line L in the plane. Such additions would necessitate using cylindrical
algebraic decomposition-based techniques (Collins, 1975), such as Grigorev and Vorobjov
(1988), Lazard (1991) and Wang (1993) which have a theoretical worst-case complexity
of O(2
n
2
), where n is the algebraic size of the problem; or alternatively require the use
of non-linear optimization techniques, all of which are slow enough in practice that they
do not represent a viable option for large problem sizes.
With regard to eciency, the following rule of thumb has therefore emerged from years
of experimentation with geometric, spatial constraint solvers in engineering design and
assembly: the use of direct algebraic/numeric solvers for solving large subsystems renders
a geometric constraint solver practically useless (see Durand, 1998, for a natural example
of a geometric constraint system with six primitive geometric objects and 15 constraints,
which has repeatedly deed attempts at tractable solution). The overwhelming cost in
geometric constraint solving is directly proportional to the size of the largest subsystem
that is solved using a direct algebraic/numeric solver. This size dictates the practical
utility of the overall constraint solver, since the time complexity of the constraint solver
is at least exponential in the size of the largest such subsystem.
Therefore, the constraint solver should use geometric domain knowledge to develop a
plan for decomposing the constraint system into small subsystems, whose solutions can
be recombined by solving other small subsystems. The primary aim of this decomposition
plan is to restrict the use of direct algebraic/numeric solvers to subsystems that are as
small as possible. Hence the optimal or most ecient decomposition plan would minimize
the size of the largest such subsystem. Any geometric constraint solver should rst solve
the problem of eciently nding a close-to-optimal decompositionrecombination (DR)
plan, because that dictates the usability of the solver. Finding a DR-plan can be done as
a pre-processing step by the constraint solver: a robust DR-plan would remain unchanged
even as minor changes to numerical parameters or other such on-line perturbations to
the constraint system are made during the design process.
In addition to optimality (eciency), other equally important (and sometimes com-
peting) issues arise from the fact that a DR-plan is a key conceptual component of the
CAD model and should aid the overall design or assembly process. These issues will be
discussed under Desirable characteristics later in this section.
A clean and precise formulation of the DR-planning problem is therefore a fundamental
necessity. To our knowledge, despite its longstanding presence, the DR-planning problem
has not yet been clearly isolated or precisely formulated, although there have been many
prior, specialized DR-planners that utilize geometric domain knowledge (e.g. Serrano
and Gossard, 1986; Crippen and Havel, 1988; Serrano, 1990; Havel, 1991; Owen, 1991,
1996; Kramer, 1992; Ait-Aoudia et al., 1993; Pabon, 1993; Homann and Vermeer, 1994,
1995; Bouma et al., 1995; Hsu, 1996; Fudos and Homann, 1996b, 1997; Latham and
Middleditch, 1996). See Homann et al. (1998) for an exposition of two primary classes of
existing methods of decomposing geometric constraint systems; representative algorithms
from these two classes are extensively analyzed in Section 4. In Part II of this paper we
370 C. M. Homan et al.
describe and analyze a new algorithm that performs well with respect to the measures
described below.
In the next two subsections we informally describe both the basic requirements of a
DR-plan(ner) that dictate its overall structure, as well as desirable characteristics of the
DR-plan(ner) that improve eciency and assist in the design process.
1.2. basic requirements of a decomposition plan
Recall that a DR-plan species a plan for decomposing a constraint system into small
subsystems and recombining solutions of these subsystems later. Therefore the rst re-
quirement of a DR-plan is that the solutions of the small subsystems in the decomposition
can be recombined into a solution of the entire system. In other words, it should be pos-
sible to substitute the (set of) solution(s) of each subsystem into the entire system in
a natural manner, resulting in a simpler system. Secondly, we would like these inter-
mediate subsystems that are solved during the decomposition and recombination to be
geometrically meaningful.
Together, these two requirements on the intermediate subsystems translate to a re-
quirement that the subsystems be geometrically rigid. A rigid or solvable subsystem of
the constraint system is one for which the set of real-zeroes of the corresponding algebraic
equations is discrete (i.e. the corresponding real-algebraic variety is zero-dimensional),
after the local coordinate system is xed arbitrarily, i.e. after an appropriate number
of degrees of freedom D are xed arbitrarily. The constant D is usually the number of
(translational and rotational) degrees of freedom available to any rigid object in the given
geometry (three in two dimensions, six in three dimensions, etc.) and in some cases, D de-
pends on other symmetries of the subsystem. An underconstrained system is not solvable,
i.e. its set of real zeroes is not discrete (non-zero-dimensional). A well-constrained system
is a solvable system where removal of any constraint results in an underconstrained sys-
tem. An overconstrained system is a solvable system in which there is a constraint whose
removal still leaves the system solvable. Solvable systems of equations are therefore well-
constrained or overconstrained. That is, the constraints force a nite number of isolated
real solutions, so one solution cannot be obtained by an innitesimal perturbation of
another. For example, Figure 1 shows a solvable subsystem of three points and three
distances between pairs of points. A consistently overconstrained system is one which
has at least one real zero.
Note. It is important to distinguish solvable from has a real solution. Although
(inconsistently) overconstrained (or even certain well-constrained) systems may have no
real solutions at all, by our denition, since their set of real zeroes is discrete, they would
still be considered solvable. In general, whenever a subsystem of a constraint system is
detected that has no real solutions, then the solution process would have to immediately
halt and inform the designer of this fact.
Informally, a geometric constraint solver which solves a large constraint system E by
using a DR-plannerto guide a direct algebraic/numeric solver capable of solving small
subsystemsproceeds by repeatedly applying the following three steps at each iteration i.
(1) Find a small solvable subsystem S
i
of the (current) entire system E
i
(at the rst
iteration, this is simply the given constraint system E, i.e. E
1
= E). This step is
Decomposition of Geometric Constraints I 371
A
B
C
D: (x
1
x
2
)
2
+ (y
1
y
2
)
2
A
2
E: (x
2
x
3
)
2
+(y
2
y
3
)
2
B
2
P
3
F: (x
3
x
1
)
2
+(y
3
y
1
)
2
C
2
P
1
P
2
Figure 1. A solvable system of equations.
E E = T (E ) E = T (E ) E = T T (E )
n 1 1 1 2 2 3
2
1 1 1
S
T (S )
S
S
T (S )
T (T (S ))
S
1 1 1
2
3
2 2
2 1 1
n . . .
Figure 2. Solving a well-constrained system by decomposition: Step 1rectangles, Step 3ovals.
indicated by a rectangle in Figure 2. The subsystem S
i
could be also chosen by the
designer.
(2) Solve S
i
using the direct algebraic/numeric solver.
(3) Using the output of the solver, and perhaps using the designers help to eliminate
some solutions, replace S
i
by an abstraction or simplication T
i
(S
i
) thereby replac-
ing the entire system E
i
by a simplication T
i
(E
i
) = E
i+1
. This step is indicated
by an oval in Figure 2. Some informal requirements on the simpliers T
i
are the
following: we would like E
i
to be (real algebraically) inferable from E
i+1
; i.e. we
would like any real solution of E
i+1
to be a solution of E
i
as well.
A constraint solver that ts the above structural descriptionwhich we shall refer to
as S in future discussionsis called a decompositionrecombination (DR) solver. (The
formal denition imposes further requirements on simplier T
i
see the next section.)
This solver terminates when the small, solvable subsystem S
i
found in Step 1 is the entire
polynomial system E
i
. An optimal DR-plan will minimize the size of the largest S
i
. If
the whole system is underconstrained, the solver should still solve its maximal solvable
subsystems.
For the purpose of planning a solution sequence a priori, we would like to execute
Steps 1 and 3 alone without access to the algebraic solver, and obtain a DR-plan, without
actually solving the subsystems. That is, i.e. we would like the constraint solver to look
as in Figure 3, with the DR-planner driving the direct algebraic/numeric solver.
372 C. M. Homan et al.
model
model
model
Master
constraint
model
Master
View/system
GD&T
plan
View/system
Other
downstreams
GD&T MPP
model
MPP
model
constr. constr.
Master model server
GD&T
CAD
View/
system
Constraint solver
Constraint model manipulator
Decomp.
planner
Algebraic
solver
CAD model
Optional design history
of constraint models
(for non-variational case)
CAD
constraint
model
decomposition
designers conceptual
consistent with
Decomposition
Constraint
Manuf. proc.
Figure 3. CAD/CAM/CAE master model architecture; this paper deals with boldface components.
To generate a DR-plan a priori, one would have to locate a solvable subsystem S
i
,
and without actually solving it, nd a suitable abstraction or simplication of it that is
substituted into the larger system E
i
to obtain an overall simpler system E
i+1
in Step 3.
On the other hand, such a DR-plan would possess the advantage of being robust, or
generically independent of particular numerical values attached to the constraints, and
of the solution to the S
i
, and usually only depends on the degrees of freedom of the
relevant geometric objects and geometric constraints.
1.3. desirable characteristics of DR-planners for CAD/CAM
We enumerate and informally describe a set C of natural characteristics desirable for a
DR-planner. These will guide the formal denition of the DR-planning problem and the
performance measures in the next section as well as the design of the new DR-planner
in Part II of this paper. We begin with four criteria that directly follow from the overall
structural description of a typical DR-planner S in the previous subsection.
(i) The DR-planner should be general, i.e. it should not be articially restricted to a
narrow class of decomposition plans; it should output a DR-plan if there is one, and
it should be able to decompose underconstrained systems as well. Furthermore, if
a DR-plan exists, the planner should run to completion irrespective of how and in
what order the solvable subsystems S
i
are chosen for the plan (a ChurchRosser
property).
Decomposition of Geometric Constraints I 373
(ii) The planner should potentially output a close-to-optimal DR-plan (i.e. where the
size of the largest solvable subsystem S
i
is close-to-minimal). This dictates eciency
of solution of the constraint system.
(iii) The DR-planner should be fast and simple; the time complexity should be low,
the planner should be fast in practice, easily implementable, and compatible with
existing algebraic solvers, CAD systems, constraint models and manipulators.
(iv) The planner should utilize and take advantage of geometric domain knowledge, as
well as other special properties of geometric constraints arising from the relevant
design or assembly application or, in some situations, from a downstream manufac-
turing application.
Besides critically aecting the speed of constraint solving, a good DR-plan is a key compo-
nent of the constraint model which participates in the overall process of design/assembly.
This is especially so, since the constraint system is a component of the CAD model which
the designer directly interacts with, and moreover, a DR-plan is nothing but a hierarchi-
cal, structural decomposition of the geometric constraint system. Therefore, maintaining
a robust DR-planwhich reects design intent at every level of renementis invaluable
in improving eciency, exibility and transparency in the overall design process. These
properties are crucial for intelligent CAD systems to facilitate designer interaction at a
conceptual, early-design phase; in fact, an eective CAD system based on spatial, varia-
tional constraints to be eective must facilitate early-design interaction. This motivates
adding the following desirable characteristics to the set C.
(v) The DR-plan should be consistent with the design intent: in particular, the designer
often has a multi-layered or hierarchical conceptual, design decomposition in mind,
reecting features or conglomerates of features in the case of product design and
parts or subassemblies in the case of assembly. See e.g. Klein (1996), Mantyla et
al. (1989), and Middleditch and Reade (1997). The designer would typically wish
to further decompose the components of her design decomposition, as well as treat
these components (recursively) as units that can be independently manipulated. For
example, the geometric objects within a component are manipulated with respect
to a local coordinate system, and a component as a whole unit is manipulated with
respect to a global (next level) coordinate system. The DR-plan should therefore be
a consistent extension and/or renement of this conceptual design decomposition.
(vi) While the DR-plan is used to guide the algebraic solver, it should remain unaected
or adapt easily to the designers intervention in the solution process, which is valu-
able for pruning combinatorial explosion: the designer can throw out meaningless
or undesirable solutions of subsystems at an early stage. Such designer interference
is also crucial for avoiding the use of inequality constraints: for example, instead of
adding on a constraint that point P is to the left of line L, the designer can simply
throw out any partial solutions that put P to the right of line L.
(vii) The DR-plan for solving a geometric constraint system should remain meaningful
in the presence of constraints other than geometric constraints, such as equational
constraints or parametric constraints expressing design intent.
Finally, the CAD system and the CAD model do not stand alone. In standard client
server based architectures (see, e.g. Bronsvoort and Jansen, 1994; Kraker et al., 1997;
Homann and Joan-Arinyo, 1998), the CAD model is just one clients view of the product
374 C. M. Homan et al.
master model (Newell and Evans, 1976; Semenkov, 1976), with which it has to be contin-
ually coordinated and made consistent. The master model in turn coordinates with other
downstream production client systems which maintain other consistent views. These
clients include geometric dimensioning and tolerancing systems (GD&T), and manufac-
turing process planners (MPP) for automatically controlled machining or assembly. Each
client view contains potentially proprietary information that must be kept secure from
the master model. Figure 3 illustrates this architecture and those parts that this paper
directly deals with.
Each client view contains its own internal view of the constraint model and therefore
coordination and consistency checks between the various views crucially involve, and
vice versa may aect, the DR-plan. This leads to the following additions to the set C of
desirable characteristics for DR-planners.
(viii) The DR-plan should be as robust as possible (see, e.g. Fang, 1992) to on-line changes
made to the constraint system, and the DR-planner should be able to quickly
alter the DR-plan in response to such changes. In particular, the DR-plan should
ideally not change at all with numerical perturbations (within large ranges) to the
constraints, thereby permitting the DR-plan to be computed as a pre-processing
step. Addition, deletion, and modication of constraints and geometric objects to
the constraint system occurs in numerous circumstances during the design process.
For example:
(a) in the process of solving the system using the DR-plan, a subsystemS
i
deemed
solvable by a degree of freedom analysis may be found to have no real solutions, or
may, in a degenerate case, turn out to be underconstrained or have innitely many
solutions, preventing a continuation of the solution process S;
(b) in history-based CAD systems, the specication of the product takes the form
of a progressive sequence of changes being made to successive partial specications
and the associated partial constraint systems;
(c) inferred or understood constraints are often added to an incomplete constraint
specication at the discretion of the CAD system;
(d) in underconstrained situations, as might occur in designing exible or mov-
ing parts and assemblies, the mapping of the conguration space of the resulting
mechanism would require the repeated addition of constraints at the discretion of
the constraint solverthe solutions to the resulting well-constrained systems would
represent various congurations of the mechanism;
(e) for a variational spatial constraint solver to be eective or even usable, it
would have to rely on extensive interactive constraint changes being made by the
designer, sometimes in the course of solving; and
(f) nally, when one of the various clients in Figure 3 makes a change to its own
view of the constraint model, this will result in consistency updates to the master
constraint model which will, in turn, result in updates to the other views of the
constraint model.
(ix) The DR-plan should isolate overconstrained subsystems which arise in assembly
problems; furthermore, with multiple (possibly proprietary) views of the constraint
model being kept by various clients as in Figure 3, constraint reconciliation often
takes place and the DR-plan should facilitate this process, and viceversa, should be
robust against this process. For a precise description of the constraint reconciliation
problem, see Homann and Joan-Arinyo (1998). The problem requires isolation
Decomposition of Geometric Constraints I 375
of overconstrained subsystems and is compounded in the case of non-variational,
history-based design.
main results and organization
Note. It should be noted that in this paper the emphasis is on the boldface components
of Figure 3: in particular, no emphasis will be placed on the direct algebraic/numeric
solver that is used to solve the small subsystems specied by the DR-plan. Furthermore,
in most of our discussion in both Parts I and II, we will restrict ourselves to variational
spatial constraints. That is, we will not explicitly deal with history-based constraint
mechanisms, although several of the desirable characteristics in C, apply indirectly or
directly (e.g. (ix)) to history-based constraint mechanisms as well.
In Sections 2 and 3, we precisely formulate the decompositionrecombination (DR)
problem and lay down formal performance measures that reect several of the nine char-
acteristics C of DR-planners given above. These are not restricted to standard complexity
measures. As noted before, to our knowledge, despite its longstanding presence, and the
existence of prior DR-planners based on various ad hoc methods, the general DR-planning
problem has not yet been clearly isolated or precisely formulated.
In Section 2, we formally dene DR-solvers and their performance measures in the
general context of polynomial systems arising from geometric constraints.
In Section 3, we give a parallel set of analogous denitions of DR-planners and their
performance measures in the context of constraint graphs that incorporate geometric
degrees of freedom.
In Section 4, we use the new performance measures to formally analyze two primary
types of existing decomposition methods that are based on constraint graphs and degree
of freedom analysis: constraint shape recognition (SR) and generalized maximum match-
ing (MM). This analysis additionally provides intuitive substance to the newly dened
performance measures.
Note. In Part II of this paper we describe and formally analyze three new DR-planners
that were designed based on the performance measures described here. The development
culminates with a new DR-planner called the modied frontier algorithm which excels
with respect to these new performance measures.
2. Formal Denition of DR-solvers Using Polynomial Systems
This section will develop adequate notation for formally dening all of the terms
that appear in italics and have been used informally so far. We will formally state the
decompositionrecombination (DR) problem for polynomial systems arising from geomet-
ric constraints. This requires formalizing the notion of a decompositionrecombination
solution sequence, and of decompositionrecombination solvers that t the description
S given in the previous section. In addition, we dene the performance measuresthat
capture some of the desirable properties C given in the last sectionfor comparing such
sequences and solvers.
Note. As mentioned in the organizational section, for the sake of gradual exposition,
we will rst assume that the constraint system is being solved even as the decomposition
376 C. M. Homan et al.
is being generatedi.e. the DR-solver exactly ts the structural description S given in
the previous section. Furthermore, we will dene the DR-solver in the general context of
polynomial equations that arise from the geometric constraints.
In Section 3 we will shift our attention to the DR-planning problem. The DR-planner
generates a decomposition plan a priori, which then drives the direct algebraic/numeric
solvertogether, they form a DR-solver. The DR-planner, however, will be dened in
the context of constraint graphs that incorporate geometric degrees of freedom. The
DR-planner and its performance measures will be analogous to the italics terms dened
here.
In order to formally dene DR-solvers and their performance measures, we need to
specify the model of real-algebraic computation used by the algebraic/numeric solver in
Figure 3. However, the issue of the model or how real numbers are represented is en-
tirely outside the focus of this manuscript for the following reason: our DR-solvers and
performance measures are robust in that they are conceptually independent of the alge-
braic/numeric solver being used or how real numbers are represented. The denition of
DR-solvers and performance measures adapts straightforwardly to other natural models
of computation. In other words, our conceptual denition of DR-solvers and performance
measures are such that if DR-solver A performs better than DR-solver B with respect to
one of our performance measures, then it continues to do so irrespective of the (natural)
model of computation being used by the algebraic/numeric solver.
Further, as pointed out in an earlier note, this paper only emphasizes the boldface
components of Figure 3, specically the DR-planner, whose operation dened in Section 3
will be seen to be purely combinatorial. In particular, no emphasis is placed on the non-
boldface components, which includes the algebraic/numeric solver, and the model of
computation it uses.
Having noted the above, the model of computation we assumefor the the sake of
completeness and formality of denitions in this manuscriptis the BlumShubSmale
model of real algebraic computation (Blum et al., 1989). Briey, in this model, real num-
bers are assumed to be representable as such as entities, without any recourse to rational
approximations and nite precision or interval arithmetic. Real arithmetic operations
such as multiplication and addition and division can be performed in constant time, and
nding an unambiguous representation of each real zero of a univariate polynomial p can
also be achieved in time polynomial in the degree of p. Furthermore, these unambiguous
representations of real zeroes of univariate polynomials are treated as entities just like
any other real number, and can for instance, be used as coecients of other polynomials.
A system of equations E is a pair (P, X) where P is a set of polynomial equations
with real coecients, and X is a set of formal indeterminates. The union of two systems
E
1
= (P
1
, X
1
) and E
2
= (P
2
, X
2
) is the system (P
1
P
2
, X
1
X
2
). An intersection
E
1
E
2
is the system (P
1
P
2
, X
1
X
2
).
Within the geometric context, a solved subsystem is a rigid or solvable system where
all the variables have already been solved for, i.e. they have typically been expressed
explicitly as polynomials of D free variables, where D represents the number of degrees of
freedom of the rigid body (within its local coordinate system) in the prevailing geometry.
In some cases, for example when the rigid body is a line in three dimensions, the number
of free variables, including redundant ones, may be greater than D.
Let E = (P, V ) with V = {y
1
, . . . , y
D
} X be such that every equation in P is of
type x
j
= F
j
(y
1
, . . . , y
D
), for all x
j
X and F
j
is a rational function (of the form
Decomposition of Geometric Constraints I 377
Q
1
/Q
2
where Q
1
and Q
2
are, as usual, polynomials with real coecients) that can be
evaluated in time polynomial in (|V |, |P|). Such a system of equations, E, is called a
solved system. The variables y
1
, . . . , y
D
V of the solved system are free variables of E,
and the variables x
1
, . . . , x
k
V are explicitly xed variables. Note that all the variables
in a solved system are xed or free, whereas a general solvable system may have free,
explicitly xed and other, implicitly xed variables.
A typical DR-solverthat follows the overall structural description S of Section 1.2
obtains a sequence E
1
, . . . , E
m
, consisting of successively simpler solvable systems. These
are general solvable systems and have successively fewer implicitly xed variables; E =
E
1
; and E
m
is a solved system. New variables y
i
may be introduced at intermediate stages
which represent free variables within subsystems S
i
that are solved with respect to these
variables. These solved subsystems represent various rigid bodies located and xed with
respect to their local coordinate systems. However, these local coordinate systems are still
constrained with respect to each other, and hence in fact only D of the newly introduced
variables are, in eect, free and the remainder are implicitly xed. Some of these newly
introduced variables may be removed at later stages. Eventually, in the solved system
E
m
, all the original variables and those newly introduced variables that might remain
become explicitly xed with respect to D free variables as well. The set of real solutions
to E
i+1
should also be a subset of solutions to E
i
, to ensure that the nal solutions to
E
m
actually represent solutions to the original system E.
We formally dene real algebraic equivalence and real algebraic inference of two ge-
ometric constraint systems as follows. Given two systems E
1
= (P
1
, X Y
1
) and E
2
=
(P
2
, XY
2
)where the set X represents the original variables that are currently implic-
itly or explicitly xed, and the sets Y
i
represent the newly introduced variables (that are
free within the solved subsystems S
i
), we say that the system E
1
= (P
1
, XY
1
) is real
algebraically inferable (in short, inferable) from the system E
2
= (P
2
, X Y
2
), if for any
real solution Y
2
= A
2
; X = B that satises the equations in P
2
, there is a corresponding
assignment A
1
of real values to Y
1
such that Y
1
= A
1
; X = B satises P
1
. Two systems
E
1
and E
2
are real algebraically equivalent if E
1
is real algebraically inferable from E
2
and E
2
is real algebraically inferable from E
1
.
Now we are ready to dene the notion of a DR-solution sequence. Let E be a system
of equations.
A DR-solution sequence of E is a sequence of systems of equations E
1
, . . . , E
m
such
that E = E
1
, E
m
is a solved system (so E
m
has a real solution), every E
i
is solvable,
each E
i
is inferable from E
i+1
. Any solvable system E which has a real solution in fact
has a DR-solution sequence. A trivial DR-solution sequence E, E

