scieee Science in your language
[en] (orig)

Revisiting variable radius circles in constructive geometric constraint solving

Abstract

Variable-radius circles are common constructs in planar constraint solving and are usually not handled fully by algebraic constraint solvers. We give a complete treatment of variable-radius circles when such a circle must be determined simultaneously with placing two groups of geometric entities. The problem arises for instance in solvers using triangle decomposition to reduce the complexity of the constraint problem.

Read accessible full text

Revisiting variable radius circles in constructive geometric constraint solving

Author: Ching-Shoei, C,Joan Arinyo, Robert
Year: 2002
Source: https://upcommons.upc.edu/bitstream/2117/97537/2/R02-25.pdf
Re isi ing Va iable Radius Ci cles in Cons uc i e
Geome ic Cons ain Sol ing
Ching-Shoei Chiang
Depa men o Compu e and In o ma ion Science
Soochow Uni e si y
Taiwan, R.O.C.
Robe Joan-A inyo
Depa amen de Llengua ges i Sis emes In o m`a ics
Uni e si a Poli `ecnica de Ca alunya
Ba celona, Ca alonia, Spain
[email p o ec ed], [email p o ec ed]
Ma ch 13, 2002
Abs ac
Va iable- adius ci cles a e common cons uc s in plana cons ain
sol ing and a e usually no handled ully by algeb aic cons ain sol e s.
We gi e a comple e ea men o a iable- adius ci cles when such a
ci cle mus be de e mined simul aneously wi h placing wo g oups o
geome ic en i ies. The p oblem a ises o ins ance in sol e s using i-
angle decomposi ion o educe he complexi y o he cons ain p ob-
lem.
This wo k offe s a se o basic cons uc i e me hods ha pe mi s o
de e mine a iable adius ci cles simul aneously wi h placing wo igid
geome ic objec s when geome ic cons ain s a e defined on bo h he
ci cum e ence and cen e poin o he cons ain ci cle. The p oblem
has been classified by he geome ic en i ies in wo g oups, one is fixed
and he o he has ansla ional and o a ional mo emen , so ha he
a iable adius ci cles sa is y he cons ain s on he geome ic en i ies
in hese wo g oups. The numbe o solu ions o each p oblem is also
gi en.
Keywo ds: Geome ic cons ain sol ing, a iable adius ci cles, con-
s uc i e sol e s, algeb aic sol e s, cyclog aphic maps.
1
1 In oduc ion
In cons ain -based geome ic design, he designe c ea es a ough ske ch o
an objec made ou o simple geome ic elemen s. Then he in ended exac
shape is specified by anno a ing he ske ch wi h cons ain s. A geome ic
cons ain sol e hen checks whe he he se o geome ic cons ain s cohe -
en ly defines he objec and, i so, de e mines he posi ion o he geome ic
elemen s.
Many echniques ha e been epo ed in he li e a u e ha p o ide pow-
e ul and efficien me hods o sol ing sys ems o geome ic cons ain s. Fo
example, see [2] and e e ences he ein o an ex ensi e analysis o wo k
on cons ain sol ing. Among hem, ou in e es ocuses on cons uc i e
echniques,
Cons uc i e sol e s ha e wo majo componen s: he analyze and he
cons uc o . The analyze symbolically de e mines whe he a geome ic
p oblem defined by cons ain s is sol able. I he p oblem is sol able, he
ou pu o he analyze is a sequence o cons uc ion s eps, known as he
cons uc ion plan, ha places each geome ic elemen in such a way ha
cons ain s a e sa isfied. A e assigning specific alues o he pa ame e s,
he cons uc o in e p e s he cons uc ion plan and builds an objec in-
s ance, p o ided ha no nume ical incompa ibili ies a ise.
The complexi y o geome ic cons ain sol ing is doubly exponen ial, a
ca ha de i es om he abili y o exp ess polynomial algeb aic equa ions
by geome ic cons ain configu a ions. As a esul , i is accep ed ha p ac-
ical sol e s a e no comple e, ha is, hey sol e a subclass o geome ic
p oblems.
A p ac ical use ul class o p oblems a e wodimensional cons ain p ob-
lems whe e he geome ic elemen s a e poin s, s aigh lines, and ci cles wi h
fixed adii, and in which he cons ain s a e like dis ance be ween wo poin s,
dis ance om a poin o a line, angle be ween wo lines, line-ci cle angency
and so on.
Va ious ex ensions o he geome ic epe oi e ha cons uc i e sol e s
can handle in wo dimensions can be conside ed. Exemples a e ci cles wi h
a iable adius, conics and B´ezie cu es. O hem, a iable adius ci cles
a e common cons uc s in wo dimensional cons ain sol ing and a e usually
no handled ully by cons uc i e sol e s. P obably hey a e he mos use ul
ex ension as hey pe mi auxilia y cons uc ion in addi ion, as explained
2
by Hoffmann and Ve mee , [9], Hoffmann and Joan-A inyo, [8], and Joan-
A inyo and So o, [10].
When he unde lying sol e is nume ical and good ini ial guesses a e
a ailable o he geome ic elemen s, a iable adius ci cles pose no pa ic-
ula p oblem. Bu he nume ical app oach o sol ing cons ain s has many
d awbacks, including eliance on good s a ing alues and he inabili y o
explo e solu ion a ian s, [2]. Wha is needed is a cons uc i e solu ion,
p e e able one in which he e is no need o sol e high-deg ee polynomials.
