Professional Documents
Culture Documents
Tn the autumn of 1948, while experi- Snelson's sculptures, in which rigid graphic and computational capabilitie
J.menting with ways to build flexible, sticks or "compression members" (as an of computeB, have now made it possi
moduUrrtoweB,ayounga~tnwmed engineer might call them) are suspend- ble to draw up a complete catalogue 0
Kenneth Snelson constructed a sort of ed in midair by almost invisible cables tensegrities with certain prescribec
sculpture that had never been seen be- or very thin wires, can still be seen types of stability and symmetry, in
fore. As ethereal in appearance as a mer around the world. A remarkable, 60- cluding some that have never beel
bile, with no obvious weight-bearing el- foot-high sculpture, ''Needle Tower," is seen before.
ements, it nonetheless retained its shape displayed at the Hirshhorn Museum
and stability. '1 was quite amazed at and Sculpture Garden in Washington, What Is a Tensegrity?
what Ihad done," Snelson recalled four D.C. The idea has penetrated into low Tensegrities have a purity and simpIic
decades later. The following sununer he art as well. A number of baby toys em- ity that lead very naturally to a mathe
showed the sculpture to his mentor, the ploy the swme principles as Snelson's matical description. Putting aside th.
not-yet-fwmous inventor, ~t and self- original tensegrities. One could even ar- physical details of the construction
styled mathematician R Buckminster gue that the first tensegrities were not every tensegrity can be modeled math
Fuller. Before long, Fuller had adapted made by human beings: A spider web ematically as a configuration of points
Snelson's invention as a centerpiece of can also be viewed as a tensegrity, al- or vertices, satisfying simple distano
his system of synergetics, even to the beit one with no rigid parts. constraints. Snelson's structures afl
point -of calling the neW objects "my Although Fuller's geodesic domes held together with two types of desigi
structures" and promoting them in his and synergetics gained him worldwide elements (engineers say members)
many inspirational, free-ranging lec- renown, most of the mathematics that which can be called cables and struts
tures. In the process, he gave them the he used was already well established. The two elements play complementar:
name by which they are known today, However, his student Snelson's discov- roles: Cables keep vertices close to
referring to their integrity under ten- ery posed genuinely new mathematical gether; struts hold them apart. Twc
sion: tensegrity. questions, which are far from being vertices connected by a cable may b.
completely resolved: What is a tenseg- as close together as desired-the~
Robert Connelly is professor and chair in the rity? Why is it stable? Can tensegrities might even be on top of one another i
Department of MathemDlics at Cornell University. be classified or listed? the tensegrity collapsed-but they ma~
He received a Ph.D. in mathematics from the BrankoGriinbaum, a mathematician never be farther apart than the lengtl
University of Michigan at Ann Arbor. He IuJs vis- at the University of Washington in of the cable joining them. Similarl}
ited the Institut des JUutes Etudes Scientifique in Seattle, was especially responsible for two vertices joined by a strut may nev
Bures-sur- Yvette, the University of Dijon and the rekindling the interest of mathemati- er be closer than the length of the strul
University of Chambery in France, Syracuse but may be arbitrarily far apart.