where E

is a solved
system equivalent to Ewill do. (Note that by E
i
we denote abstract algebraic systems,
rather than their computer representations that could only have rational coecients and
therefore may only have approximate DR-solution sequences.)
The DR-problem is the problem of nding a DR-solution sequence of a given constraint
system. A DR-solver is a constraint solver or algorithm that solves the DR-planning
problem. A DR-solver is general if it always outputs a DR-solution sequence when given
a solvable system as input. Aside from being general, we would also like a DR-solver
to have the ChurchRosser property, i.e. the DR-solver should terminate irrespective
of the order in which the solvable subsystems S
i
are chosen. In other words, at each
step, a solvable subsystem S
i
can be chosen greedily to satisfy the (easily checkable)
requirements of the algorithm. This prevents exhaustive search.
378 C. M. Homan et al.
Next, we formally dene a set of performance measures for the DR-solution sequences
and DR-solvers. These performance measures are designed to capture the characteristics
C of constraint solvers given in Section 1.3, that are desirable for engineering design
and assembly applications. (Note that many of these measures are Boolean, i.e. either
a certain desirable condition is met or not.) In particular, we would like a DR-solution
sequence Q = E
1
, . . . , E
m
of a solvable system E to have several properties. To describe
these properties we formalize a simplifying map from each system E
i
to its successor
E
i+1
. In fact, for generality, we choose these maps T
i
called subsystem simpliersto
map the set of subsystems of E
i
onto the set of subsystems of E
i+1
.
First, in order to reect the ChurchRosser property in (i), and points (iv) and (vi)
of C, we would like these subsystem simpliers T
i
to be natural and well-behaved, i.e. to
obey the following simple and non-restrictive rules.
(1) If A is a subsystem of B, then T
i
(A) is a subsystem of T
i
(B).
(2) T
i
(A) T
i
(B) = T
i
(A B).
(3) T
i
(A) T
i
(B) = T
i
(A B).
Second, in the description of the typical DR-solver S given in Section 1.2, each system
E
i+1
in the DR-solution sequence is typically to be obtained from E
i
by replacing a
solvable subsystem S
i
in E
i
(located during Step 1 of S), by a simpler subsystem (during
Steps 2 and 3). For a manipulable DR-solution sequence (again satisfying points (i), (iv),
(v) and (vi) of C), we would like E
i+1
to look exactly like E
i
outside of S
i
. This leads to
another set of properties desirable for the subsystem simpliers T
i
.
(4) Each E
i
= S
i
R
i
U
i
, 1 i m, where S
i
is solvable, R
i
is a maximal subsystem
such that S
i
and R
i
do not share any variables, S
i
, R
i
, U
i
do not share any equations,
and all variables of U
i
are either variables of S
i
or R
i
. For any A R
i
, T
i
(A) = A.
Thirdly, in order to address points (iv)(vi), and (i) of C simultaneously, i.e. permitting
generality of the subsystem S
i
that is replaced by a simpler system during the ith step,
while at the same time making it convenient for the designer to geometrically follow
and manipulate the decomposition, we would like the subsystem simpliers to satisfy the
following property.
(5) For each i, all the pre-images of S
i
, T
1
j
. . . T
1
i1
(S
i
), 1 j i 1, are algebraically
inferable from S
i
, and furthermore, they are solvable or rigid subsystems for the
given geometry (recall that the denition of solvable depends on the geometry).
It follows from (2) and (3) above that the inverse T
1
i
(A) =