Recen ly, Hoffmann and Chiang, [6, 7], epo ed on an ex ension o he
basic cons uc s o deal wi h a iable adius ci cles in cons uc i e sol e s.
He e, he cons ain s on he a iable adius ci cle a e placed only on he
ci cum e ence.
In his wo k we u he ex end he basic cons uc ions o conside con-
s ain p oblems in which a iable adius ci cles occu wi h cons ain s de-
fined on hei cen e poin s.
The es o he pape is o ganized as ollows. In Sec ion 2 we gi e a sho
o e iew on ela ed wo k. Nex in Sec ion 3 we define a minimal se o ools
a use in e ace should p o ide o define a iable adius ci cles. Sec ion 4
ecalls he undamen al concep s o he cyclog aphic model geome y we
will make use o . Gene al algo i hms o he cons uc ions which sol e he
p oblem conside ed he e a e gi en in Sec ion 5. In Sec ion 6 we p esen
sol ing es a egies o keep o a minimum he complexi y o he algo i hms
implemen a ion. Finally we offe some conclusions in Sec ion 7.
2 P io Wo k
The e is a pauci y o published wo ks epo ing on a iable adius ci -
cles in cons uc i e geome ic cons ain sol ing. Ramana han, [14], s ud-
ied he Apollonious p oblem which consis s in cons uc ing a ci cle an-
gen o h ee gi en ci cles. The wo k add essed wo p oblems: De ising a
coo dina e-independen enume a ion me hod o he eigh possible solu ions
and pe o ming he compu a ions efficien ly. The echnique was applied o
cons ain -based, a iable adius fi ing o fille s o wo lines.
Joan-A inyo and So o desc ibed in [10] a hyb id echnique ha allows o
sol e cons ain p oblems in ol ing geome ic elemen s wi h mo e han wo
deg ees o eedom. In pa icula i is shown how he me hod sol es a iable
3
Co
e1
e2
e3
C(Co, )Rigid clus e
Figu e 1: Va iable adius ci cle C(C0, ) a ached o one clus e .
adius ci cles a ached o one geome ic objec which is de e mined up o
posi ion and o ien a ion, om now on e e ed o as a clus e [3], h ough
h ee cons ain s. See Figu e 1.
Hoffmann and Chiang ecen ly, [6, 7], epo ed on a mo e gene al ap-
p oach o cons uc i ely sol ing cons ain p oblems in ol ing a iable a-
dius ci cles. The app oach uses cyclog aphic maps, a special case o Lague e
geome y, [5], and handles he si ua ion whe e he a iable adius ci cle,
C(C0, ), (see Figu e 2) is a ached o wo clus e s, S1and S2,whichsha e
a common geome ic elemen , E. The o al numbe o deg ees o eedom
ha need o be canceled o S1,S2and he a iable adius ci cle o define
a clus e is ou : h ee o he ci cle i sel plus one due o he possible ela-
i e mo ion be ween S1and S2along E. These cons ain s a e canceled by
a aching he a iable adius ci cle o each clus e h ough wo cons ain s.
No e ha , o he wise he p oblem could be educed o he p e ious case. A
limi a ion o he me hod is ha i only conside s cons ain s placed on he
ci cum e ence o he a iable adius ci cle.
The DCM is a comme cial sol e , [1], which pe mi s sequen ial cons uc-
ions o a iable adius ci cles when hey a e a ached o one gi en clus e
h ough h ee cons ain s. As a as we know, [11, 12], no de ails ha e been
disclosed abou how he a iable adius ci cles a e handled.
3 De ini ion o Va iable Radius Ci cles
To define geome ic p oblems in ol ing ci cles wi h a iable adius, he use
in e ace should p o ide an app opia e se o ools. A minimal se o ools
4
Co
e
e11
e12
e1m
e21 e22
S1
C(Co, )
S2
e2n
Figu e 2: Va iable adius ci cle C(C0, ) a ached o wo clus e s.
would include an explici command o igge he a iable adius ci cle de -
ini ion along wi h ope a ions o place geome ic cons ain s on i s ci cum-
e ence and on i s cen e .
We assume ha he geome ic elemen s om which cons ain p oblems
a e buil a e poin s, s aigh lines and ci cles. A sufficien se o cons ain s
o define a iable adius ci cles includes angencies and dis ances.
3.1 Cons ain s Placed on he Ci cum e ence
We define he angency cons ain s placed on he ci cum e ence as ollows.
See Figu e 3. The ci cum e ence o he a iable adius ci cle can be angen
o
P
Qo
C
Co d
d
d
on
LQ
Figu e 3: Cons ain s placed on he ci cum e ence o a a iable adius ci cle.
5

•The ci cum e ence o a fixed adius ci cle.
•A s aigh line.
•A poin . This is he usual on cons ain .
Dis ance cons ain s placed on he ci cum e ence o he a iable adius ci cle,
see Figu e 3, a e defined as ollows
•Dis ance o a fixed adius ci cle Q: The minimum dis ance be ween
he wo ci cles measu ed along he s aigh line defined by hei cen e
poin s.