cians in such questions, with a wonder-
University, the University of Montreal, E6tviJs
University in Budapest and Bielefeld University in ful set of mimeographed notes written The last point may seem surprisinl
Bielefeld, Germany. His research interests include in the early 1970s, called ''Lectures on at fiBt, because in most real tensegri
discrete geometry, particulDrly the study of rigid Lost Mathematics." In 1980, one of us ties the struts cannot get either longe
and flexible frameworks, distance geometry, pack- (Connelly) proved a conjecture of or shorter. In fact, the tenn "bar" ha
ings and coverings, and the shapes of asteroids. Griinbaum's that allows the systematic been used to describe a design elemen
Allen Back is Director of the Matll Department construction of stable planar tensegri- of fixed length. However, we havi
Instructional Computing lAb at Cornell. He ties. But the wonder and beauty of found that, most of the time, bars cat
received his Ph.D. in mathematics from tire Snelson's sculptures surely lies in their be replaced by struts without sacrific
University of California, Berkeley. His interests three-dimensional nature. One of the ing stability. Moreover, the concept 0
iflclllde differential geometry, topology, solid mod- struts can be applied to other prob
motivations of our recent work, there-
eling, robotics, dyruzmical systems and the produc-
fore, was to find a proper three-dimen- lerns, such as the packing of spherica
tion of mathemDtical software. Address for
Connelly: Department of Mathematics, White sional generalization. The mathematical balls. (In any such packing, the center
Hall, Cornell University, ItIUZCiJ, NY 14853. tools of group theory and representa- of the balls must keep a minimum dis
Iflternet: connelly@math.comell.edu. tion theory, coupled with the powerful tance but can be as far apart as desired
;'i .!I\,i;~;;;:•·•
physicist Robert Hooke (1635-1703)
.D~¥~:~>Fi''''.'·
'.i .. ' f.".'~;";" .....5;...•
,<..;,'., " .J:,!:.•.;,.,,~V,:(X,'k,.X2}" ~ (y '1· ··.,y,2}..+ ;~~~.:.«w ,(z.l·· . l: W. 2) '.'
found that the force needed to displaCE
a spring was proportional to the dis-
..~--:.-.
:.,~..~'.'\:~::~.'~ <~./..":.-:> .'t,"""',~:
:.:~.':'>:~'.,~~ '."':..',.' :-..,~,; >:',: ,:,,:.~~.;~ '~~~.~,~~.~.~,•.~;.~::. :~'>
> ~-,',:~.::,.
...; :~,:;' :::;t:
I',· • ,'. -,' . '-.
placement from its rest position. (HE
wrote this empirical observation, late!
Figure 3. Points in n-dimensional Euclidean space can be identified with their coordinates in
known as Hooke's law, as an anagram
a Cartesian coordinate system. The number of coordinates is the dimension. Distance in n-
dimensional space is calculated by analogy with the Pythagorean fOIDlula from plane geom-
"ceiinosssttuu." The unscrambled ana-
etry. A tensegrity can be defined in any dimension because its points and the distance con- gram-"Ut tensio, sic vis"-translat~
straints governing the struts and cables can aU be stated in tenns of coordinates. Here exam- from Latin as, "As the extension, so i!
ples are shown for 1- through 4-dimensional space. the force.") Although Hooke did nol
trengths in such a way that the config- one minimum point. If a spider web sat-
:ration represents a unique minimum isfies the equilibrium condition, it has to Figure 4. Mathematical spider web consists
)r the corresponding energy function, be the one. Thus the principle of least only of cables, with no struts. Thn!e vertices
:te sum of the energy functions of all are pinned (slladed rectangles) and cannot
work applies to show that the spider
move relative to the background. In order for
:te cables. Then any comparable spider web is super stable.
the web to be super stable, the three cables
,reb that does not increase any cable Note that this method does not just coming from the pinned vertices must deter-
?ngth must have the same energy or provide a ''local'' result only valid for mine lines that go through a single point.
maller, because each cable contributes small (or even bounded) perturbations.
:te swme, or less, to the total. But since
:te given configuration is supposed to
epresent a unique minimum energy,
:te two configurations must be identi-
al. Consequently, such a spider web y y
)ne that minimizes some energy func-
.on) is super stable. This is called the
rinciple of lenst work, in honor of the sim-
ar principle that is used in structural
x x
ngineering.