B where the union

is taken over all B E
i
, such that T
i
(B) A.
The above property states that while the subsystem simpliers enjoy a high degree
of generality and are completely free to map solvable subsystems into solvable systems,
they should never map (convert) subsystems that are originally not solvable into one of
the chosen, solvable systems S
i
, at any stage i. In other words, in the act of simplify-
ing, the subsystem simpliers should not create one of the solvable subsystems S
i
out
of subsystems that were originally not solvable. A DR-solution sequence that satises
the above properties is called valid. Thus a valid DR-solution sequence for a geometric
constraint system E is specied as a sequence of E
1
, . . . , E
m
such that E
1
= E, E
m
is a
Decomposition of Geometric Constraints I 379
solved system, every E
i
is solvable and inferable from E
i+1
, every E
i
= S
i
R
i
U
i
as
described above.
The motivation given for each of the properties above makes it clear that valid DR-
solution sequences encompass highly general but geometrically meaningful solutions of
the original constraint system E. Next we turn to point (ii) of C, i.e. optimality, which
also competes with generality (point (i)). For optimality, we would like to minimize the
size of the largest solvable subsystem S
i
in the DR-solution sequence. Formally, the size
of the DR-solution sequence Q is equal to max
1im
|S
i
|, where the size |S
i
| is equal
to the total number of its variables less the number of its explicitly xed variables. The
optimal DR-size of the algebraic system E is the minimum size of Q, where the minimum
is taken over all possible DR-solution sequences Q of E. An optimal DR-solution sequence
Q of E is a solution sequence such that the size of Q is equal to the optimal DR-size of
E. The approximation factor of the DR-solution sequence Q of the system E is dened
as the ratio of the optimal DR-size of E to the size of Q.
As a general rule for optimality, it is clear that the larger the choice for solvable
subsystems S
i
of E
i
available at any stage i, the more likely that one can nd a small
solvable S
i
in E
i
. In other words, we would like to make sure that the subsystem simplier
does not destroy solvability of too many subsystems starting from the original system E.
Note that while the denition of DR-solution sequence makes sure that the entire system
E
m
is solvable if E
1
is, it does not require the same for subsystems of E
i
. (In fact, even
if all the E
i
in a DR-solution sequence are algebraically equivalent, this would still not
imply that solvability is preserved for the subsystems.) In addition, while the denition
of a DR-solution sequence ensures that E
i+1
has a real solution if E
i
has one, it does
not ensure the same for subsystems of E
i
. The next two denitions capture properties
of subsystem simpliers that preserve subsystem solvability (resp. solutions) to varying
degrees, thereby helping the optimality of the DR-solver, i.e. (ii) of C. The DR-solution
sequence Q = E
1
, . . . , E
m
is solvability preserving if and only if for all A E
i
, A is
solvable (resp. has a real solution) and (A S
i
= or A S
i
) T
i
(A) is solvable
(resp. has a real solution). A DR solution sequence Q = E
1
, . . . , E
m
is strictly solvability
preserving if and only if for all A E
i
, A is solvable (resp. has a real solution) T
i
(A)
is solvable (resp. has a real solution). Requiring such a solvability preserving simplier
places a weak restriction on the class of valid DR-solution sequences, but on the other
hand, this restriction cannot eliminate all optimal DR-solution sequences. Furthermore,
solvability preservation helps to ensure the ChurchRosser property in (i) of C.
Note. To be more precise, solvability preservation should be termed solvability and
solution-existence preservation, but we choose the shorter phrase.
In fact, for DR-solution sequences to be optimal, we would prefer that for all i 1,
S
i
does not contain any strictly smaller subsystem, say B that is solvable, and has not
been found (during Step 1 of the description S in Section 1.2) and simplied/replaced
at an earlier stage j < i. The DR-solution sequence Q = E
1
, . . . , E
m
is complete if and
only if for every non-trivial solvable B S
i
, B = T
i1
T
i2
. . . T
j
(S
j
) for some j i 1.
While the completeness requirement restricts the class of valid DR-solution sequences,
it only eliminates sequences that either have size greater than optimal or the same size
as some optimal sequence that does simplify B. In addition to aecting optimality, i.e.
(ii) of C, completeness also strongly reects (ix): completeness prevents a DR-solver from
380 C. M. Homan et al.
overlooking an overconstrained subsystem inside a well-constrained subsystem, which is
also useful for constraint reconciliation (see Homann and Joan-Arinyo, 1998).
So far, we have discussed performance measures that measure the desirability of DR-
solution solution sequences. Next we formally dene directly analogous performance mea-
sures for DR-solvers which generate DR-solution sequences. A DR-solver A is said to be
valid, solvability preserving, strictly solvability preserving, complete if and only if for every
input constraint system E, every DR-solution sequence produced by A is valid, solvability
preserving, strictly solvability preserving or complete, respectively.
Note. Purely as a tool to help analysis and exposition, often we assume that DR-solvers
for producing DR-solution sequences are randomized or non-deterministic in a natural
way, i.e. those steps in the algorithm where arbitrary choices of equations or variables
are made (for example, to be the lowest numbered equation or variable) are now taken
to be randomized or non-deterministic choices.
The next denition formalizes performance measures related to characteristic (ii) of
C that dierentiates randomized DR-solvers for which some random choice leads to an
optimal DR-solution sequence, from inferior DR-solvers where no choice would lead to
an optimal DR-solution sequence. The worst-choice approximation factor of a DR-solver
A on input system E is the minimum of the approximation factors of all DR-solution
sequences Q of E obtained by the algorithm A over all possible random choices. The
best-choice approximation factor of the algorithm A on input E is the maximum of the
approximation factors of all the DR-solution sequences Q of E obtained by the algorithm
A over all possible random choices.
3. Formal Denition of a DR-planner Using Constraint Graphs
The DR-solution sequence dened in the previous section embeds a DR-plan that is
intertwined with the actual solution of the system. Therefore, a DR-solver that simply
outputs a DR-solution sequence (even one that is strictly solvability preserving, complete,
etc.), may not be modular as shown in Figure 3.
How does one construct a DR-solver that rst generates a DR-plan before using it to
drive the general algebraic/numeric solver? First note that repeatedly applying Steps 1
and 3 of the DR-solver in the description S in Section 1.2 could potentially generate a
DR-plan without actually solving any subsystem (and without applying Step 2), provided
that the following can be done:
(a) solvability of a subsystem S
i
of the system E
i
in Step 1 can be determined generi-
cally, without actually solving it; and
(b) a solvable subsystem S
i
can be replaced in Step 3 without actually solving S
i
i.e.
by a hypothetical solved subsystemto give the new system E
i+1
. For this we need
a consistent and adequate abstraction of the systems E
i
, and of their subsystems,
as well as a way to abstract a hypothetical solved subsystem. In other words, we
need to adapt the subsystem simplier maps described in the last section, so that
the iteration can proceed with Step 1 again.
By thus modifying the description S of the DR-solver, we obtain a DR-planner which
can be employed as a pre-processing step to output a DR-plan instead of a DR-solution
Decomposition of Geometric Constraints I 381
2
2
1
1 1
E
F D
2
Figure 4. A constraint graph.
sequence. This DR-plan can thereafter be used to direct a series of applications of Step 2,
leading to a complete DR-solution sequence. This modularity helps, for example, towards
manoeuverability by the designer (point (vi)), and towards compatibility of the DR-solver
with existing solvers (point (v)).
In order to formally dene such a DR-plan, we follow a common practice and view the
constraint system as the constraint hypergraph: this abstraction permits us to build the
DR-planner on the foundation of generalized degree of freedom analysis which is known to
work well in estimating generic solvability of constraint systems without actually solving
them. (This is explained more precisely after the formal graph-theoretic denitions are in
place.) Hence using constraint graphs facilitates both tasks (a) and (b) described above.
In other words a DR-planner that is based on a generalized degree of freedom analysis
is robust in the following sense: changing the numerical values of constraints is not
likely to aect the DR-plan. Thus the motivation for using constraint graphs includes
all of the desirable characteristics (iii) to (vi) of the set C in Section 1.3. In addition,
geometry is more visibly displayed via constraint graphs than via equations, thereby
helping interaction with the designer.
Note. DR-plans and planners and their performance measures could also be dened
directly in terms of the algebraic constraint systems, just as we dened DR-solvers.
In this paper, however, due to the reasons mentioned above, we dene DR-plans and
planners entirely within the context of constraint graphs and degree of freedom analysis.
3.1. constraint graphs and solvability
First we dene the abstraction that converts a geometric constraint system into a
weighted graph. Recall that a geometric constraint problem consists of a set of geometric
objects and a set of constraints between them.
A geometric constraint graph G = (V, E, w) corresponding to a geometric constraint
problem is a weighted graph with n vertices (representing geometric objects) V and
m edges (representing constraints) E; w(v) is the weight of vertex v and w(e) is the
weight of edge e, corresponding to the number of degrees of freedom available to an
object represented by v and the number of degrees of freedom removed by a constraint
represented by e, respectively. For example, Figure 4 is a constraint graph of a constraint
problem shown in Figure 1. Note that the constraint graph could be a hypergraph, each
hyperedge involving any number of vertices.
Now we introduce denitions that will help us to relate the notion of solvability of the
geometric constraint system to the corresponding geometric constraint graph.
382 C. M. Homan et al.
A subgraph A G that satises

eA
w(e) +D

vA
w(v) (1)
is called dense, where D is a dimension-dependent constant, to be described below. The
function d(A) =

eA
w(e)

