13
Pacific Journal of Mathematics ON l -SIMPLICIAL CONVEXITY IN VECTOR SPACES TUDOR ZAMFIRESCU Vol. 22, No. 3 March 1967

1BDJGJD +PVSOBM PG .BUIFNBUJDT

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

Pacific Journal ofMathematics

ON l-SIMPLICIAL CONVEXITY IN VECTOR SPACES

TUDOR ZAMFIRESCU

Vol. 22, No. 3 March 1967

Page 2: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

PACIFIC JOURNAL OF MATHEMATICS

Vol. 22, No. 3, 1967

ON /-SIMPLICIAL CONVEXITY IN VECTOR SPACES

TUDOR ZAMFIRESCU

The paper is concerned with a generalized type of con-vexity, which is called /-simplicial convexity. The name isderived from the simplex with I vertices, an Z-simplicialconvex set being the union of all (ΐ — l)-simplexes withvertices in another set, i varying between 1 and I. The basicspace is a linear one.

For convex sets the Z-order (which is a natural numberassociated to an ί-simplicial convex set) is a decreasing func-tion of I. Several inequalities between I- and ά-orders areestablished. In doing this the case of a convex set and thatof a non convex set are distinguished.

Besides these inequalities, the main result of the paper isthe proof of non monotonicity of the border, given by anexample in a 34-dimensional linear space.

Let us consider a real vector space and recall some notations anddefinitions, given in [2].

The convex cover (hull) of a set M is gf(M). We denote&({xu , xs}) by &(xlf , x8), xlf , xs being not necessarily distinctpoints in the space.

The operation £^, called Z-simplicial convex cover, and defined by

Sφί) = {U if (^, . . . , ^ ) : f t G J ί , l ^ i ^ l }

(for arbitrary set M and natural number I ^ 2) will play a significantrole throughout the paper. The operation S^ is studied in [1], whereis denoted by conz. It is easy to verify the following elementaryproperties of this operation:

(l) ^ ς Λ = ^ ς o ^ ; ,

(2) £smP = ^ 1 , and

(3) &>m(M)a&(M),for arbitrary m, n, p ^ 2 and M. S^fM) is an increasing1 functionof I. It is also an increasing function of M with respect to the in-clusion ordering. Let us denote S^m({xu , xs}) by S^m{xu , xs).

A set K is said to be Isίmplicίal convex if there exists a subsetM such that K = SφΛ).

The l-order (l-simplίcial convexity order) of an i-simplicial con-vex set K is

ωt(K) = supmin {k: Sf\(M) = K} .M

1 By increasing and decreasing functions we mean here not necessarily strictlyincreasing and strictly decreasing ones.

565

Page 3: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

566 TUDOR ZAMFIRESCU

A set K is said to be simplicial convex if there exists a numberI such that K is ϊ-simplicial convex.

The degree of a simplicial convex set K is

δ(K) = min {I: K is i-simplicial convex} .

The order {simplicial convexity order) of a simplicial convex setK is

= sup ωι(K) .I

The power of a simplicial convex set K of finite order is

Δ(K) = min {I: O(K) = ω,(iί)} .

It is proved in § 1 of both [1] and [2] that

gf (M) = <9ϊίl08ιnl+1(M)

in an π-dimensional vector space.

2* Relations between Z-simplicial and Zosimplicial convexityorders* In this section some inequalities concerning ί-order and k-order of a same set will be established.

THEOREM 1. The l-order and k-order of a convex set K satisfythe inequalities'.

[log, (ZV- 1 + 1)] ωk(K) [log, (W*> - 1)1 + 1 .

Proof. Consider a set M such that

Sφί) - K .

If k< l\

K - SftM) c ^

S i n c e i f i s c o n v e x t h e i n v e r s e i n c l u s i o n h o l d s t o o a n d K = S ^ ( )L e t x e K . S i n c e

^ωι{K)(M) = K ,

x belongs to a simplex with vertices xl9 , xs e M(s ^ W^)- Thereexists a linear manifold of dimension lωι{K) — 1 containing g*(ίc,, , a?β).Following the last remark of §1,

%>(xl9 . . . , » . ) = ^ ^ * ( I d > i ( Λ : > - 1 ) ^ 1 ( a ? 1 > , a?.) c

h e n c e

Page 4: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

ON Z-SIMPLICIAL CONVEXITY IN VECTOR SPACES 567

The inverse inclusion holds owing to the convexity of K, whence

ω{K) = K .

Thus

ωk(K) £ [log, (ί«ι<*> - 1)] + 1 .

The symmetry in I and k gives

ω^K) ^ [log, (£«*<*> - 1)] + 1 ^ log, (£«*<*> - 1) + 1 ,

i.e.