How does one recognize when a spi- a
.er web has an energy function that is
lrecisely minimized by the given con-
.guration? One answer is the equilibri-
m of stresses-another concept bor-
)wed from engineering. Again, it helps
) imagine the cables that meet at any
iven vertex as springs, each one tug-
ing in a different direction. Remember
:tat this force can be made "stronger"
r "weaker" by a proportionality con-
tant, which we will call the stress in the
able; thus the strength of each cable's
19 is equal to its length times its stress.
b
:the stresses are picked in just the right
lay, so that the tug-of-war wmong all
:te cables leading to a given vertex is a
raw, then the stresses at that vertex are
1 equilibrium. If that happens at every
npinned vertex, then the whole spider y
leb is said to be in equilibrium. y
These stresses are simply numbers
:tat are assigned to cables; they need
.ot have anything to do with the size of
~e cable or even its physical or me- x
hanical characteristics. However, it is
nportant that all the stresses are posi- c
ive: If some are zero, then the given Figure S. Affine transformations include stretches (a), flips (b) and projections (c'. Note that a
onfiguration may not be a unique min- projection may cause many struts and cables to overlap. AU of these transformations have
num of the energy function, and the the property that an eqUilibrium stress for the original tensegrity also serves as an equilibri-
.rinciple of least work will not apply. um stress for the transformed tensegrity.
-..
··· i
!
each of the cables and struts.
Incidentally, although the energy func-
.................. .......
........
.' ...
~ tion just described is not physically real·
.'.'
istic, it is not completely divorced from
..........
.... .....~ -(
... engineering reality. If one analyzes the
local static properties of a structure, the
quadratic energy described here is one
Figure 6. Super stable planar tensegrities can be generated by a 1980 Iheorem of Connelly. If of two terms that enter into a descriptiot:\
the cables form a strictly convex polygon, If the struts are inlemal diagonals, and If Ihere Is a of the structure under sufficiently sman
positive stress for each cable, a negative stress for each sinal and eqUilibrium al each vertex, perturbations. When the stability can be
then the lensegrlty Is super slable. The stabUlty Is not always intuitively obvious. A configu· detected by such a quadratic approxima-
ration inspired by a rigidity theorem proved by French mathematician Augustin Louis tion, the structure is called prestress stable
Cauchy in 1813 is shown at top left. An example of a dass suggested by the Branko in the engineering literature.
GrIlnbaum of the University of Washington is shown at top right. A configuration that is If we consider only small perturba-'
super stable only If the vertices lie on an ellipse appears at boHom left. lions of the physical energy of a pre-
stress stable structure, the second term
(which we have not described) only
adds to the stability of the structure. In·
deed, for a super stable tensegrity, as
long as there is no catastrophic buckling
of the struts or breaking of the cables, in·
creasing the stress tends to stabilize the
tensegrity. This is not necessarily the
case for a tensegrity that is only pre-
stress stable.
Figure 7. Super stable three·dimensional
tensegrlties can be generated from the action
of a symmetry group on one strut and two
cables. In Ihis example, which has the sym·
metry group of half the symmetries of a
cube, each strut (green) can be superimposed
on each other one by a rotation or a reflec-
tion. Similarly, each red cable can be super-
imposed on each other red cable, and each
blue cable on each other blue cable. The
whole tensegrity can be seen as being made
up of six Identical "stretchers" joined
together at their ends with the red cables. If
the predetermined ratio of the stress in the
blue cables to the stress in the red cables is
increased, the blue cables shorten, and the
configuration approaches that of the baby'
toy tensegrlty.
century by F. Georg Frobenius and Isai in the same orbit if at least one symme- congruences is defined by first perfonn-
Shur, was originally motivated by a prob- try of the tensegrity superimposes one ing one motion and then the other. i
lem in algebra, but soon found a wide va- of the pair onto the other. In Figure 7, Group theory is something like the
riety of applications, especially to the each cable or strut in a given orbit has ancient Oriental game of go. There are
physics of the then new theory of the the same color. The grouP. of symme- only a few simple rules, and they are
atom. A typical success story begins with tries of these tensegrities is transitive easy to learn. But their consequences
a fairly complicated mathematical model on the struts (because there is only one can take many yeaB of intense study to
of a structure with some form of (ge0- orbit, colored green) but not quite tran- master. Like certain move sequences (or I
metric) symmetry. Representation theory sitive on the cables (because there are "joseki") in go, certain groups occur of-.