vA
w(v) is called the density of a graph A.
The constant D is typically

d+1
2

where d is the dimension. The constant D captures

the degrees of freedom associated with the cluster of geometric objects corresponding
to the dense graph. In general, we use words subgraph and cluster interchangeably.
For planar contexts and Euclidean geometry, we expect D = 3 and for spatial contexts
D = 6, in general. If we expect the cluster to be xed with respect to a global coordinate
system, then D = 0.
A dense graph with density strictly greater than D is called overconstrained. A graph
that is dense and all of whose subgraphs (including itself) have density at most D is
called well-constrained. A graph G is called well-overconstrained if it satises the follow-
ing: G is dense, G has at least one overconstrained subgraph, and has the property that
on replacing all overconstrained subgraphs by well-constrained subgraphs, G remains
dense. A graph that is well-constrained or well-overconstrained is said to be solvable. A
dense graph is minimal if it has no proper dense subgraph. Note that all minimal dense
subgraphs are solvable, but the converse is not the case. A graph that is not solvable
is said to be underconstrained. If a dense graph is not minimal, it could in fact be an
underconstrained graph: the density of the graph could be the result of embedding a
subgraph of density greater than D.
In order to understand how solvable constraint graphs relate to solvable constraint sys-
tems it is important to remember that a geometric constraint problem has two aspects
combinatorial and geometric. The geometric aspect deals with actual parameters of the
geometric constraint problem, while the combinatorial aspect deals with only the abstrac-
tions of objects and constraints. Unfortunately, at the moment it is not known how to
completely separate the two aspects, except for some special cases. In this paper we will
limit ourselves to the geometric constraint problems where generally there is a correspon-
dence between solvable constraint systems and solvable constraint graphs. In order to do
that, we need to introduce and formalize the notion of generic solvability of constraint
systems. Informally, a constraint system is generically (un)solvable if it is (un)solvable for
most choices of coecients of the system. More formally we use the notion of genericity
of e.g. Cox et al. (1998). A property is said to hold generically for polynomials f
1
, . . . , f
n
if there is a non-zero polynomial P in the coecients of the f
i
such that this property
holds for all f
1
, . . . , f
n
for which P does not vanish.
Thus the constraint system E is generically (un)solvable if there is a non-zero poly-
nomial P in the parameters of the constraint systemsuch that E is (un)solvable when
P does not vanish. Consider for example Figure 5. Here the objects are six points in 2D
and the constraints are eight distances between them. This system is unsolvable since the
edge BC can be continuously displaced, unless the length of AD (induced by the lengths
of AE, DE, DF, AF and EF) is equal to |AB| + |BC| + |CD|. Since we can create an
appropriate non-zero polynomial P(AB, BC, . . . , EF), such that P() = 0 if and only if
|AD| = |AB| +|BC| +|CD|, this system is generically unsolvable.
While a generically solvable system always gives a solvable constraint graph, the con-
verse is not always the case. In fact, there are solvable, even minimal dense graphs whose
corresponding systems are not generically solvable, and are in fact generically not solv-
Decomposition of Geometric Constraints I 383
B
C
D
E
F
A
Figure 5. Generically unsolvable system.
A
B
C D
E
F
G
H
3
3
3
3
3
3
3
3
Figure 6. Generically unsolvable system that has a solvable constraint graph.
able (note that the position of the not changes the meaning, the latter being stronger
than the former). Consider for example Figure 6. The constraint system is shown on the
left, it consists of eight points in 3D and 18 distances between them. The corresponding
constraint graph is shown on the right, the weight of all the edges is 1, the weight of the
vertices is 3, the geometry-dependent constant D = 6. Note that this system is gener-
ically unsolvable since rigid bodies ABCDE and AFGHE can rotate about the axis
passing through AE. This example can also be transformed into a constraint problem in
2D, where the objects are circles and the constraints are angles of intersections (Saliola
and Whiteley, 1999).
Another example is the graph K
7,6
that in four dimensions represents distances be-
tween pairs of points. The constraint graph is minimal dense but it does not represent a
generically solvable system.
It should be noted that in two dimensions, according to Lamans (1970) theorem,
if all geometric objects are points and all constraints are distance constraints between
these points, then any minimal dense subgraph represents a generically solvable system.
In addition, there exists a purely combinatorial characterization of solvable systems in
one dimension based on connectivity of the constraint graphs. However, as examples
above indicate, the generalization of Lamans theorem fails in higher dimensions even for
the case of points and distances.
There is a matroid-based approach to verifying whether solvability of the constraint
graph implies solvability of the constraint system. It begins by checking whether a sub-
384 C. M. Homan et al.
modular function f(E) = 2|V |3, dened on sets of edges of the constraint graph, creates
a matroid (Whiteley, 1992, 1997), i.e. checking whether the created function is positive
on single edges. Thereafter this approach checks whether the corresponding geometric
structure is generically rigid. Some matroid-based approaches to determining generic solv-
ability are fast in practice, for example those to be discussed later in Section 4.2 under
Generalized Maximum Matching. However, so far no match has been established (for
all cases) between the created matroid and generic rigidity of the particular geometric
problem.
There are several attempts at characterization of generic solvability in dimension three
and higher for the case of points and distances (Graver et al., 1993; Tay, 1999). One such
characterization, the so-called Henneberg construction, checks whether a given constraint
graph can be constructed from the initial basic graph by applying a sequence of standard
replacements. A characterization due to Dress checks whether the constraint graph sat-
ises a certain inclusionexclusion type rule. A characterization due to Crapo examines
whether the constraint graph can be decomposed into a union of certain edge-disjoint
trees. All of these characterizations though interesting and useful, are so far unproven
conjectures.
Note. Due to the above discussion, we restrict ourselves to the class of constraint sys-
tems where solvability of the constraint graph in fact implies the generic solvability of
the constraint system. (As pointed out earlier, the converse is always true, with no as-
sumptions on the constraint system.)
As was indicated above, this class is far from empty, it contains all constraint problems
involving points and distances in 2D, problems resulting from Cauchy triangulations of
the polyhedra in 3D as well as body-and-hinge structures in 3D. Moreover, it should
be emphasized that while existing applications stop at nding subgraphs representing
solvable constraint systems, we are interested in the entire problem of decomposition
and recombination, optimizing the size of the largest subsystem to be solved, i.e. we are
interested in nding an optimal DR-plan. In addition, note that for the class of generi-
cally solvable constraint systems and corresponding graphs, the DR-planning problem is
already, in general, dicult (see a later note about NP-hardness, after the denition of
the optimal DR-plan).
3.2. formal definition of DR-plans using constraint graphs
Informally, stated in terms of constraint graphs, the DR-planning problem involves
nding a sequence of graphs G
i
a DR-plansuch that the original constraint graph
G = G
1
and every G
i
contains a minimal solvable subgraph S
i
, which is simplied or
abstracted into a simpler subgraph T
i
(S
i
) and substituted into G
i
to give an overall
simpler graph G
i+1
= T
i
(G
i
). (While T
i
(S
i
) should be simpler than S
i
, it should also be
somehow equivalent to S
i
, for example by having the same density value.) If the original
graph G
1
is well-constrained, then the process terminates when G
m
= S
m
. (If not, the
process terminates with the decomposition of G
m
into a maximal set of minimal solvable
subgraphs.)
Consider for example Figure 7 which shows one simplication step. On the left is the
constraint graph G
1
, the weight of all vertices is 2, the weight of all edges is 1, the
geometry-dependent constant D = 3. Then S
1
= {A, B, C} is a solvable subgraph. On
Decomposition of Geometric Constraints I 385
S
1
A
B
C
G
H
I
D E
F
V
I
H
E D
F
G
Figure 7. Original geometric constraint graph G
1
and simplied graph G
2
.
the right is the simplied graph T
1
(G
1
) = G
2
, after subgraph S
1
is replaced by a vertex
{V } = T
1
(S
1
). Since the density of the subgraph S
1
is 3, the weight of the vertex {V }
could be set to 3.
A sequence of simplication steps is shown in Figure 8. The top part depicts a ge-
ometric constraint graph G, where the weight of each edge is 1, the weight of each
vertex is 2, and the dimension-dependent constant D is equal to 3 (this corresponds
to points and distances in 2D). One of the plans for decomposing and recombining
G (and the geometric constraint system that G represents) into small solvable sub-
graphs (representing solvable subsystems), is to decompose G into dense subgraphs
S
1
= {A, B, C}, S
2
= {D, E, F}, S
3
= {G, H, I}, S
5
= {J, K, L}, S
6
= {M, N, O}; repre-
sent their solutions appropriately in a simplied graph so that they can be recombined,
one possibility is to represent them as vertices P, Q, R, S, T of weight 3 each; recursively
decompose the simplied graph into S
4
= {P, Q, R}, S
7
= {S, T}; represent and recom-
bine their solutions as vertices U, W of weight 3; and so on until the entire graph is
represented and recombined as a single vertex.
A corresponding DR-plan is shown at the bottom part of Figure 8. Note that there
could be more than one DR-plan for a given constraint graph. For example, another
possible DR-plan for a constraint graph described above is shown in Figure 9.
An optimal DR-plan will minimize the size of the largest solvable subgraph S
i
found
during the process. That is, it will minimize the maximum fan-in of the vertices in the
DR-tree shown in Figures 8 and 9, where by fan-in of a vertex we mean the number
of immediate descendants of the vertex. With this description, it should be clear that
DR-plans obtained using the weighted, constraint graph model are generically robust
with respect to the changes made to the geometric constraints; as long as the number of
degrees of freedom attached to the objects and destroyed by the constraints remains the
same, the same DR-plan will work for the changed constraint system as well. Thus, such
DR-plans satisfy the initial robustness requirements of characteristic (viii) of the set C
described in Section 1.3.
Next we formally dene a DR-plan for constraint graphs, and the various performance
measures that capture desirable properties of DR-plans and DR-planners. The develop-
ment is parallel to that in Section 2, where a DR-solution sequence and various perfor-
386 C. M. Homan et al.
S
4
S
2
S1
S
3 S5
S
6
S
S
S
S
S S
S
S
1
2
3
4
8
7
5
6
7
S
S
8
B
A C
D E
F
G
H
I
J
K
L M
N
O
A B C D E F G H I J K L M N O
Figure 8. Geometric constraint graph and a DR-plan.
mance measures for these sequences and for DR-solvers were motivated and dened in
terms of polynomial systems.
Note. Due to the strong analogy of DR-planners to DR-solvers dened in the previous
section, the discussion here is more terse. The following are useful generic correspondences
to keep in mind after reading the note at the end of Section 3.1, and the paragraphs
preceding it:
Solvable subgraphs solvable subsystems;
DR-plans DR-solution sequences;
DR-planner DR-solver;
subgraph simplier subsystem simplier.
Let G be a solvable constraint graph. A DR-plan Q for G is a sequence Q of graphs
G
1
, . . . , G
m
such that G
1
= G, G
m
is a minimal solvable graph, and every G
i
is solvable.
An algorithm is a general DR-planner if it outputs a DR-plan when given a solvable
constraint graph as input.
The union (resp. intersection) of two subgraphs A and B is the graph induced by the
Decomposition of Geometric Constraints I 387
S
2
S
1
S
3 S
5
S
4
S
S
S
S
S
S S S
8
6
7
5
4
1
2
3
S
6
S
7
S
8
A B C D E F G H I J K L M N O
B
A C
D E
F
G
H
I
J
K
L M
N
O
Figure 9. Another possible DR-plan.
union (resp. intersection) of sets of vertices of A and B. All subgraphs are understood to
be vertex induced.
The mapping from the graph G
i
to G
i+1
is called a subgraph simplier and is denoted
by T
i
. This mapping should have the following properties.
(1) If A is a subgraph of B, then T
i
(A) is a subgraph of T
i
(B).
(2) T
i
(A) T
i
(B) is the same as the graph T
i
(A B).
(3) T
i
(A) T
i
(B) is the same as T
i
(A B).
As in the case of DR-solution sequences, assume that every constraint graph G
i
in the
DR-plan can be written as S
i
R
i
U
i
, where S
i
is minimal solvable, R
i
is a maximal
subgraph such that S
i
and R
i
do not have common vertices, S
i
, R
i
, U
i
do not have
common edges and all vertices of U
i
are either vertices of S
i
or vertices of R
i
. Analogous
to properties (4) and (5) of the subsystem simplier in the previous section, we would
like the subgraph simplier to have the following additional properties.
(4) For every A R
i
, T
i
(A) = A.
(5) All the pre-images of S
i
, i.e. T
1
j
T
1
j+1
. . . T
1
i1
(S
i
) for all 1 j < i 1, are solvable.
388 C. M. Homan et al.
A DR-plan that satises the above rules is called valid. The size of the DR-plan Q of G
is the maximum of the sizes of S
i
. The size of an arbitrary subgraph A G
i
is computed
as follows.
Size(A) = 0
For 1 j i 1
B = T
i1
T
i2
. . . T
j
(S
j
)
If A B = 0
then Size(A) = Size(A) +D, A = A\ B
end if
end for
Size(A) = Size(A) +