•Dis ance o a s aigh line L: The minimum dis ance be ween he
ci cle and he line measu ed along he pe pendicula o he gi en line
h ough he ci cle cen e poin .
•Dis ance o a poin P: Dis ance be ween he ci cle and he poin
measu ed along he s aigh line defined by he poin and he cen e
o he ci cle.
3.2 Cons ain s Placed on he Cen e Poin
Cen e poin s o a iable adius ci cles ha e no p i ileges o e o he poin s.
The e o e he se o cons ain s ha apply o gene ic poin s apply also o
cen e poin s o ci cles wi h a iable adius. We assume ha he cons ain s
a ailable a he use in e ace a e, see Figu e 4,
•The cen e poin can be a a gi en dis ance om ano he geome ic
elemen .
•The cen e poin can be on ( angen o) ano he geome ic elemen .
To acili a e he use in e ac ion, o he cons ain s could be added o he
epe oi e so a p esen ed.
4 The Cyclog aphic Model
Se e al geome ic design p oblems can be sol ed in a su p isingly simple
way i one uses Lague e geome y. A specific case o his geome y, known
6
P
Qo
C
Co
don
on
LQ
on
dd
Figu e 4: Cons ain s placed on he cen e o a a iable adius ci cle.
as he cyclog aphic model, esul s pa icula ly use ul o sol e he p oblem
we ha e a hand, [14].
Fo he sake o comple ness, we ecall he undamen al concep s o he
cyclog aphic model o he embedding o space R2in R3we will make use
o . Fo a gene al and mo e in dep h discusion on he cyclog aphic model
and i s applica ions o compu e aided geome ic design see Hoffmnann [5],
and Po mann and Pe e nell [13].
The undamen al elemen s in R2we conside a e ays and cycles.A ay
is an o ien ed s aigh line. A cycle is an o ien ed ci cle o a poin (cycle
wi h adius 0). The o ien a ion is fixed by a uni no mal ec o field in he
case o ays and by a signed adius in he case o he cycle.
The basic ela ion is ha o o ien ed con ac be ween cycles and ays.
Re e o Figu e 5. An o ien ed cycle and a ay a e in o ien ed con ac , i
hey a e angen and he uni no mals coincide a he poin o con ac . Fo
a poin and a ay, o ien ed con ac equals incidence.
L
C
ab
L
C
Figu e 5: Con ac ay-cycle. a) O ien ed. b) No o ien ed.
7
γ
C
a
π/4
C
(a, b)
b
π/4
π/4
P
γ
P
Figu e 6: Cyclog aphic maps. a) Cycle. b) Poin .
Le C(a, b, ) deno e a cycle wi h cen e poin (a, b) and signed adius
. We assume ha when >0, he cycle is o ien ed coun e clockwise; i
<0, he cycle is o ien ed clockwise. When = 0, he cycle ep esen s a
poin and is conside ed o ha e bo h o ien a ions simul aneously.
Wi h each cycle C(a, b, ) he e is an associa ed cyclog aphic map,de-
no ed by γC, defined as he cone whose apex is he poin (a, b, )inR3,
whose axis is pa allel o he Zaxis and whose angle is equal o π/4. Fig-
u e 6 illus a es his concep .
Conside he line in R2whose equa ion is ax +by +c=0. No e ha ,
depending on he o ien a ion, a s aigh line can suppo wo diffe en ays.
The o ien a ion o a ay, gi en by i s di ec ion ec o , is defined as he ec o
[b, −a]; ha is, he esul o o a ing clockwise by 90◦ he ec o [a, b], which
is no mal o he line. We shall deno e a ay by L(a, b, c)o jus L.
Wi h each ay L(a, b, c) he e is an associa ed cyclog aphic map, deno ed
by γL, defined as he plane in R3which in e sec s he XY plane a Land
a an angle wi h [b, −a]equal oπ/4. See Figu e 7.
n
LL
π/4
γ
L
Figu e 7: Cyclog aphic map o a ay.
8
The dis ance o a poin o a ay is measu ed as a posi i e quan i y i
he poin is o he le o he ay as seen in he ay’s o ien a ion. The
angle be ween a pai o ays, ∠(Li,L
j) is measu ed om he di ec ion o Li
clockwise o he di ec ion o Lj.
5 Sol ing Va iable Radius Ci cles
The se o cons ain s gi en in Sec ion 3 e e o he ools a ailable a he use
in e ace o p o ide a iendly in e ac ion. To acili a e he sol ing p oces, we
fi s show how o ans o m he p oblem defined a he use in e ace in o an
equi alen p oblem, whe e dis ance cons ain s on he a iable adius ci cle
a e exp essed as angencies.
Nex , o handle he cons ain s defined on he cen e poin o a iable
adius ci cles in a uni o m and consis en way, we ex end he cyclog aphic
model wi h wo new auxilia y maps.
Then we gi e gene al algo i hms ha compu e a iable adius ci cles ha
a e a ached o wo igid clus e s h ough cons ain s placed on bo h, he
ci cum e ence and he cen e poin o he ci cle. Following [6, 7], we conside
wo diffe en scena ios: 1) The geome ic elemen Esha ed by clus e s S1
and S2, see Figu e 2, is a s aigh line and ela i e mo ion is ansla ional,
and 2) Eis a ci cle o poin and he ela i e mo ion o clus e s is a o a ion.