It follows that

[log, (W^- 1 + 1)] ^ log, (IV*)-1 + 1) ^ ωΛ(X)

and both inequalities are obtained.

In [2], we have established that in general, λ -simplicial convexitydoes not imply ϊ-simplicial convexity for I < k. However, if k is apower of I this implication holds.

THEOREM 2. For all natural numbers k, I, q Ξ> 2 satisfyingk = lq, non convexity and k-simplicial convexity imply l-simplicialconvexity; also the k-simplicial and l-simplicial convexity orders verifythe inequalities

qωk ^ ωι ίg qωk + q — 1 .

Proof. Let M be such that

= C ,

where C is a non convex, fe-simplicial convex set. Then

whence C is Z-simplicial convex and

To prove the other inequality, suppose that there is a set Mr

such that

^ m ( M ' ) = C

with

Page 5: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

568 TUDOR ZAMFIRESCU

w& ^ Qθ)k(C) + q .

Then, either m — qo)k(C) + g and

or m > qωk(C) + g and

both impossible. Following Theorem 5 of [2],

ω,(C) = sup {m: ^m(M) = C} .M

Hence

ω,(C) ^ gωΛ(C) + q - 1 ,

which concludes the theorem.A different inequality is obtained, if k = i9, for a convex set.

We have, indeed, by Theorem 1, ωk ^ log* (iωί — 1) + 1, i.e.

and ωk ^ logfc (ίωr 2 + 1), i.e.

k^ ^ ί ' r 1 + 1 .

Hence

whence

q(ωk - 1) < ωι

and

iffω* ^ lωrι + 1 ,

whence

qωk > ωι - 1 .

Therefore

These inequalities and those of Theorem 2 show that, for arbi-trary sets, A -simplicial convexity implies Z-simplicial convexity (fork = lq) and

qωk — q + 1 <; ωt ^ qωk + q - 1 .

Page 6: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

ON Z-SIMPLICIAL CONVEXITY IN VECTOR SPACES 569

3* M o n o t o n i c i t y of o)t(K) for c o n v e x K. I t is proved in § 4

of [2] t h a t ωk ^ ωι if k is a multiple of I. Moreover, we shall prove

t h a t for convex K, ωk(K) ^ ωx(K) if k ^ I.

THEOREM 3. For a convex set K, the l-simplicίal convexity orderis decreasing on I and Ω(K) = ω2(K).

Proof. Prove that ω^K) is a decreasing function of I. Suppose,on the contrary, that i > j and

ωά{K) .

It follows that

ωt(K) - 1 ^ ωά{K)

and

But, from Theorem 1,

ωάK) ^ log, (j"J^ - 1) + 1 ,

whence

/j/ωi(K)-ί < jo)j(K) _ ]_ ^ jo)j(K)

The contradiction shows that ωz(if) is decreasing of I, for convexK. Since iΓ is 2-simplicial convex,

Ω(K) - ωδ(iC)(iί) = ω2(K) .

4* Non monotonicity of o)ι% It may be conjectured that ωt isin general a decreasing function of i, i.e. ωz(C) is also decreasingfor non convex C. Then the inequality of Theorem 11 would betrivially implied by Theorem 6, both of [2], and Ω and δ would equalrespectively ωs and A.

On the other hand one can believe that Theorem 2 can be obtainedfrom two more general inequalities, for non convex sets, like

(*) [log, (lω* + 1 ) 1 - 1 ^ 0 ) * ^ [log, (1-1+1 - 1)]

in the same way as, for convex sets,

qωk — q + 1 ^ ωι <; qωk (k = lq)

is implied by Theorem 1.It should be noted that each inequality of (*) would imply that

ωι is decreasing. If, on the contrary i < j and ω{ < ω3Ί then

Page 7: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

570 TUDOR ZAMFIRESCU

which contradicts the two inequalities

and

ωό < [log, (i^1 - 1)]

Also,

iωi+1 < jωj

contradicts the two inequalities

jωj + l <; ίωi+1

and

The conjecture that o)ι is, in general, a decreasing function ofI (and with it also relation (*)) is disproved by the following counter-example.

PROPOSITION. Let 7 be a 34-dimensional real vector space andC be the union of the subsimplexes with 18 vertices of a simplex with35 vertices in 5Γ Then C is both 2-simplicial convex and 3-simplicialconvex, ω2(C) = 1 and ω3(C) = 2. Also Ω(C) = 2 and Δ{C) = 3.