vA
w(v)
In other words, the image of any of the S
j
contributes D to the size of A, where D
is a geometry-dependent constant, and the vertices of A that are not in any such im-
age contribute their original weight. The optimal size of the constraint graph G is the
minimum size of Q, where the minimum is taken over all possible DR-plans of G. An
optimal DR-plan of G is the DR-plan that has size equal to the optimal size of G. The
approximation factor of DR-plan Q of the graph G is dened as the ratio of the optimal
size of G to the size of Q.
Note. The problem of nding the optimal DR-plan for a constraint graph with un-
bounded vertex weights is NP-hard. This follows from a result in the authors paper
(Homann et al., 1997) showing that the problem of nding a minimum dense subgraph
is NP-hard, by reducing this problem to the CLIQUE. The CLIQUE problem is ex-
tremely hard to approximate (Hastad, 1996), i.e. nding a clique of size within a n
1
factor of the size of the maximum clique cannot be achieved in time polynomial in n,
for any constant (unless P = NP). However our reduction of CLIQUE to the optimal
DR-planning problem is not a so-called gap-preserving reduction (or L-reduction); thus
how well this problem could be approximated is still an open question.
The denition of solvability preservation is largely analogous to the case of DR-solvers,
but is additionally motivated by the following. In the case of DR-solvers, one condition
on a solvability preserving simplier is that it preserves the existence of real solutions for
certain subsystems. Here, in the case of DR-plans, a natural choice is to correspondingly
require that the simplier does not map well-constrained subgraphs or overconstrained
subgraphs to underconstrained and vice versa. The DR-plan Q of G is solvability pre-
serving if and only if for all A G
i
, A is solvable and (A S
i
= or A S
i
)
T
i
(A) is solvable. The DR-plan Q of G is strictly solvability preserving if and only if for
all A G
i
, A is solvable T
i
(A) is solvable. The DR-plan Q of G is complete if and
only if for all non-trivial solvable B S
i
, B = T
i1
T
i2
. . . T
j
(S
j
) for some j i 1.
Next, we formally dene DR-planners and their performance measures. An algorithm
is said to be a DR-planner if it outputs a DR-plan when given a solvable constraint
graph as input. As before, we consider DR-planners to be randomized algorithms. A
randomized DR-planner A is said to be valid, solvability preserving, strictly solvability
preserving, complete if and only if for every G every DR-plan produced by A is valid,
solvability preserving, strictly solvability preserving or complete, accordingly. The worst-
choice approximation factor of a DR-planner A on input graph G is the minimum of
Decomposition of Geometric Constraints I 389
the approximation factors of all DR-plans Q of G obtained by the DR-planner A over
all possible random choices. The best-choice approximation factor of the algorithm A on
input graph G is the maximum of the approximation factors of all the DR-plans Q of G
obtained by the DR-planner A over all possible random choices.
In addition to the above performance measures, we dene three others that reect the
ChurchRosser property, the ability to deal with underconstrained subsystems, as well
as the ability to incorporate an input, design decomposition provided by the designer.
A DR-planner is said to have the ChurchRosser property, if the DR-planner terminates
with a DR-plan irrespective of the order in which the dense subgraphs S
i
are chosen.
A DR-planner A adapts to underconstrained constraint graphs G if every (partial)
DR-plan produced by A terminates with a set of solvable subgraphs Q
i
such that each
solvable subgraph Q
i
has no supergraph that is solvable, and moreover, no subgraph of
G that is disjoint from all of the Q
i
s is solvable.
A conceptual design decomposition P is a set of solvable subgraphs P
i
, which are par-
tially ordered with respect to the subgraph relation. A DR-planner A is said to incorpo-
rate a design decomposition P, if for every DR-plan Q produced by A, the corresponding
sequence of solvable subgraphs S
i
embeds a topological ordering of P as a subsequence
recall that a topological ordering is one that is consistent with the natural partial order
given by the subgraph relation on P.
When a DR-plan incorporates a design decomposition P, the level of a cluster P
i
in
the partial ordering of P can now be viewed as a priority rating which species which
component of the design decomposition has most inuence over a given geometric object.
In other words, a given geometric object K is rst xed/manipulated with respect to the
local coordinate system of the lowest level cluster P
i
P containing K. Thereafter, the
entire cluster P
i
can be treated as a unit and can be independently xed/manipulated
in the local coordinate system of the next level cluster P
j
containing (the simplication
of) P
i
, etc.
Finally, we summarize how the above formal performance measures capture the infor-
mally described characteristics in the set C given in Section 1.3. The property of being a
general DR-planner refers to whether the method successfully terminates with a DR-plan
in the general case, and reects characteristic (i); since we use constraint graphs which
yield robust DR-plans that can be obtained eciently, the property of being a general
DR-planner also reects (iii) and (viii); dealing with underconstrained graphs also re-
ects (i); incorporating input, design decompositions reects (v); validity inuences the
ChurchRosser property and reects (i) as well as (iv), (v), (vi); solvability preservation,
strict solvability preservation inuences the ChurchRosser property and reects (i), (ii),
(iv), (v); completeness is based on criteria (ii) and (ix); worst- and best-choice approxi-
mation factors are based on (ii) and complexity directly reects (iii).
4. Performance Analysis of Prior DR-planners
We concentrate on two primary types of prior algorithms for constructing DR-plans
using constraint graphs and geometric degrees of freedom.
Note. Due to reasons discussed in Section 3, we leave out those graph rigidity-based
methods for distance constraints in dimensions three or greater such as Tay and Whiteley
(1985) and Hendrickson (1992) as well as methods such as Crippen and Havel (1988),
Havel (1991) and Hsu (1996) since they are non-deterministic or rely on symbolic compu-
390 C. M. Homan et al.
tation, or they are randomized or generally exponential and based on exhaustive search.
Graph rigidity- and matroid-based methods for more specic constraint graphs are dis-
cussed in Section 4.2 under the so-called maximum matching-based algorithms. Also,
for reasons discussed in the Introduction, we leave out methods based on decomposing
sparse polynomial systems.
The rst type of algorithm, which we call SR for constraint shape recognition (e.g.
Owen, 1991, 1996; Homann and Vermeer, 1994, 1995; Bouma et al., 1995; Fudos and
Homann, 1996b, 1997), concentrates on recognizing specic solvable subgraphs of known
shape, most commonly, patterns such as triangles. The second type, which we call MM for
generalized maximum matching (e.g. Kramer, 1992; Ait-Aoudia et al., 1993; Pabon, 1993;
Latham and Middleditch, 1996), is based on rst isolating certain solvable subgraphs by
transforming the constraint graph into a bipartite graph and nding a maximum general-
ized matching, followed by a connectivity analysis to obtain the DR-plan. In this section,
we give a formal performance analysis of SR- and MM-based algorithmschoosing a rep-
resentative algorithm (typically the best performer) in each classusing the performance
measures dened in the previous section.
Informally, one major drawback of the SR and MM algorithms is their inability to
perform a generalized degree of freedom analysis. For example, SR would require an
innite repertoire of patterns. In the case of spatial constraints, some elementary patterns
have been identied (Homann and Vermeer, 1994, 1995). In the case of extending the
scope of planar constraint solvers, adding free-form curves or conic sections requires
additional patterns as well (Homann and Peters, 1995; Fudos and Homann, 1996a).
Similarly, a decomposition of underconstrained constraint graphs into well-constrained
components is possible for SR algorithms but only subject to the pattern limitations. In
many cases, MM algorithms will output DR-plans with larger, non-minimal subgraphs
S
i
that may contain smaller solvable subgraphs. This aects the approximation factors
adversely.
This inability to nd general minimal dense subgraphs also aects their ability to
deal with overconstrained subgraphs that arise in assemblies, which is in turn needed to
perform constraint reconciliation (Homann and Joan-Arinyo, 1998).
4.1. constraint shape recognition (SR)
Consider the algorithm of Fudos and Homann (1996b, 1997) which relies on the
following strong assumption (SR1): all geometric objects in two dimensions have 2 degrees
of freedom and all constraints between them are binary and destroy exactly 1 degree of
freedom. Thus, in the corresponding constraint graph, the weight of all the edges is 1 and
of all the vertices is 2. Because of this assumption, the SR algorithm ignores the degrees
of freedom and relies only on the topology of the constraint graph.
description of the algorithm
We give a terse description of the algorithmthe reader is referred to Fudos and
Homann (1996b, 1997) for details. Our description is meant only to put the algorithm
into an appropriate DR-planner framework that is suited for the performance analysis.
The algorithm consists of two phases. During Phase One, SR uses the bottom-up
iterative technique of Itai and Rodeh (1978): in the current graph G
i
(where G
1
= G),
Decomposition of Geometric Constraints I 391
specic solvable graphs (clusters) are found that can be represented as a union of three
previously found clusters that pairwise share a common vertex. Such congurations of
three clusters are called triangles. The vertices of G
i
represent clusters and edges of G
i
represent constraints between clusters (initially due to SR1, every vertex and every edge
of G is a cluster). More specically, once a triangle formed by three clusters has been
found, the three vertices in G
i
corresponding to these three clusters are simplied into
one new vertex in the simplied graph G
i+1
. The edges of G
i+1
are induced by the edges
of the three old vertices. This is repeated for k steps until there are no more triangles
left. If there is only one cluster left in G
k
, then the algorithm terminates. Otherwise G
k
serves as an input to Phase Two of the algorithm.
Before we describe Phase Two, we note that a so-called cluster graph C
i
corresponding
to G
i
is used as an auxiliary structure for the purpose of nding triangles. Hence the
partial DR-plan produced during Phase One of the algorithm is of the form: (G
1
, C
1
), . . . ,
(G
k
, C
k
). The vertices of the cluster graph C
i
correspond to
vertices of the original graph G
1
;
cluster vertices in the graph G
i
; and
edges in G
1
which have not been included into one of the cluster vertices in the
graphs G
i1
, . . . , G
1
.
In particular, the rst cluster graph C
1
contains one vertex for every vertex and every
edge of G
1
. The edges of the cluster graph C
i
connect the vertices of G
i
that represent
clusters to the original vertices from G
1
that are contained in these clusters. Due to the
structure of cluster graphs, triangles in G
i
are found by looking for specic 6-cycles in
the corresponding cluster graph C
i
. These 6-cycles consist of three cluster vertices and
three original vertices. The new cluster graph C
i+1
is constructed from C
i
by adding a
new vertex c
i
representing the newly found cluster S
i
, and connecting it by edges to the
original vertices from G
1
that are in the cluster S
i
. We also remove the three old cluster
vertices in C
i
which together formed the new cluster S
i
. We note that this is the only way
in which cluster vertices are removed from cluster graphs. In particular, situations may
arise where two clusters (that share a single vertex) are represented by the same vertex
in the graph G
i
, but they are represented by distinct vertices in the cluster graph C
i
.
During Phase Two, SR constructs the remainder of the DR-plan (G
k
, C
k
), . . . ,
(G
m
, C
m
). First SR uses a global topdown technique of Hopcroft and Tarjan (1973)
to divide G
k
into a collection of triconnected subgraphs. These subgraphs are found
by recursively splitting the original graph using separators of size at most 2. Thus the
triconnected components can be viewed as the leaves of a binary tree T each of whose
internal nodes corresponds to a vertex separator of size at most 2.
For every triconnected subgraph S in G
k
, a new cluster vertex is created in the cluster
graph C
k
: similar to Phase One, this vertex replaces all vertices that represent existing
clusters contained in S.
The next pair (G
k+1
, C
k+1
) in the DR-plan is found as in Phase One by nding a tri-
angle in G
k
, or eectively locating a 6-cycle in C
k
. However, triangles in G
k
have already
been located in the course of constructing the tree Tthese triangles were originally split
by the vertex separators at the internal vertices. Thus, if the original constraint graph
G is solvable, the remainder of the DR-plan (G
k+1
, C
k+1
), . . . , (G
m
, C
m
) is formed by
repeated simplications along a bottom up traversal of the tree T.
If the original graph was underconstrained, then it is still possible to construct a
392 C. M. Homan et al.
5
1
5
1
G
G
C
C
M
O
F
G
D
LN
OM
FO
DG
P
AD
A
D
DG
AE
FO
OM
G
F
O
A B C
D E F
G H
I J
K
L M
N O O N
M L
K
J I
H G
P
Figure 10. The original graph G = G
1
, the cluster graph C
1
and the simplied graphs G
5
, C
5
.
DR-plan of its maximal well-constrained subgraphs, by introducing additional constraints
and making the original graph solvable. That is, in order to complete the bottom up
traversal of the tree T in Phase Two, additional constraints need to be introduced, to
make the whole graph solvable. For details, see Fudos and Homann (1996b, 1997).
example
Consider Figure 10. During Phase One, the triangles ADE, ABE, BCE and CEF will
be discovered and simplied as P. During Phase Two, the remainder of the graph will
be divided into triconnected subgraphs P, PGKILN and OKHMJP, then PGKILN
and OKHMJP are simplied and nally the union of P, PGKILN and OKHMJP is
simplied.
defining the simplifier map
The same simplier is used throughout Phase One and Two: replace a subgraph S
i
consisting of a triangle of clusters in G
i
(or triconnected subgraphs during Phase Two)
by one vertex in G
i+1
representing this triangle. The subgraph S
i
is found as a 6-cycle in
the cluster graph C
i
. The cluster graph C
i+1
is constructed as described in Phase One.
More formally, recalling the denitions in Section 3: let G be a constraint graph; the
rst graph G
1
in the DR-plan is the original graph G. Let G
i
= (V, E) be the current
graph and let S
i
be a cluster found. Let A G
i
, A = (V
A
, E
A
). Then the image of A
under the subgraph simplier T
i
is T
i
(A) = A, if the intersection of A and S
i
is empty;
otherwise T
i
(A) = (V
T
i
(A)
, E
T
i
(A)