allows one to break down the complicat- two orbits, colored red and blue). ten enough to have their own names.:
ed model into a predetermined, small With this in mind, it is especially easy The simplest, and perhaps most ubiq-
number of much more tractable models, to check whether there is an equilibrium uitous, is a group with two elements i
each of which can be treated more or less stress for the cables and struts. If there is called ~. Its elements can be thought of:
independently. Fan Chung and Shlomo any energy function at all that is mini- as the numbeB +1 and -1, with the op-'
Sternberg (1993) gave a very nice exam- mized by the given configuration, then eration of multiplication. (Note that ai
ple of such an application: the analysis of there is one that has the swme symmeby product of either +1 or-1 with either +11
the infrared spectra of the buckyball, a as the tensegrity itself. Thus the same or -1 again gives +1 or -1.) Or they can·;
molecule with 60 carbon atoms that has stress can be assigned to each cable or be thought of as the words "even" and i
the symmetry of a regular dodecahedron. strut in any given orbit. Similarly, the bal- "odd," with the operation of addition. I
In our situation, the representation theory ance of stresses only needs to be checked Or they can be thought of as the identity i
is called on in almost the swme way, ex- at one vertex, because all vertices are motion and reflection in a mirror.
cept that the underlying mathematical alike under the group of symmetries. For Of the many guises an abstract group!
objects are different. almost all the exwmples considered here, can assume, among the most conve-!
Suppose that a tensegrity can be r0- it turns out to be easy to check that the nient ones are sets of linear functions.]
tated about some line in such a way that only affine motions that preserve the ca- Rotations and reflections are examples:
the rotated tensegrity is indistinguish- ble and strut constraints are congruences. of linear functions. H the elements of thej
able from the original: each vertex su- Hence only one difficulty remains before abstract group are thought of as actors
perimposed on a vertex, each cable su- we can apply the least work principle: and the linear functions as roles, then·
perimposed on a cable and each strut We have to make sure that the energy is the playbill, which assigns certain actors
superimposed on a strut. All the mo- a minimum. That is where group repre- to certain roles, is called a representation:
tions, or congruences, that superimpose sentations enter the picture. of the group. The abstract group itself
the tensegrity on itself in this fashion can be used to do group calculations
form a mathematical structure called Group Representations and effectively provide a common point'
the group of symmetries of the tensegrity. So far, groups have appeared in only one of reference. A representation, on the
(It may include some reflections and guise: the group of symmetries of a other hand, may have more structure,'
other congruences, and always includes tensegrity. But groups can also be de- and may give deep and subtle infonna-,
the "identity," a Zen-like motion that fined in the abstract, without reference tion about the abstract group. 11
simply leaves everything untouched.) to any particular physical object. From An essential concept in dealing with
If a tensegrity has enough symme- this point of view, a group is simply a set representations is that of eqllivalpHce•1
tries that any vertex can be superim- whose elements can be "multiplied" and Consider, Corexwmple, the group of
posed onto any other by a congruence, that obeys certain rules, such as the exis- metries of the word MOM. There are
then the group of symmetries is said to tence of an identity element. In the case ways to superimpose this word on
be transitive on the vertices. Similarly, just presented, the elements of the group Either leave it alone (the identity motioNI
any pair of vertices (or any pair of ca- were motions that superimpose a tenseg- or reflect it about a vertical line
bles or any pair of struts) is said to be rity on itself, and "multiplication" of two the center of the O. This
for four dimensions; and the nontrivi,ll Suppose that we have a tensegrity tion), corresponds to the regular
and trivial1-dimensional representations whose group of symmetries acts transi- sentation of the symmetry group.
account for one more dimension each. In tively on the vertices. Its energy function, was not at all obvious to us at first:
total, we get 9 + 9 + 4 + 1 + 1 = 24 dimen- which was defined using the stresses after doing several special cases did
sions accounted for. (motivated from the equilibrium condi- discover that Frobenius's theorem