Note. This proposition and its proof, are the simplest that wecould find to provide our counter-example. We ask for simpler ones.In fact, we have found such a simpler example in a vector space ofsmaller dimension, but the proof was much more complicated. How-ever it should be interesting to find the smallest dimension of thespace in which such an example can be found, even if its proof isdifficult.

Proof. Let S)(i = 1, , (f)) be the subsimplexes with j vertices(j = 2, , 34) of the given simplex S. Thus

c = u sι8.i

(1) The 2-simplicial and 3-simplicial convexity of C follow from

c =

(2) Prove that ω2(C) = 1. Suppose that 6^\M) = C.(a) First, we shall prove that

Page 8: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

ON Z-SIMPLICIAL CONVEXITY IN VECTOR SPACES 571

Suppose there exists

then y belongs to the interior of a simplex S)a with 10 ^ j ^ 18.(a) If 10 ^ i ^ 17, let Sjj-i be the simplex with 35 - j vertices

disjoint from S)a. If <Pζ(M) contains also a point z in the interiorof a subsimplex S]-j of S^-y, with ί ^ 19, then &(y> z) intersectsthe interior of gf (S*α U S\ί.ά), which is impossible; hence

ΓΊ Silj c {U SlLj: S ^ - c Sjt,} c { u S j : S 5 c Sβ_,.}

for 10 ^ j ^ 16 or

^t(M) n sii c {a?lf..., xw] n s;8

b c {u si: s^ c si8

δ}

and if

o 1 8 L_ o35_j ,

then

Sζ(M) ΓΊ Sί8

rf c {U Si: S i c Stlj} Π Sί ί - { U S j S j c Sit} .

It follows t h a t

^2\M)n s; 8

d c{us; 6 :s j 6 cs; 8

d } ^ s;8

d,

whence

sit,

absurd.(b) If j — 18, then the segment joining y with a vertex &% of S

that does not belong to Slξ, meets the interior of &{S[t U { s}), absurdagain.

(β) Now, prove that

M(£\JSΪ.

Suppose that M e U* S{. Then

whence

impossible. Hence there exists x e M — (J;

Page 9: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

572 TUDOR ZAMFIRESCU

(a) If x is an interior point of a simplex S'f with 5 ^ j <£ 8,then, following (a), M does not intersect the interior of any subsimplexS)f(l ^ 5) of

Hence, if Sίί c SU-j, then

n Sίί c J5f { U SI: S\ c Sjί} = {U S;6: S;6 c

whence Si2(M) ;z3 SI*, absurd.(b) If x is an interior point of the simplex Slk then, for

xreSlk, &(x,xr) intersects the interior of ί£(S\k U {xr}), absurd.Combining (a) and (β) we obtain that M(t\JiS\ and Si{M)a\Ji Si

contradicting one other. Therefore any Ma y such that S^2(M) — Cdoes not exist. Also, £ik(M) Φ C for k ^ 3, because £>i\s<ik-\M)) Φ C.Hence ω2(C) = 1.

(3) Prove that ω3(C) = 2. Suppose there exist a subset I c T "and a number & 3 such that S^k(M) = C Then M must containthe vertices xlf , x3δ of S. But

=) ?3(M) 3 £%\xu , α;35) -

whence Sik(M) Φ C, absurd. Because

ω,(C) = 2.(4) Prove that Ω(C) = 2 and J(C) = 3. If C would be 4-simplicial

convex, then 2α>4(C) 1, following Theorem 2, which is not possible.If C is i-simplicial convex, with I ^ 5, then suppose that M and k ^ 2satisfy 5>f(M) - C. Then

^f2(ikf) => ^(a?!, , x35) = U SU ,

absurd. Hence ω,(C) = 1.Thus β(C) = ω3(C) = 2 and Δ(C) = 3. The proof is complete.

5. Concluding remarks. Most of the results of [2] and of thepresent paper are purely algebraic and hence valid in vector spacesover arbitrary ordered fields.

Besides the operation f, the function ωι whose study was begunin § 2 of [2] and continued here is of special interest in the investiga-tion of ί-simplicial convexity. In §§ 3-4 of [2] we established someelementary facts concerning degree, order, and power. In order toinitiate a systematic study of simplicial convexity more informationabout these three functions should be obtained.

Page 10: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

ON Z-SIMPLICIAL CONVEXITY IN VECTOR SPACES 573

REFERENCES

1. W. Bonnice and V. L. Klee, The generation of convex hulls, Math. Annalen 152(1963), 1-29.2. T. Zamίirescu, Simplicial convexity in vector spaces, Bull. Math. Soc. Sci. Math.R.S.R. 9 (1965), 137-149.

Received March 7, 1966.