) where V
T
i
(A)
is the set of all vertices of A that are
Decomposition of Geometric Constraints I 393
T
1
T
3
C
C
2
FE
FC
CE
CB
2
C
C
B
E
D
A
D
A
AB
EB
DE
AD
AE C
1
A
D
E
B
C
EB
AB
CB
CE
G
2
G
1
G
3
C
1
C
2
E
B
J
O N
K
M L
I
H G
F E D
C B A B C
F
G
I J
K
L
O
C
F
G H
I J
K
L
O
1
C
M
N
H
2
C
M
N
Figure 11. Action of the simplier on G
i
and C
i
during Phase One.
not vertices of S
i
plus a vertex c
i
that replaces the cluster S
i
. The set of edges E
T
i
(A)
is
a set of all edges of A that are not edges of S
i
; the edges of E
A
that have exactly one
endpoint in S
i
will have this endpoint replaced by the vertex c
i
; and the edges of A that
have both endpoints in S
i
are removed.
Since the cluster S
i
in G
i
is located by nding a 6-cycle in C
i
, we need to describe how
C
i+1
is constructed from C
i
, i.e. the eect that T
i
has on C
i
(formally, T
i
is a map from
(G
i
, C
i
) to (G
i+1
, C
i+1
)). To obtain C
i+1
, we start with C
i
and rst add a new vertex c
i
representing the cluster S
i
, which is connected by edges to all the original vertices that
it contains. Finally, vertices in C
i
that represent clusters entirely contained in S
i
are
removed, and edges adjacent to these vertices are also removed.
Figures 11 and 12 illustrate the action of the subgraph simplier T
i
on both G
i
and C
i
.
a further concession for SR
Observe that the SR algorithm is not a general DR-planner when input geometric
constraint graphs do not comply with assumption SR1. For example, for the graph shown
in Figure 13, during Phase One, SR would not nd any triangles and during Phase Two,
it would conclude that the graph is underconstrained, even though the graph is well-
constrained.
SR implementations may remedy some cases similar to this one. For instance, weight
2 edges to weight 2 vertices often arise from incidence constraints between geometric
elements of the same type (two coincident points or two coincident lines), and such cases
are easily accounted for by working with equivalence classes of such vertices. Moreover,
most planners will evaluate the density of a graph before announcing under- or over-
constrained situations. Thus, an implementation of the SR algorithms of Fudos and
Homann (1996b, 1997) may or may not construct a DR-plan for the graph of Figure 13
394 C. M. Homan et al.
C
8
G
G
5
T
6
T
7
T
8
T
5
7
C
6
C
9
F
K
D
R
Q
P
P
IL
LN
G
K
L
N
A
J
O
F
K
L
D
A
DG
DI
DN
IK
OF
I
D
P
Q
KH
KO
KJ
JF
G
I
P
O N
K
J
H
M L
Figure 12. Action of the simplier during Phase Two.
3 2
2
2
2
2
2
1
1
2
2
2
2
1
1
Figure 13. This solvable graph would not be recognized as solvable by SR.
depending on the original problem statement. It is clear that such heuristics enlarge the
class of solvable graphs, but fall short of decomposing all solvable constraint graphs.
However, even when input graphs do comply with SR1, and SR checks the overall
density of a graph, SR could still mistake a graph that is not solvable for solvable.
Figure 14 shows an example which the SR algorithm may process incorrectly: since the
graph contains no solvable triangles, SR proceeds immediately to Phase Two. During
Phase Two, the triconnectivity of the entire graph could be erroneously construed to mean
that the graph is solvable. In fact, the graph does have density 3, which supercially
seems to support such a conclusion. However, the graph is certainly not well-constrained:
the eight vertices on the right form a subgraph of density 2; that is, an overconstrained
subgraph. Moreover, if this overconstrained subgraph is replaced a subgraph of density
3, the resulting graph has density 4 revealing that it is not well-overconstrained either,
and therefore not solvable.
Figure 14 demonstrates that the density calculation heuristics is insucient to deter-
mine the existence of minimal dense subgraphs of triconnected constraint graphs. What
is needed is a general algorithm for nding minimal dense subgraphs.
In order to give a performance analysis of the SR algorithm for those classes of constraint
graphs where it does produce a satisfactory DR-plan, we make a strong concession (SR2)
Decomposition of Geometric Constraints I 395
Figure 14. Weight of all vertices is 2, weight of all edges is 1.
that: only triangular structures are acceptable for the remainder of this section.
That is, we modify the denitions of validity, solvability preserving, strictly solvability
preserving and completeness by replacing the words solvable subgraph by the following
recursive denition: either a subgraph that can be simplied into a single vertex by
a sequence of consecutive simplications of triangles of solvable subgraphs (using the
simplier described in Section 4.1), or a vertex, edge or triconnected subgraph.
performance analysis
In this section, we analyze the SR algorithm with respect to the various performance
measures dened in Section 3.
Claim 4.1. Under the concession SR2, the SR algorithm is valid.
Proof. We will show that every pre-imageof the solvable cluster S
i
found at the
ith iteration of the SR algorithmis also solvable, using the stricter SR2 denition
of solvable. It can be easily checked that the other requirements necessary for validity
from Section 3 also clearly hold.
Let S
i
be a cluster that has been found in G
i
by the SR algorithm (by locating a
6-cycle in C
i
). Then from the description of the algorithm, and by assertion SR2, there
is a sequence of simplications, say T
i+1
, . . . , T
m
whose composition maps S
i
to a single
vertex. For every pre-image A G
j
A = T
1
j
. . . T
1
i1
(S
i
), for j i 1, the composition
T
m
T
i
T
i1
T
j
will map A to a single vertex, hence A is solvable. 2
Claim 4.2. The SR algorithm is strictly solvability preserving under the concession SR2,
and therefore is solvability preserving as well.
Proof. Let A be a solvable subgraph, i.e. there is a sequence of simplications
A
1
, . . . , A
m
such that A = A
1
and A
m
consists of only one vertex. If the intersection
of A and the currently found cluster S
i
is empty, then T
i
(A) = A and it remains solv-
able. If this intersection is not empty, then a subgraph B = A S
i
is simplied into a
new cluster vertex c
i
= T
i
(S
i
). We need to show that in this case T
i
(A) remains solv-
able as well. Suppose that some subgraph of B is a vertex of a 6-cycle in one of the
cluster graphs formed during the simplication A
1
, . . . , A
m
. Then clearly the new cluster
vertex c
i
could perform the same function: i.e. it could also be a vertex in that 6-cycle
of one of the cluster graphs formed during the simplication A
1
, . . . , A
m
. In addition, if
A was triconnected, then so is T
i
(A). This proves that there is a sequence (essentially a
396 C. M. Homan et al.
A B
Figure 15. Solvable graph containing two triconnected subgraphs not composed of triangles.
Figure 16. Solvable graph consisting of n/3 solvable triangles.
sequence A
1
, . . . , A
m
modied by replacing B by c
i
) which terminates in a single vertex,
thus demonstrating the density of T
i
(A). 2
Claim 4.3. Even under concession SR2, SR is not complete.
Proof. We will show that there are cases when the SR picks large, non-minimal solv-
able subgraphs S
i
to simplify, ignoring smaller solvable subgraphs of S
i
.
Consider Figure 15. If subgraphs A and B are triconnected but not composed of
triangles, then they are solvable, but since the entire graph is triconnected neither A
nor B will be simplied by SR. However, since the whole graph AB is triconnected, it
will chosen by SR as S
1
during Phase Two. 2
Claim 4.4. The (worst- and) best-choice approximation factor of SR under concession
SR2 is at most O(
1
n
).
Proof. Consider Figure 16. The entire graph consists of n/3 triangles. During Phase
One, SR will successfully locate and simplify each one of them. However, during Phase
Two, SR will not be able to decompose the entire solvable graph into smaller solvable
ones (since the entire graph is triconnected) so the size of the corresponding DR-plan
is O(n). On the other hand, the optimal DR-plan would simplify neighboring pairs of
triangles, one pair at a time, thus the optimal size is a small constant. 2
Next, we consider three other performance measures discussed in Section 3.
Claim 4.5. Under the concession SR2, the algorithm SR adapts to underconstrained
graphs.
Proof. Suppose that the graph G is underconstrained. Let A be a solvable subgraph
that is not contained in any other solvable graph. Since Phase Two of the SR algorithm
Decomposition of Geometric Constraints I 397
is topdown, either SR will nd A and simplify it as one of the S
i
s, or A S
i
= for
one of the S
i
. In either case, SR adapts to G. 2
Observation 4.6. Under the concession SR2, the SR algorithm has the ChurchRosser
property, since the new graph G
i+1
remains solvable if G
i
is solvable, regardless of
the choice of the solvable S
i
that is simplied at the ith stage.
Claim 4.7. Under the concession SR2, the SR algorithm can incorporate design decom-
positions P if and only if P fullls the following requirement: any pair of solvable
subgraphs P
k
and P
t
in P satisfy P
k
P
t
or P
t
P
k
or P
k
P
t
contains no edges.
Proof. For the if part we consider the most natural modication of the original SR
algorithm, and nd a topological ordering O of the given design decomposition P
which is a set of solvable subgraphs of the input graph G, partially ordered under the
subgraph relationsuch that O is embedded as a subplan of the nal DR-plan generated
by this modied SR algorithm; i.e. O forms a subsequence of the sequence of solvable
subgraphs S
i
, whose (sequential) simplication gives the DR-plan.
We take any topological ordering O of the given design decomposition P and create
a DR-plan for the rst solvable subgraph P
1
in P. That is, while constructing the
individual DR-plan for P
1
, we ignore the rest of the graph. This individual DR-plan
induces the rst part of the DR-plan for the whole graph G. In particular, the last graph
in this partial DR-plan is obtained by simplifying P
1
using the simplier described in the
section Dening the Simplier Map (and treating P
1
exactly as SR would treat a cluster
S
j
found at some stage j). Let G
i
be the last graph in the DR-plan for G created thus
far. Now, we consider the next subgraph P
2
in the ordering O, and nd an individual
DR-plan for it, treating it not as a subgraph of the original graph G, but as subgraphs
of the simplied graph G
i
. This individual DR-plan is added on as the next part of the
DR-plan of the whole graph G.
The crucial point is that the simplication of any subgraph, say P
k
, will not aect any
of the unrelated subgraphs P
t
, t k, unless P
k
P
t
. This is because, by the requirement
on P, P
k
and P
t
share no edges. Therefore, when the cluster vertex for P
k
is created,
none of the clusters inside P
t
are removed.
The processof constructing individual DR-plans for subgraphs in the decomposition
P and concatenating them to the current partial DR-planis continued until a partial
DR-plan for the input graph G has been produced, which completely includes topological
ordering O of the decomposition P as a subplan. Let G
p
be the last graph in this partial
DR-plan. The rest of the DR-plan of G is found by running the original SR algorithm
on G
p
and the corresponding cluster graph C
p
.
For the only if part, consider Figure 17. Let P={P
0
, P
1
, P
2
}, where P
0
= ABD, P
1
=
BCD, P
2
= ABCD. Then SR cannot produce any DR-plan that would incorporate P as
subplan. 2
4.2. generalized maximum matching (MM)
Consider the algorithms of Ait-Aoudia et al. (1993), Pabon (1993), Kramer (1992), Ser-
rano and Gossard (1986) and Serrano (1990) as well as graph rigidity- and matroid-based
methods for distance constraints in two dimensions (Imai, 1985; Gabow and Westermann,
398 C. M. Homan et al.
A
C
D
B
P
0
P
2
P
1
A B D B C D
P
2
P
0
A B D C
Figure 17. Constraint graph, intended and actual decompositions.
1988; Hendrickson, 1992) as well as more general constraints (Sugihara, 1985) all of which
more or less use (generalized) maximum matching (or equivalent maximum network ow)
for nding solvable subgraphs in specialized geometric constraint graphs. These methods
either assume that the constraint graph has zero density or they reduce the weight of an
arbitrarily selected set of vertices in order to turn the graph into one of zero density.
In this paper we will analyze what we consider to be the most general algorithm of
this kind (Pabon, 1993) (it generalizes the algorithm of Ait-Aoudia et al., 1993, although
the latter provides a more complete analysis), supplemented (by us, as suggested by a
reviewer) with a method from Hendrickson (1992).
Note that while the algorithm of Pabon (1993) both locates solvable subgraphs and de-
scribes how to construct a corresponding DR-plan, Sugihara (1985), Hendrickson (1992)
and Gabow and Westermann (1988) only describe algorithms that allow one to verify
whether a given graph is solvable, but do not explicitly state how to use these algo-
rithms for successively decomposing into small solvable subgraphs, i.e. for constructing
DR-plans.
While neither of the algorithms analyzed in this paper perform well according to our
previously dened set of criteria, later, in Part II of this paper, we will describe our
network ow-based algorithm MFA that improves the performance in several key areas.
description of the algorithm
As in the case of SR, we give a terse description of the MM algorithmthe reader
is referred to Ait-Aoudia et al. (1993) and Pabon (1993) for details. Our description
is meant only to put the algorithm into an appropriate DR-planner framework that is
suited for the performance analysis.
By using maximum ow, the input constraint graph is decomposed into a collection of
subgraphs that are strongly connected. The ow information also provides a partial order-
ing of the strongly connected subgraphs representing the order in which these subgraphs
should be simplied. This ordering in turn species the DR-plan. It is important to note
that these strongly connected components represent a sequence of solvable subgraphs of
the original constraint graph, only if the input constraint graph is well-constrained.
Consider Figure 18. All the vertices have weight 2, all the edges have weight 1, and
the geometry is assumed to be in two dimensions (i.e. the geometry-dependent constant
D = 3). The output of the MM algorithm is:
a set of vertices whose total weight is reduced by 3 units, say weight of vertices
Decomposition of Geometric Constraints I 399
A
B
C
D A
B
C
D
Figure 18. Original constraint graph and its decomposition into strongly connected components.
AC
AB
BC
CD
BD
A
B
C
D
Sink
1
1
1
1
1
2
1
1
1
AC
AB
BC
CD
BD
A
B
C
D
Sink
1
1
1
1
1
1
1
1
2
Figure 19. Modied bipartite graph and maximum ow in this graph.
A, B and C to be reduced by 1 unit each, (this corresponds to xing 3 degrees of
freedomthe number of degrees of freedom of a rigid body in two dimensions);
two strongly connected components ABC and D;
the DR-plan, i.e. the information that the subsystem represented by ABC should
be simplied/solved rst and then its union with D should be simplied/solved.
In order to produce such an output, the MM algorithm rst constructs the network
X = (V X, EX) corresponding to the original weighted constraint graph G = (V, E)