5.1 P oblem T ans o ma ion
Fi s we conside he dis ance cons ain s placed on he ci cum e ence o he
a iable adius ci cle o be de e mined, C(C0, ). The ci cle-line dis ance
cons ain , dis(C, L)=d, see Figu e 8, is equi alen o a angency cons ain
be ween he ci cle Cand a line Lwhich has been ansla ed a dis ance d
along i s no mal. Le Q(Q0, ) be a ci cle wi h cen e Q0andfixed adius
. he dis ance cons ain dis(C, Q(Q0, )) = dis ans o med in o he an-
gency cons ain (C, Q(Q0, +d)). Finally, he dis ance cons ain be ween
Cand poin P,dis(C, P)=d, is ans o med in o he equi alen angency
cons ain (C, Q(P,d)).
Dis ance cons ain s placed on he cen e poin C0o C(C0, )a e ans-
o med in o angency (on) cons ain s as ollows. The cen e poin -poin
dis ance, dis(C0,P)=dis ans o med in o (C0,Q(P,d)). The cen e
poin -line cons ain dis(C0,L)=dis ans o med in o (C0,L
), whe e L
9
P oblem 2cons ain s 1cons ain
E1: E(LL,LL)(1,4) E11: E(LL,L’L’)(1,2) E13: E(LL,LL’)(1,2)
E12: E(LL’,LL’)(1,2)
E2: E(CL,LL)(2,8) E21: E(CL,L’L’)(2,4) E25: E(CL,LL’)(2,4)
E22: E(CL’,LL’)(2,4) E26: E(CL’,LL)(2,4)
E23: E(C’L,LL’)(2,4) E27: E(C’L,LL)(2,4)
E24: E(C’L’,LL)(2,4)
E3: E(CL,CL)(4,16) E31: E(CL,C’L’)(4,8) E35: E(CL,CL’)(4,4)
E32: E(CL’,CL’)(4,4) E36: E(CL,C’L)(4,16)
E33: E(C’L,CL’)(4,16)
E34: E(C’L,C’L)(4,4)
E4: E(CC,LL)(4,16) E41: E(CC,L’L’)(2,4) E44: E(CC,LL’)(2,4)
E42: E(CC’,LL’)(4,16) E45: E(CC’,LL)(4,16)
E43: E(C’C’,LL)(2,4)
E5: E(CC,CL)(8,32) E51: E(CC,C’L’)(4,8) E55: E(CC,CL’)(4,4)
E52: E(CC’,CL’)(8,16) E56: E(CC,C’L)(4,16)
E53: E(CC’,C’L)(8,16) E57: E(CC’,CL)(8,16)
E54: E(C’C’,CL)(4,8)
E6: E(CC,CC)(16,64) E61: E(CC,C’C’)(4,8) E63: E(CC,CC’)(8,16)
E62: E(CC’,CC’)(16,64)
Table 1: Gene al me ge p oblem classifica ion.
we will deno e by Land C he geome ic elemen s in a me ge p oblem on
which cen e cons ain s ha e been defined. Fo example C(LL,CL) will
deno e ha one line in he fixed clus e and he ci cle in he mo ing clus e
suppo cen e cons ain s wi h espec o he a iable adius ci cle.
A gene al classifica ion o he ansla ional and o a ional me ge p ob-
lem is shown in Table 1. The fi s column indica es he p oblems wi h
ci cum e ence cons ain s only, he second and hi d columns show he di -
e en p oblems wi h wo and one cen e cons ain s espec i ely. The pai
(m, n) a e each p oblem indica es he maximum numbe o solu ions, m
o ansla ional and n o o a ional p oblems. We will explain la e on how
hese figu es ha e been de i ed.
16

6.2 Deg ee o he Equa ions
To simpli y he p oblem, we assume ha in he ansla ional case he s aigh
line sha ed by he clus e s is coinciden wi h he X-axis and ha in he
o a ional p oblem he cycle sha ed by he clus e s is cen e ed a he o igin.
Using he no a ion al eay in oduce in Sec ion 5 and wi h simple geom-
e y, we ha e he ollowing heo em
Theo em 6.1 The equa ions o γL,τL,γL(d), and τL(d) a e deg ee one.
The equa ions o γC,τC,γC(d), τC(d), γL(θ), and τL(θ) a e deg ee wo.
And, he equa ions o γC(θ)andτC(θ) a e deg ee 4 equa ions.
No ice ha he equa ions o γL,τL,γC and τC ha e 3 a iables x, y, z,
and he equa ions o γL(d), τL(d), γC(d), τC(d), γL(θ), τL(θ), γC(θ)and
τC(θ) add one mo e a iable, d.