INSTITUTE OF MATHEMATICS, BUCHAREST

Page 11: 1BDJGJD +PVSOBM PG .BUIFNBUJDT
Page 12: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

PACIFIC JOURNAL OF MATHEMATICS

EDITORSH. SAMELSON

Stanford University-Stanford, California

J. P. JANS

University of WashingtonSeattle, Washington 98105

J. DUGUNDJI

University of Southern CaliforniaLos Angeles, California 90007

RICHARD ARENS

University of CaliforniaLos Angeles, California 90024

ASSOCIATE EDITORSE. F. BECKENBACH B. H. NEUMANN F. WOLF K. YOSIDA

SUPPORTING INSTITUTIONS

UNIVERSITY OF BRITISH COLUMBIACALIFORNIA INSTITUTE OF TECHNOLOGYUNIVERSITY OF CALIFORNIAMONTANA STATE UNIVERSITYUNIVERSITY OF NEVADANEW MEXICO STATE UNIVERSITYOREGON STATE UNIVERSITYUNIVERSITY OF OREGONOSAKA UNIVERSITYUNIVERSITY OF SOUTHERN CALIFORNIA

STANFORD UNIVERSITYUNIVERSITY OF TOKYOUNIVERSITY OF UTAHWASHINGTON STATE UNIVERSITYUNIVERSITY OF WASHINGTON

* * *AMERICAN MATHEMATICAL SOCIETYCHEVRON RESEARCH CORPORATIONTRW SYSTEMSNAVAL ORDNANCE TEST STATION

Mathematical papers intended for publication in the Pacific Journal of Mathematics should betypewritten (double spaced). The first paragraph or two must be capable of being used separatelyas a synopsis of the entire paper. It should not contain references to the bibliography. Manu-scripts may be sent to any one of the four editors. All other communications to the editors shouldbe addressed to the managing editor, Richard Arens at the University of California, Los Angeles,California 90024.

50 reprints per author of each article are furnished free of charge; additional copies may beobtained at cost in multiples of 50.

The Pacific Journal of Mathematics is published monthly. Effective with Volume 16 the priceper volume (3 numbers) is $8.00; single issues, $3.00. Special price for current issues to individualfaculty members of supporting institutions and to individual members of the American MathematicalSociety: $4.00 per volume; single issues $1.50. Back numbers are available.

Subscriptions, orders for back numbers, and changes of address should be sent to Pacific Journalof Mathematics, 103 Highland Boulevard, Berkeley 8, California.

Printed at Kokusai Bunken Insatsusha (International Academic Printing Co., Ltd.), 7-17, Fujimi2-chome, Chiyoda-ku, Tokyo, Japan.

PUBLISHED BY PACIFIC JOURNAL OF MATHEMATICS, A NON-PROFIT CORPORATIONThe Supporting Institutions listed above contribute to the cost of publication of this Journal,

but they are not owners or publishers and have no responsibility for its content or policies.

Page 13: 1BDJGJD +PVSOBM PG .BUIFNBUJDT

Pacific Journal of MathematicsVol. 22, No. 3 March, 1967

Wai-Mee Ching and James Sai-Wing Wong, Multipliers and H∗

algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387P. H. Doyle, III and John Gilbert Hocking, A generalization of the Wilder

arcs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397Irving Leonard Glicksberg, A Phragmén-Lindelöf theorem for function

algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401E. M. Horadam, A sum of a certain divisor function for arithmetical

semi-groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407V. Istratescu, On some hyponormal operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413Harold H. Johnson, The non-invariance of hyperbolicity in partial

differential equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419Daniel Paul Maki, On constructing distribution functions: A bounded

denumerable spectrum with n limit points . . . . . . . . . . . . . . . . . . . . . . . . . . . 431Ronald John Nunke, On the structure of Tor. II . . . . . . . . . . . . . . . . . . . . . . . . . . 453T. V. Panchapagesan, Unitary operators in Banach spaces . . . . . . . . . . . . . . . . 465Gerald H. Ryder, Boundary value problems for a class of nonlinear

differential equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477Stephen Simons, The iterated limit condition and sequential

convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505Larry Eugene Snyder, Stolz angle convergence in metric spaces . . . . . . . . . . . 515Sherman K. Stein, Factoring by subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523Ponnaluri Suryanarayana, The higher order differentiability of solutions of

abstract evolution equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543Leroy J. Warren and Henry Gilbert Bray, On the square-freeness of Fermat

and Mersenne numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563Tudor Zamfirescu, On l-simplicial convexity in vector spaces . . . . . . . . . . . . . . 565Eduardo H. Zarantonello, The closure of the numerical range contains the

spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575

PacificJournalofM

athematics

1967Vol.22,N

o.3