(after the weights of A, B, C were reduced by 1 unit each). The set of vertices V X is
the union of V X
1
and V X
2
where the vertices in V X
1
correspond to the vertices in
V , vertices in V X
2
correspond to the edges in E. An edge ex EX would be created
between vx
1
V X
1
and vx
2
V X
2
if the vertex in V corresponding to the vx
1
is
an endpoint of an edge in E corresponding to vx
2
. The edge ex has innite capacity.
All the vertices in V X
1
are connected to the special vertex called Sink. The capacity of
connecting edges is equal to the weight of the corresponding vertices in V . For example,
the left half of Figure 19 shows the bipartite graph corresponding to the constraint graph
of Figure 18. Next the maximum ow in the network X is found, with vertices in V X
2
being source vertices of capacity equal to the weights of the corresponding edges in E.
See the right half of Figure 19 (thick edges have non-zero ow).
The maximum ow found in X induces a partition of the original graph G into a
partially ordered set of strongly connected componentsgiving a sequence of solvable
subgraphs of G, provided G is well-constrainedas in Figure 18, according to the fol-
lowing rules: if the ow in X from the vertex z V X
2
that is connected to the vertices
x, y V X
1
is sent toward x, then in the graph G, the edge corresponding to z (between
vertices x and y) becomes an oriented edge directed from y to x. If the ow from z is
sent toward both x and y, then the edge is bidirected toward both x and y. (Recall that
400 C. M. Homan et al.
a strongly connected component S in this directed graph is a subgraph such that for
any two vertices a, b S, there is an oriented path from a to b.) The partial ordering of
components is induced as follows. Let K and L be two strongly connected components
in the oriented version of G. If all the edges between vertices of K and L are pointing
toward L, then L should be simplied after K.
defining the simplifier map
We capture the transformations performed by the MM DR-planner described above,
by describing its simplier maps (recall the denitions in Section 3).
Let G = (V, E) be the geometric constraint graph. Denote by G
1
= (V
1
, E
1
= E)
the directed graph after weights of some vertices have been reduced as described above
and a partial ordering of strongly connected components has been found. First S
1
G
1
is located such that S
1
is strongly connected. Then S
1
is simplied into the vertex v
1
of weight 0. The other vertices of G
1
remain unchanged. Edges of G
1
that had both
endpoints outside of S
1
are unchanged, edges that had both endpoints in S
1
are removed,
edges that had exactly one endpoint in S
1
have this endpoint replaced by v
1
. In the
next step, C
2
the next strongly connected component in a topological ordering of the
components in G
2
= T
1
(G
1
)is located. The subgraph S
2
is set to be {v
1
} C
2
. Then
S
2
is simplied into the vertex v
2
of weight 0. This process is continued until G
k
consists
of one vertex. The simplier maps are formally dened as follows.
T
i
(S
i
) = v
i
where v
i
is the vertex in G
i+1
of weight 0.
If B G
i
, B S
i
= , then T
i
(B) = B.
If B G
i
, B S
i
= , then the image of B, T
i
(B) = {v
i
} (B \ S
i
).
redefining solvability
Claim 4.8. The MM algorithm is a not a general DR-planner.
Proof. While the MM algorithm can correctly classify as solvable and decompose
well-constrained and well-overconstrained graphs and correctly classify underconstrained
graphs that have no overconstrained subgraphs as being unsolvable, it is unable to cor-
rectly classify an underconstrained graph that has an overconstrained subgraph. Con-
sider Figure 21, the weight of all the edges is 1, of the vertices as indicated. Graph
ABCDE has density 3, it contains overconstrained subgraphs AC, BC, CD, CE, and
removal of say edge AC will result in ABCDE becoming non-dense, hence ABCDE
is not well-overconstrained and not solvable. The MM algorithm may or may not de-
tect this, depending on the initial choice of vertices whose weights are to be reduced.
Suppose that the weight of the vertex E was reduced by 1 and the weight of the ver-
tex D by 2. Then the corresponding maximum possible ow f and the correspond-
ing decomposition into strongly connected components are shown in Figure 22. Note
that it is impossible to simplify the strongly connected component A, and thus there
can be no DR-plan for this initial choice of vertices for reducing weights. On the other
hand, if the weight of A had been reduced by 2 and weight of D by 1, then the max-
imum ow is larger than f (in fact, no choice of vertices for weight reduction can give
a larger ow) and it can be checked that MM would yield a DR-plan, see Figure 23
(S
1
= A, S
2
= AC, S
3
= ABC, S
4
= ABCD, S
5
= ABCDE). 2
Decomposition of Geometric Constraints I 401
A
C
D
B
1
*
Sink
1
1
1
1
2
2
A
D
B
C
2
2
4
AB
AD
BD
BC
DC
*
*
2
*
Figure 20. Constraint graph and network ow with three extra ow units at AB.
2
2
1
2
A
B
C
E
D
2
Figure 21. The existence of a DR-plan depends on the initial choice of vertices whose weights are
reduced.
Thus MM runs into problems in the presence of overconstrained subgraphs, unless
G happens to be well-overconstrained, which cannot be a priori checked without re-
lying on an algorithm for detecting overconstrained subgraphs and replacing them by
well-constrained ones. Therefore it is necessary and sucient to locate overconstrained
subgraphs and replace them by well-constrained ones, in order to guarantee that MM
will work.
The author of Pabon (1993) does not specify how to do this. The following modication
similar to that of Hendrickson (1992) could be used, as was suggested by a reviewer, and
completed here.
First a maximum ow in the unmodied network X (i.e. where vertex weights are not
reduced) is found. After that, for every source vertex v V X
2
, except one vertex v
m
,
the following two steps are repeated.
Three extra units of ow are sent from v, possibly rearranging existing ows in X.
These 3 units of ow are removed, without restoring original ows.
For vertex v
m
, 3 units of ow are sent but not removed.
For example, consider Figure 20. The constraint graph is shown on the left, the weights
of all vertices are 2, the weights of all edges are 1. The resulting network ow is shown
on the right, assuming that AB was the last vertex v
m
. The extra three units of ow and
their destination vertices A and B are marked by asterisks. This particular ow induces
a weight reduction of A by 1 unit and of B by 2 units.
This modication identies overconstrained subgraphs since it is impossible to send
all three extra units from at least one edge of an overconstrained graph. These overcon-
402 C. M. Homan et al.
AC
AB
BC
A
B
C
D
Sink
1
1
1
1
1
1
CD
CE
ED
E
1
2
2
1
0
1
0
2
1
A
B
C
D
E
2
Figure 22. Maximum ow and decomposition given the bad initial choice.
0 2
1
A
B
C
D
E
2 1
Figure 23. Decomposition given the good initial choice.
strained graphs have to be modied by the designer to become well-constrained, and the
ow algorithm is run again.
This is important because otherwise the DR-algorithm cannot proceed, since in un-
derconstrained graphs with overconstrained subgraphs, strongly connected components
do not necessarily correspond to a sequence of solvable subgraphs.
To reect the modication above, we make the following concession for MM, (MM1):
the input constraint graph G has no overconstrained subgraphs in G. A subgraph A G
is solvable if it has zero density, after the vertices for weight reduction by D are chosen
(these can be chosen arbitrarily, provided there are no overconstrained subgraphs).
performance analysis
In this section, we analyze the MM algorithm with respect to the various performance
measures dened in Section 3.
Note that despite the concession MM1, the MM algorithm has the following drawbacks:
the DR-plan is uniquely determined, once vertices whose weights are reduced are chosen.
For example, in Figure 20 after the weight of A is reduced by 1 and the weight of B
by 2 units, solvable subgraphs to be simplied are S
1
= {A, B}, S
2
= {T
1
(S
1
), D}, S
3
=
{T
2
(S
2
), C}. While the subgraph BCD is also solvable (if say initially weights of vertices
B and C were reduced instead of A and B), it cannot be chosen as one of the S
i
after
the weights of A and B are reduced.
This causes MM to have bad worst-choice and best-choice approximation factors and
to be unable to incorporate general designer decomposition.
The following is straightforward from the description of the simpliers.
Decomposition of Geometric Constraints I 403
L
n/2
L
n/2 1
L
3
3
3
3
3
2
3
3
3
3
2
L
2
L
1
R
n/2
R
n/2 1
R
3
R
2
R
1
2
C
Figure 24. Bad best-choice approximation.
Claim 4.9. Under the concession MM1, the MM algorithm is a valid DR-planner.
Claim 4.10. Under the concession MM1, the MM algorithm is strictly solvability pre-
serving (and therefore solvability preserving).
Proof. Suppose that a subgraph A of the input graph G is solvable, i.e. d(A) = 0. Let
S
i
be the solvable subgraph to be simplied at the current stage. Let B = AS
i
, C =
A\B. Since we assume that G does not contain any overconstrained subgraphs, d(B) 0
and d(AS
i
) 0 d(AS
i
) = d(A)+d(S
i
)d(B) 0 0+0d(B) 0 d(B) = 0.
Thus d(T
i
(A)) = d(T
i
(C)) +d(T
i
(B)) = d(C) + 0 = d(A) d(B) = 0, therefore T
i
(A) is
also solvable. 2
Claim 4.11. Under the concession MM1, the MM algorithm is complete.
Proof. Let A be a proper solvable subgraph of the S
i
, i 2. Since A is solvable,
there cannot be any edges outside of A pointing toward A (this is because there is no
room for ows of outside edges toward the vertices of A). Recall that S
i
is the union
of C
i
{v
i1
}, where C
i
is the rst strongly connected component in the topological
ordering of the components at the stage i, and v
i1
is the simplication of S
i1
. However,
unless A = v
i1
, the rst strongly connected component at this stage would have been
A\ {v
i1
}, not C
i
, which contradicts the choice of C
i
at stage i. 2
Claim 4.12. Under the concession MM1, the best-choice (and worst-choice) approxima-
tion factor of MM is at most O(
1
n
).
Proof. To prove the bound on the best-choice approximation factor consider Figure 24.
The left and right columns contain n/2 vertices each. The weights of all the vertical edges
are 2, the weights of all other edges are 1, the weights of the vertices are as indicated,
and the geometry-dependent constant D = 3.
Note that all solvable subgraphs in Figure 24 could be divided into three classes. The
404 C. M. Homan et al.
rst class consists of the subgraphs CL
1
L
2
; CL
1
L
2
L
3
; . . . ; CL
1
L
2
. . . L
n/21
L
n/2
. The
second class consists of the subgraphs CR
1
R
2
; CR
1
R
2
R
3
; . . . ; CR
1
R
2
. . . R
n/21
R
n/2
.
The third class contains the solvable subgraphs that contain both L and R vertices. There
is only one element in this classthe entire graph CL
1
L
2
. . . L
n/2
R
1
R
2
. . . R
n/2
. There is
an optimal DR-plan of constant size that takes S
1
= CL
1
L
2
, S
2
= S
1
L
3
, . . . , S
n/21
=
S
n/22
L
n/2
. After that it takes S
n/2
= CR
1
R
2
, S
n/2+1
= S
n/2
R
3
, . . . , S
n
=
S
n1
R
n/2
. Finally it takes S
n+1
= S
n/21
S
n
.
However all DR-plans found by MM will have size O(n). The reason for this is that MM
is unable to simplify solvable subgraphs on the left of the Figure 24 independently from
the solvable subgraphs on the right. More formally, let S
1
be the rst subgraph simplied
by MM under some DR-plan Q. If S
1
belongs to the third class of solvable subgraphs,
then the size of Q is O(n). Otherwise, without loss of generality we can assume that S
1
belongs to the rst class. According to the denition of MM, the simplication of S
1
is
a vertex v
1
of weight 0. After this simplication any strongly connected component that
contains some R
i
should also contain all of the R
1
. . . R
n/2
. Hence there is an S
i
in Q
such that R
1
R
2
. . . R
n/2
S
i
. Hence the size of Q is O(n). 2
Next, we consider three other performance measures discussed in Section 3.
Claim 4.13. Under the concession MM1, a simple modication of the MM algorithm
is able to adapt to underconstrained graphs. This modication, however, increases the
complexity by a factor of n.
Proof. Suppose that the graph G is underconstrained. Consider the maximum ow
found by MM in the network X corresponding to G, as described in the section Redening
Solubility. There are two cases.
The rst case is when the last vertex v
m
in X corresponds to an edge of some solv-
able subgraph A
1
. Then all vertices v
i
in X corresponding to vertices of A
1
have their
capacities completely lled, since A
1
is solvable. On the other hand, at least one vertex
v of G will not have its capacity lled completely, since G is underconstrained. Let W be
the set of such vertices v. The new modication of MM would proceed by removing all
vertices of X corresponding to vertices of W and edges adjacent to W, as well as ows
originating at such edges. Once this is done, a new set W is recomputed and removed,
until all remaining vertices of X that represent vertices of G have their capacities lled
completely. These vertices comprise a solvable subgraph A
1
such that no supergraph of
A
1
is solvable. Once the subgraph A
1
(and its DR-plan Q
1
) is found, it could be removed
from G and the process applied recursively to G\ A
1
to nd a solvable subgraph A
2
, its
DR-plan Q
2
, etc.
The second case is when the last vertex v
m
in X is not contained in any solvable
subgraph. Then constructing and removing sets W as described above will completely
exhaust G, without nding any solvable graphs. The new modication of MM would
proceed by nding another maximum ow in network X corresponding to G such that
the last vertex v
i
m
is dierent from v
m
. Again two cases are considered for vertex v
i
m
.
If all v
i
m
G are not contained in any solvable subgraph, then G does not contain any
solvable subgraph and the process terminates.
Note that this modication ohand increases the complexity of MM by a factor of n,
and there does not seem to be any obvious way to prevent this factor.
This modication of MM outputs a DR-plan Q = Q
1
, . . . , Q
k
for a set of solvable
Decomposition of Geometric Constraints I 405
subgraphs A
1
, . . . , A
k
such for any i no supergraph of A
i
is solvable and there are no
solvable subgraphs B G such that B A
j
= , j. Thus this modication of MM is
able to adapt to underconstrained graphs. 2
Observation 4.14. (i) Under the concession MM1, MM has the ChurchRosser prop-
erty since simplifying any S
i
that is solvable at the current stage preserves the density of
the whole graph G
i+1
(i.e. G
i+1
is solvable if and only if G
i
is).
(ii) Under the concession MM1, MM is able to incorporate a design decomposition P
specied by the designer if and only if for every P
k
, P
t
P such that P
k
P
t
= either
P
k
P
t
or P
t
P
k
.
Proof is similar to the corresponding proof for the SR algorithm in Claim 4.7.
4.3. comparison of performance
Next, we give a table comparing the SR and MM DR-planners with respect to the
performance measures of Section 3. Underconstr. refers to the ability to deal with
underconstrained graphs, Design decomposition refers to the ability to incorporate
design decompositions specied by the designer.
Performance measure SR MM
Generality No Yes