As men ioned in Sec ion 5, he a iable adius ci cle can be ound by
in e sec ing ou su aces, each su ace being a γ-map o a τ-map, depend-
ing on whe he he cons ain on he a iable adius cycle is a ci cum e ence
cons ain o a cen e cons ain . The sys em o equa ions can be easily gen-
e a ed by gene a ing he equa ions o he maps associa ed o each geome ic
elemen in ol ed. Fo example, he solu ion o he p oblem L(CL,CL)can
be figu ed ou by finding he in e sec ion
γC1∩τL
2∩τC
3(d)∩γL4(d)
Since equa ions τL
2,andγL4(d) a e deg ee one and γC1and τC
3(d)a e
deg ee wo equa ions, om Bezou ’s heo em, [4], we know he e a e 4 so-
lu ions o his sys em. By using Bezou ’s heo em, e e y subp oblem Eij
o p oblem Ei, see Table 1, has he same numbe o solu ions. Fo example,
e e y subp oblem o L1, including L11,L12 and L13, has only one solu ion,
and e e y subp oblem o C1, including C11,C12 and C13, has ou solu-
ions. The numbe o solu ions o each p oblem is summa ized in Table 1.
In he ollowing sec ion, we will de i e s a egies o each me ge p oblem
subclass o educe hese numbe s o compu e a a iable adius ci cle.
No ice ha he numbe o solu ions gi en is based on ou app oach
by using he γcyclog aphic maps and τmaps. I we do no use hese
maps, he numbe o solu ions mus be mul iplied by 8, he numbe o
essen ially dis inc o ien a ions o lines and cycles. Fo example, he numbe
o solu ions o he p oblems L(LL, LL)andC(LL, LL) would be 8 and 32,
espec i ely.
17
6.3 Algo i hms o he T ansla ional Me ge P oblem
The subp oblems gi en in Table 1 a e indi idualized as ansla ional me ge
p oblems by eplacing Eij wi h Tij. To de i e s a egies o educe he de-
g ee o he equa ions o be sol ed, we g oup he ansla ional me ge p oblems
in ol h ee diffe en classes as ollows:
1. P oblems wi h wo cen e cons ain s defined in he same clus e . This
includes T11, T21, T24, T31, T41, T43, T51, T54 and T61.
2. P oblems ha can be ans o med in o he in e sec ion o h ee planes
and one su ace. This includes T12, T13, T22, T23, T25, T26, T27,
T32, T34, T35, T44, T55, and pa o he p oblems in he p e ious
class.
3. P oblems ha can be ans o med in o he in e sec ion o one γC,one
τC, and wo planes. This includes T33, T36, T42, T45, T52, T53,
T56, T57, T62 and T63.
In he ollowing h ee subsec ions we p esen , om easy o ha d, he
solu ion o he p oblems lis ed abo e.
6.3.1 Two cen e cons ain s in he same clus e
This p oblem can be sol ed by o cing he clus e wi h wo cen e cons ain s
o be he mo ing clus e , so he p oblem becomes L(E1E2,E
3E
4). No ice
ha i he mo ing clus e has wo cen e cons ain s he clus e fixed has
no cen e cons ain s. F om he ac ha he cone-cone in e sec ion can
be ans o med in o cone-plane in e sec ion, [6, 7], he in e sec ion o he
geome ic elemen s in he fi s clus e can always be ep esen ed by he
in e sec ion o he map o he fi s elemen E1wi h a fixed plane. The
algo i hm o sol e he p oblem is:
Algo i hm L(E1E2,E
3E
4)
1. Find he poin (x, y)=τE
3∩τE
4.
2. Subs i u e he poin (x+d, y) in o he equa ion o he plane
gene a ed in he fi s clus e and find z(d)whichisadeg ee
one equa ion o a iable d.
3. Subs i u e he poin (x+d, y, z(d)) in o he equa ion o E1
o yield one equa ion in d.
18
4. Sol e he equa ion o find he alue o d.
5. The a iable adius ci cle seeked has (x+d, y)
as cen e poin and z(d)as adius.
The ollowing h ee heo ems gi e he ools needed o figu e ou he
in e sec ion o τE
3and τE
4. The fi s heo em is i ial and applies when
bo h E
3and E
4a e s aigh lines.
Theo em 6.2 Two diffe en s aigh lines Li=[ai,b
i,e
i]andL2=[aj,b
j,e
j]
which a e no pa allel, in e sec a poin
(−eibj+ejbi,−aiej+ajei,a
ibj−ajbi)
The second heo em educes he line-ci cle in e sec ion o wo line-line
in e sec ions.
Theo em 6.3 Le L=[a, b, e] be a s aigh line and C=(x, y, )aci cle.
In e sec ing Land Cis equi alen o in e sec ing Land L=[a,b
,e
]whe e
a=a(2 )+b(1 − 2)
b=−a(1 − 2)+b(2 )
e=−M(1 − 2)−N(2 )
and
=±−(ax +by +e− )/(ax +by +e+ )
M=bx −ay
N=ax +by
Fu he mo e, he in e sec ion poin s a e (Ax+Cx , Ay+Cy , 1) whe e
Ax=bM −ae, Cx=b 2−(N+e)2
Ay=−aM −be, Cy=−a 2−(N+e)2
No ice ha is eal i and only i − ≤ax +by +d≤ and ha each
diffe en alue o ep esen s a diffe en line.
No ice u he ha when in e sec ing a ci cle Cwi h a line L(d)=
[a, b, e−ad], he abo e heo em applies jus eplacing eby e−ad o eplacing
19
xby x−d. In his case, he ela ion be ween and dbecomes ad(1 − 2)=
(e− )(1 − 2)+N(1 + 2).
The hi d heo em ans o ms he in e sec ion o wo ci cles in o a line-
ci cle in e sec ion.
Theo em 6.4 Le Ci=(xi,y
i,
i)andCj=(xj,y
j,
j)be woci cles.
In e sec ing Ciand Cjis equi alen o in e sec ei he ci cle Cio Cjwi h
he line L=[a, b, e]whe e
a=xj−xi
b=yj−yi
e=1
2(x2
i+y2
i− 2
i−x2
j−y2
j+ 2
j)
The numbe o solu ions in each case can be easily calcula ed. Two lines
in e sec a mos in one poin and a line and a ci cle o wo ci cles in e sec
a mos in wo poin s. I he clus e fixed has wo cycles, we can ans o m
he in e sec ion o hei cyclog aphic maps in o he in e sec ion o one cone
wi h a plane, which has wo equa ions o deg ee one. This inc eases he
numbe o solu ions by a ac o o 2. The e o e, he e is one solu ion o
T11, wo solu ions o T21, T24, T41, T43, and ou solu ions o T31, T51,
T54 and T61.
6.3.2 In e sec ion o h ee planes and one su ace
In [6, 7], cone-cone in e sec ions a e ans o med in o cone-plane in e sec-
ions. Le Π(C1,C
i) deno e he plane ha con ains he in e sec ion cu es
o he cones γC1and γCio τC1and τCi. an conside , o example, he
subp oblem E(CC,CC). Fi s h ee planes, namely Π(C1,C
2), Π(C1,C
3)
and Π(C1,C
4) a e in e sec ed. Then his in e sec ion is subs i u ed in he
map γC1. We apply his app oach he e o find he in e sec ion o ou cones,
wi h he deg ee o eedom ha he hi d and ou h cones mo e along he
X-axis o o a e abou he o igin. We use he simila app oach o sol e he
p oblem in his c i e ion. No ice ha he in e sec ion o τC1and τC2can
be con e in o he in e sec ion o τC1wi h a plane τL.
Table 2 summa izes he p oblems we conside he e. The hi d, ou h
and fi h columns gi e he planes, gene a ed as γand τmaps, whose common
poin defines he dis ance d ha he mo ing clus e mus be ansla ed along
he X-axis. This common poin is hen subs i u ed in he equa ion gi en in
20
P oblem Equa ion Π1Π2Π3Deg ee
T12: L(LL’,LL’) γL1τL
2γL3(d)τL4(d) 1
T13: L(LL,LL’) γL1γL2γL3(d)τL4(d) 1
T22: L(CL’,LL’) γC1τL2γL3(d)τL4(d) 2
T23: L(C’L,LL’) γC
1γL2γL3(d)τL4(d) 2
T25: L(CL,LL’) γC1γL2γL3(d)τL4(d) 2
T26: L(CL’,LL) γC1τL2γL3(d)γL4(d) 2
T27: L(C’L,LL) τC
1γL2γL3(d)γL4(d) 2
T32: L(CL’,CL’) γC1τL2P(C1,C
3(d)) τL4(d) 4
T34: L(C’L,C’L) τC
1γL2P(C
1,C
3(d)) γL4(d) 4
T35: L(CL,CL’) γC1γL2P(C1,C
3(d)) τL4(d) 4
T44: L(CC,LL’) γC1P(C1,C
2)γL3(d)τL4(d) 2
T55: L(CC,CL’) γC1P(C1,C
2)P(C1,C
3(d)) τL4(d) 4
Table 2: Classifica ion o he h ee planes and one su ace in e sec ion p ob-
lem.
he second column, yielding an equa ion on he a iable dwhose deg ee is
shown in he las column. The specific algo i hm o sol e he p oblem can
be w i en as ollows.
Algo i hm L(E1E2,E
3E4)
1. Find he poin (x(d),y(d),z(d),w(d)) = Π1∩Π2∩Π3.
2. I ci cum(E1,C) hen
M:= γE1
else i cen e (E1,C) hen
M:= τE
1
endi
3. Gene a e one equa ion wi h one a iable dby eplacing
(x(d),y(d),z(d),w(d)) in he equa ion o M.
4. Sol e he sys em o equa ions o find a iable d.
5. The a iable adius ci cle seeked has (x(d)/w(d),y(d)/w(d))
as cen e poin and z(d)/w(d)as adius.
6.3.3 In e sec ion o one γ-cylinde one τ-cylinde and wo planes
The Table 3 summa izes he p oblems we conside in his sec ion. These
p oblems a e ans o med in o he in e sec ion o wo planes wi h a cone,
21

P oblem γC τCΠ1Π2Deg 1 Deg 2 Deg ee
T33: L(CL,CL)γC1τC
3(d)τL2γL4(d)(1,1,0) (0,0,0) 4
T36: L(CL,CL)γC1τC
3(d)γL2γL4(d)(1,1,0) (0,0,0) 4
T42: L(CC,LL
)γC1τC
2γL3(d)τL4(d)(1,0,0) (0,0,0) 4
T45: L(CC,LL)γC1τC
2γL3(d)γl4(d)(1,0,0) (0,0,0) 4
T52: L(CC,CL
)γC1τC
2dP(C1,C
3(d)) τ(L4(d)) (2,2,0) (0,0,1) 8
T53: L(CC,CL)γC1τC
2P(C
2,C
3(d)) γL4(d)(2,2,0) (0,1,1) 8
T56: L(CC,CL)γC1τC
3(d)P(C1,C
2)γL4(d)(1,1,0) (0,0,0) 4
T57: L(CC,CL)γC1τC
2P(C1,C
3(d)) τL4(d)(2,2,0) (0,1,1) 8
T62: L(CC,CC)γC1τC
2P(C1,C
3(d)) P(C
2,C
4(d)) (2,3,0) (0,1,1) 18(16)
T63: L(CC,CC)γC1τC
2P(C1,C
3(d)) P(C3,C
4)(d)(2,2,0) (0,1,1) 8
Table 3: Classifica ion o he γC,τCand wo planes in e sec ion p oblem.