Underconstr. No(Yes

) Yes
,
Design decomposition No(Yes
,
) No(Yes

,
)
Validity No(Yes

) Yes

Solvability No(Yes

) Yes

Strict solvability No(Yes

) Yes

Complete No(No

) Yes

Worst approximation factor 0 (O(

1
n
)

) O(
1
n
)

Best approximation factor 0 (O(

1
n
)

) O(
1
n
)

ChurchRosser No(Yes

) Yes

Complexity O((m+n)
2
) O(n(m+n))

The superscript * refers to a narrow class of DR-plans: those that require the solvable
subsystems S
i
to be based on triangles or a xed repertoire of patterns. The superscript
refers to results that were not true for the original MM algorithm developed by Ait-
Aoudia et al. (1993) and Pabon (1993) and proved in this paper through a modication
of MM described in Section 4.2. The superscript refers to a restricted class of graphs
in which there are no overconstrained subgraphs. It also refers to a further modication
of MM described in Claim 4.13, which however, increases the complexity by a factor of
n. The superscript refers to strong restrictions on the design decompositions that
can be incorporated into DR-plans by SR and MM.
Acknowledgements
We thank two anonymous reviewers for their careful reading of the paper, and one
of them for suggesting the modication which we completed and which signicantly
improves the MM algorithm, as described in Section 4.2.
406 C. M. Homan et al.
References
Ait-Aoudia, S., Jegou, R., Michelucci, D. (1993). Reduction of constraint systems. In Compugraphics, pp.
8392. Alvor, Portugal.
Blum, L., Shub, M., Smale, S. (1989). On a theory of computation and complexity over the real numbers:
NP-completeness, recursive functions and universal machines. Bull. Am. Math. Soc., 21, 146.
Bouma, W., Fudos, I., Homann, C., Cai, J., Paige, R. (1995). A geometric constraint solver. Comput.
Aided Des., 27, 487501.
Bronsvoort, W. F., Jansen, F. W. (1994). Multiview feature modeling for design and assembly. In Shah,
J., Mantyla, M., Nau, D. S. eds, Advances in Feature Based Modeling, pp. 315330. Amsterdam,
Elsevier Science.
Canny, J. (1993). Improved algorithms for sign determination and existential quantier elimination.
Comput. J., 36, 409418.
Canny, J., Emiris, I. (1993). An ecient algorithm for the sparse mixed resultant. In Cohen, G., Mora, T.,
Moreno, O. eds, Proceedings of the 10th International Symposium on Applied Algebra, Algebraic Al-
gorithms, and Error Correcting Codes, LNCS 263, pp. 89104. Berlin, Springer-Verlag.
Chou, S. C., Gao, X. S., Zhang, J. Z. (1996). A method of solving geometric constraints. Technical
Report, Wichita State University, Department of Computer Science.
Collins, G. (1975). Quantier elimination for real closed elds by cylindrical algebraic decomposi-
tion, LNCS 33, pp. 134183. Berlin, Springer-Verlag.
Cox, D., Little, J., OShea, D. (1998). Using Algebraic Geometry. New York, Springer-Verlag.
Crippen, G., Havel, T. (1988). Distance Geometry and Molecular Conformation. New York, John Wiley
& Sons.
Durand, C. (1998). Symbolic and numerical techniques for constraint solving. Ph.D. Thesis, Purdue
University, Computer Science Department.
Fang, S. (1992). Robustness in geometric modeling. Ph.D. Thesis, University of Utah.
Fudos, I. (1995). Geometric constraint solving. Ph.D. Thesis, Purdue University, Department of Com-
puter Science.
Fudos, I., Homann, C. M. (1996a). Constraint-based parametric conics for CAD. Comput. Aided Des.,
28, 91100.
Fudos, I., Homann, C. M. (1996b). Correctness proof of a geometric constraint solver. Int. J. Comput.
Geom. Appl., 6, 405420.
Fudos, I., Homann, C. M. (1997). A Graph-Constructive Approach to Solving Systems of Geometric
Constraints. ACM Trans. Graph., 16, 179216.
Gabow, H., Westermann, H. (1988). Forests, frames and games: Algorithms for matroid sums and ap-
plications. In Proceedings of the Twentieth Annual ACM Symposium on the Theory of Computing,
Chicago, Illinois, U.S.A., 24 May 1998, pp. 407421. New York, ACM Press.
Gao, X. S., Chou, S. C. (1998a). Solving geometric constraint systems. I. A global propagation approach.
Comput. Aided Des., 30, 4754.
Gao, X. S., Chou, S. C. (1998b). Solving geometric constraint systems. II. A symbolic approach and
decision of RC-constructibility. Comput. Aided Des., 30, 115122.
Graver, J., Servatius, B., Servatius, H. (1993). Combinatorial Rigidity, Graduate Studies in Mathematics.
Grigorev, D. Y., Vorobjov, N. N. (1988). Solving systems of polynomial inequalities in subexponential
time. J. Symb. Comput., 5, 3764.
Hastad, J. (1996). Clique is hard to approximate within n
1
. In Proceedings of the 37th IEEE Sympo-
sium on Foundations of Computer Science, pp. 627636. Burlington, IEEE press.
Havel, T. (1991). Some examples of the use of distances as coordinates for Euclidean geometry. J. Symb.
Comput., 11, 579594.
Hendrickson, B. (1992). Conditions for unique graph realizations. SIAM J. Comput., 21, 6584.
Homann, C. M. (1997). Solid modeling. In Goodman, J. E., ORourke, J. eds, CRC Handbook on
Discrete and Computational Geometry. Boca Raton, FL, CRC Press.
Homann, C. M., Joan-Arinyo, R. (1998). CAD and the product master model. Comput. Aided Des.,
30, 905919.
Homann, C. M., Lomonosov, A., Sitharam, M. (1997). Finding solvable subsets of constraint graphs. In
Proceedings of Principles and Practice of Constraint Programming 97, Linz, Austria, LNCS 1330,
pp. 463477. Berlin, Springer-Verlag.
Homann, C. M., Lomonosov, A., Sitharam, M. (1998). Geometric constraint decomposition. In Brud-
erlin, Roller eds, Geometric Constraint Solving. Berlin, Springer-Verlag.
Homann, C. M., Peters, J. (1995). Geometric constraints for CAGD. In Daehlen, M., Lyche, T., Schu-
maker, L. eds, Mathematical Methods for Curves and Surfaces, pp. 237254. Vanderbilt University
Press.
Homann, C. M., Rossignac, J. (1996). A road map to solid modeling. IEEE Trans. Vis. Comput.
Graphics, 2, 310.
Homann, C. M., Vermeer, P. J. (1994). Geometric constraint solving in R
2
and R
3
. In Du, D. Z.,
Hwang, F. eds, Computing in Euclidean Geometry, 2
nd
edn. Singapore, World Scientic Publishing.
Decomposition of Geometric Constraints I 407
Homann, C. M., Vermeer, P. J. (1995). A spatial constraint problem. In Workshop on Computational
Kinematics, France. Sophia-Antipolis, INRIA.
Hopcroft, J. E., Tarjan, R. E. (1973). Dividing a graph into triconnected components. SIAM J. Comput.,
2, 135158.
Hsu, C. (1996). Graph-based approach for solving geometric constraint problems. Ph.D. Thesis, Univer-
sity of Utah, Department of Computer Science.
Imai, H. (1985). On combinatorial structures of line drawings of polyhedra. Discrete Appl. Math., 10,
7992.
Itai, A., Rodeh, M. (1978). Finding a minimum circuit in a graph. SIAM J. Comput., 4, 413423.
Khovanskii, A. G. (1978). Newton polyhedra and the genus of complete intersections. Funktsionalnyi
Analiz i Ego Prilozheniya, 12, 5161.
Klein, R. (1996). Geometry and feature representation for an integration with knowledge based systems.
In Geometric Modeling and CAD. Florida, U.S.A., Chapman-Hall.
Klein, R. (1998). The role of constraints in geometric modeling. In Bruderlin, Roller eds, Geometric
Constraint Solving and Applications. Berlin, Springer-Verlag.
Kraker, K. J., Dohmen, M., Bronsvoort, W. F. (1997). Maintaining multiple views in feature modeling.
In ACM/SIGGRAPH Symposium on Solid Modeling Foundations and CAD/CAM Applications, pp.
123130. New York, ACM press.
Kramer, G. (1992). Solving Geometric Constraint Systems. Cambridge, MA, U.S.A., MIT Press.
Laman, G. (1970). On graphs and rigidity of plane skeletal structures. J. Eng. Math., 4, 331340.
Latham, R., Middleditch, A. (1996). Connectivity analysis: a tool for processing geometric constraints.
Comput. Aided Des., 28, 917928.
Lazard, D. (1981). Resolution des syst`emes dequations algebriques. Theor. Comput. Sci., 15, 77110.
Lazard, D. (1991). A new method for solving algebraic systems of positive dimension. Discrete Appl.
Math., 33, 147160.
Mantyla, M., Opas, J., Puhakka, J. (1989). Generative process planning of prismatic parts by feature
relaxation. In Advances in Design Automation, Computer Aided and Computational Design, pp.
4960. Sacramento, CA, U.S.A., ASME.
Middleditch, A., Reade, C. (1997). A kernel for geometric features. In ACM/SIGGRAPH Symposium
on Solid Modeling Foundations and CAD/CAM Applications. New York, ACM press.
Newell, M. E., Evans, D. C. (1976). Modeling by computer. In IFIP Working Conference on CAD
Systems, Austin, TX, pp. 291297. Amsterdam, North-Holland.
Owen, J. (1991). Algebraic solution for geometry from dimensional constraints. In ACM Symposium
Found. of Solid Modeling, Austin, TX, pp. 397407. ACM.
Owen, J. (1996). Constraints on simple geometry in two and three dimensions. Int. J. Comput. Geom.
Appl., 6, 421434.
Pabon, J. A. (1993). Modeling method for sorting dependencies among geometric entities. United States
Patent 5,251,290.
Renegar, J. (1992). On the computational complexity and the rst order theory of the reals, Part I. J.
Symb. Comput., 13, 255299.
Ruiz, O. E., Ferreira, P. M. (1996). Algebraic geometry and group theory in geometric constraint satis-
faction for computer-aided design and assembly planning. IIE Trans. Des. Manuf., 28, 281294.
Saliola, F., Whiteley, W. (1999). Constraint congurations in CAD: circles, lines and angles in the plane.
Preprint, York University.
Semenkov, O. I. (1976). An experimental CAD/CM system. In 3rd International IFIP/IFAC Conference
on Programming Languages for Machine Tools, Stirling, Scotland, pp. 397403. Amsterdam, North-
Holland.
Serrano, D. (1990). Managing constraints in concurrent design: rst steps. In Proceedings of Computers
in Engineering, Boston, MA, 1990, pp. 159164.
Serrano, D., Gossard, D. C. (1986). Combining mathematical models with geometric models in CAE
systems. Computers in Engineering, Chicago, IL. ASME, 1, 277284.
Sridhar, N., Aggarwal, R., Kinzel, G. L. (1993). Active occurence matrix based approach to design
decomposition. Comput. Aided Des., 25, 500512.
Sridhar, N., Aggarwal, R., Kinzel, G. L. (1996). Algorithms for the structural diagnosis and decomposi-
tion of sparse, underconstrained, systems. Comput. Aided Des., 28, 237249.
Sturmfels, B. (1993). Sparse elimination theory. In Proceedings of the Computational Algebraic Geometry
and Commutative Algebra, pp. 377396. Cambridge, Cambridge University Press.
Sugihara, K. (1985). Detection of structural inconsistency in systems of equations with degrees of freedom
and its applications. Discrete Appl. Math., 10, 297312.
Tay, T. (1999). On the generic rigidity of bar frameworks. Adv. Appl. Math., 23, 1428.
Tay, T., Whiteley, W. (1985). Generating isostatic frameworks. Topologie Structurale, 11, 2169.
Wang, D. (1993). An elimination method for polynomial systems. J. Symb. Comput., 16, 83114.
Whiteley, W. (1992). Matroids and rigid structures. In Matroid Applications, volume 40 of Encyclopedia
Math. Appl., pp. 153. Cambridge, Cambridge University Press.
408 C. M. Homan et al.
Whiteley, W. (1997). Rigidity and scene analysis. In Handbook of Discrete and Computational Geome-
try, pp. 893916. Florida, U.S.A., CRC Press.
Originally Received 17 February 1998
Accepted 11 June 2000