γC, and a cylinde , τC. The planes Π1and Π2, a e gi en in he ou h and
fi h columns, he second column lis s he γmap and he hi d column he
τmap.
No e ha when he mo ing clus e is ansla ed along he X-axis, he
line common o planes Π1and Π2defines he dis ance d. Assume ha
Π1=(a1(d),b
1(d),c
1(d),w
1(d))
and
Π2=(a2(d),b
2(d),c
2(d),w
2(d))
a e wo planes in Table 3, he pa ame ic o m o he line whe e Π1and Π2
in e sec is
L=(a1(d),b
1(d),c
1(d),w
1(d)) + s(a2(d),b
2(d),c
2(d),w
2(d))
The deg ees on he a iable do each componen in (a1(d),b
1(d),c
1(d)) and
(a2(d),b
2(d),c
2(d)) a e shown in he six h and se en h column in Table 3.
The algo i hm o finding he solu ion o his class o p oblems is he ol-
lowing
Algo i hm L(CE2,E
3E4)
1. Acco ding o he subp oblem a hand and ollowing Table 3,
gene a e γC,τC,Π
1and Π2)
2. Find he pa ame ic o m o he line L=Π
1∩Π2.
3. Gene a e wo equa ions in dand sby subs i u ing he
explici o m o L in o he implici o ms o γC and τC.
22
4. Figu e ou dby sol ing he sys em o wo equa ions de i ed
in s ep 3.
5. Poin (a1(d),b1(d)) + s(a2(d),b2(d)) is he cen e o he
a iable adius ci cle and c1(d)+s∗c2(d) is he adius.
Applying Bezou ’s heo em o he equa ions de i ed in s ep 3 in he
Algo i hm L(CE2,E
3E4), we find he maximun numbe o solu ions o he
p oblem, which is gi en in he eigh column o Table 3. No ice ha compa ed
o he sys em o equa ions gene a ed in Sec ion 6.2, his app oach yields 2
mo e solu ions o he p oblem L(CC,CC).
6.4 Algo i hms o he Ro a ional Me ge P oblem
The classifica ion o o a ional clus e s is he same as in he ansla ional
case. All wha is needed is o eplace Eij wi h Rij in Table 1 o indica e
ha he p oblem has he same cons ain pa e n bu he mo ing clus e is
now o a ed. We also sepa a e hese p oblems in o h ee diffe en classes.
They a e:
1. P oblems wi h wo cen e cons ain s defined in he same clus e . This
includes R11, R21, R24, R31, R41, R43, R51, R54 and R61.
2. P oblems which can be ans o med in o he in e sec ion o h ee
planes and one su ace. This includes R12, R13, R22, R23, R25, R26,
R27, R32, R34, R35, R44, R55, and pa o he p oblems in he p e-
ious class.
3. The emainde o he p oblems, including R33, R36, R42, R45, R52,
R53, R56, R57, R62, R63.
6.4.1 Two cen e cons ain s in he same clus e
I he clus e wi h wo cen e cons ain s is he one ha will be o a ed, he
p oblems in his class a e o mally sol ed in he same way as L(E1E2,E
3E
4).
The in e sec ion o he geome ic elemen s in he clus e fixed can always
be ep esen ed by he in e sec ion o he map o he fi s elemen E1wi h
afixedplane. I θis he o a ion angle and = an(θ/2) he algo i hm o
sol e he p oblem is
23
Algo i hm C(E1E2,E
3E
4)
1. Find he poin (x, y)=τE
3∩τE
4.
2. Ro a e he poin (x, y)byanangleθyielding (x( ),y( )).
3. Replace (x( ),y( )) in o he equa ion o he plane gene a ed
in he fi s clus e and find z( ) which is a deg ee one
equa ion in .
4. Subs i u e (x( ),y( ),z( )) in o he equa ion o E1 o yield
one equa ion in .
5. Sol e he equa ion o find he alue o .
6. The seeked ci cle is cen e d on (x( ),y( )) and z( ) is he adius.
6.4.2 In e sec ion o h ee planes and one su ace
P oblems in his class, C(E1E2,E
3E4), a e sol ed as p oblems in he co e-
sponding ansla ional class L(E1E2,E
3E4) eplacing he ansla ion wi h
he o a ion αI = an(θ/2), he algo i hm is
Algo i hm C(E1E2,E
3E4)
1. Find he poin (x( ),y( ),z( ),w( )) = Π1∩Π2∩Π3.
2. I ci cum(E1,C) hen
M:= γE1
else i cen e (E1,C) hen
M:= τE
1
endi
3. Gene a e one equa ion wi h one a iable by eplacing
(x( ),y( ),z( ),w( )) in he equa ion o M.
4. Sol e he sys em o equa ions o find a iable .
5. The a iable adius ci cle seeked has (x( )/w( ),y( )/w( ))
as cen e poin and z( )/w( )as adius.
To jus i y he deg ee educ ion, we need he ollowing esul .
Theo em 6.5 Conside h ee planes wi h cons an coefficien s excep o :
Π1=[a2,b
2,c
2,−d2]
Π2=[a3(1 − 2)−b3(2 )+e3(1 + 2),a
3(2 )+b3(1 − 2)+ 3(1 + 2),
c3(1 + 2),−d3(1 + 2)]
Π3=[a4(1 − 2)−b4(2 )+e4(1 + 2),a
4(2 )+b4(1 − 2)+ 4(1 + 2),
c4(1 + 2),−d4(1 + 2)]
24
Then Π1,Π2and Π3in e sec in a poin (x( ),y( ),z( ),w( )) whose com-
ponen s a e exp essions o deg ee 2 in .
Now i is easy o show ha p oblems in his class ha e wo solu ions
when he fi s componen E1is a ay, such as in p oblems C12 and C13.
O he wise, hey ha e 4 soul ions, such as in p oblems C22, C23, C25, C26,
C27, C32, C34, C35, C44 and C55. The numbe o solu ions is gi en in he
Table 1.
6.4.3 Remainde P oblems
The emainde subp oblems o o a ional me ge p oblems a e C33, C36,
C42, C45, C52, C53, C56, C57,C62 and C63.
Un o una ely, he s a egy p esen ed in Sec ion 6.3 does no wo k he e
because he deg ee o he pa ame ic line gene a ed a an in e media e s ep
does no allow o educe he final deg ee o γC and τC.
We need o de i e some esul s be o e p esen ing ou algo i hm. Le C
be a cycle cen e ed a (x1,y
1)wi h(signed) adiusz1and le Land Lbe
ays wi h equa ions a2x+b2y+d2=0,anda2
2+b2
2= 1 espec i ely. Le
δ=(a2,b
2,d
2)·(x1,y
1,1) = (a2x1+b2y1+d2).
Theo em 6.6 The in e sec ion poin s o γC and γL a e gi en by
⎧
⎪
⎨
⎪
⎩
x(u)=x1+z11−u2
1+u2−δ
w(u)(1 −u2)
y(u)=y1+z12u
1+u2−δ
w(u)(2u)
z(u)=2z1−δ
w(u)(1 + u2)
No ice ha when z1= 0, he locus o he in e sec ion poin s in he abo e
heo em can be w i en in homogeneous o m as:
⎧
⎪
⎪
⎨
⎪
⎪
⎩
x(u)=x1w(u)−δ(1 −u2)
y(u)=y1w(u)−δ(2u)
z(u)=δ(1 + u2)
w(u)=(a2,b
2,−1) ·(1 −u2,2u, 1+u2)
Theo em 6.7 The in e sec ion poin s o γC and τLa e
⎧
⎪
⎪
⎨
⎪
⎪
⎩
x(u)=x1w(u)−δ(1 −u2)
y(u)=y1w(u)−δ(2u)
z(u)=z1w(u)−δ(1 + u2)
w(u)=(a2,b
2,0) ·(1 −u2,2u, 1+u2)
25
whe e Π = [x1−x0,y
1−y0,z
0−z1,(x2
0+y2
0−z2
0−x2
1−y2
1+z2
1)/2].
Theo em 8.12 The cyclog aphic map o he ay [a, b, d] o a ed abou he
o igin by θhas he o m
[a(1 − 2)−b(2 ),a(2 )+b(1 − 2),c(1 + 2),d(1 + 2)]
whe e = an(θ/2), −π<θ<π,andc=−√a2+b2.Whenθ=π, he
cyclog aphic map becomes [−a, −b, c, d].
Theo em 8.13 Le L=[a, b, d]bea ay.ThemapsγL(θ)andτL(θ)ha e
he o m
[a(1 − 2)−b(2 )+e(1 + 2),a(2 )+b(1 − 2)+ (1 + 2),c(1 + 2),d(1 + 2)]
whe e = an(θ/2), −π<θ<π,e= =0,andc=−√a2+b2,c=0
o he γL(θ)andτL(θ) espec i ely. When θ=π, heγL(θ)andτL(θ)
becomes [−a, −b, c, d].
Theo em 8.14 Le C
1=(x1,y
1,z
1)andC
3=(x3,y
3,z
3). The in e sec ion
plane o τC
1and τC
3(θ) has he o m
[a(1 − 2)−b(2 )+e(1 + 2),a(2 )+b(1 − 2)+ (1 + 2),c(1 + 2),d(1 + 2)]
whe e = an(θ/2), −π<θ<π,a=x3,b =y3,e =−x1, =−y1,c =0,
and d=(x2
1+y2
1−z2
1−x2
3−y2
3+z2
3)/2. When θ=π, he cyclog aphic map
becomes [−a+e, −b+ ,0,d].
32