* , is . Ai ABSTRACT OF THE THESIS OF CORNELIUS HENRY 74110 Date. thesis is presented for the MASTER of SCIENCE in MATHEMATICS _. Apl 8, 1963 ,- Title ON THE STRUCTURE OF MINIMUM SURFACES AT THE BOUNDARY Abstract approved - 1 '' In this paper we consider the behavior of certain surfaces at certain boundary points. The surfaces under consideration satisfy a topological definition and are of 2- dimension in 3- dimensional Euclidean space with the boundary a finite set of straight line It is shown that the surface of minimum area with a given segments. boundary is locally Euclidean at all non-vertex. boundary points. The key to the proof is a theorem in 1 which itself concerns the behavior of a set of points under very restricted conditions. It is shown in 1 that for almost all interior points the conditions of the lemma are satisfied. This paper first shows that the part of the given surface interior to some sphere centered at any non-vertex boundary point lies near a plane passing through the point in question. Secondly it is shown that for any point of the surface lying in the sphere the surface interior to any smaller neighborhood of that point lies near a plane. "Near" here refers to nearness with respect to the radius of the sphere or neighborhood in consideration. - From these conditions we may construct a bounded point set satisfying the hypothesis of the aforementioned theorem. The theorem of this paper follows immediately. a S ì1t i6 ^ ' " !. 41 ON THE STRUCTURE OF .0 MINIMUM SURFACES AT THE BOUNDARY by CORNELIUS HENRY TJORIUR 1,Li '.r6l,¡r.iA;a5..7 : ,.1, 1 .. o') ; A THESIS submitted 0 to I)' :, :r i5'1 N STATE UNIVERSITY j ', ; ^'N ('tri June 1963 , _ . . in partial fulfillment of the requirements for the degree of STIR OF SCIENCE e ,'íy..R1 " , . 1: f ." 1 . tr I'4' r . esd of Qaxint and in Ctin °f Sc1i° an of )se Ct Cts $PIIh1!H áf liaioy. Graduate ccrsitss Graduate ScbpOt si.s 3. 4tes Zy4ea by S ea soä SI pr il 8' 1963 's AC1CNOWLSDG!021T ', author wishes to acknowledge the aid and understanding of Bristol, without whose help The Mr. S. R. Reifenberg, University of this paper would not have been possible, .-r;NV .t,. . .. ; ;. :.1;.. { e 4 ON THE STRUCTURE OF MINIMUM SURFACES AT THE BOUNDARY The problem of Plateau was solved for surfaces of varying type One of the problems considered previously was to prove that in [1]. for almost all interior points the surface of minimum area is locally Euclidean. In this paper we consider surfaces satisfying the same definition as previously but with added restrictions. For a detailed description and some examples of surfaces in this class the reader is referred to the original paper. Therefore such a discussion will be omitted here and only the definition will be restated. Definition: Let G be á compact Abelian group,. set in n- dimensional Euclidean space and A g be it group jdrr- Let j/ jut class j. A ,ice closed closed subset of S. Let x . 3 Lt written miss) . Let x be the g the inclusion blialanhin I l ; S be a Then there is defined the, Cech homology non- negative integer. [s A. G) a Let . m- 1(A ;G) --*1Ç any subgroup of 4 EEO) AG daD L if D with boundary Then we . se S is a surface of L. We will consider here only surfaces such that G is the group of integers modulo -2, n-3, and m-2. Further we restrict the boundary to be a finite set of straight line segments. We will show that the sur- face of minimum area with a given boundary is locally Euclidean at all non -vertex boundary points. Notation: I. Throughout this paper we will adopt the following symbols: S(x,r) is the closed solid sphere, center x, radius r. . J 2 x,r) is the surface of the sphere. o is a minimal surface satisfying our restrictions. 1(x,r) @ So(x,r). (X,r) is the set of points whose distance from X does not exceed r. $0414,$). Z(x,r) B is the set of boundary points of So. V is the set of vertex points of So. - 0 AtX dimensional Hausdorff spherical measure de- is the Suppose X is a set in N- dimensional Euclidean space. fined as: ri< Consider a set of spheres S(xi,ri) such that X Ci S(xi,ri). Let /\dX be the lower bound of taken over all such sets of spheres; where and W m - IT if m = 2. Let A mZ _ lim ¡-ro = W m P71 b and j' . 2 E w if m = 1 X. S(P0,R0) or S(Po,ro) will have no vertex points in the . interior. Given Lemma 1. ro r °° E -rr ) r 2 __. y E 1 where Po > 0 and S E BAY, - exists there an r> 0, -,o-- such that < n 2R 2 + E . The proof of this is similar to the relevant parts of : Chapter 5. Lemma a. hEif B ) where aEs(x.r) [1]J twit ,k,I Skg á surface contained ia B consists of m rays bE (x.r)), plus a set R1, = xa L(x.r) C S(x.r) with á,s,d Ar = - s(x.r). 1. boundary xb (where If fi az 3 a 4 ira Aux-0 < 0 ) VII... ü) S*C Cj'_ follows treat eiromrif ,. A it 3 141)(r.» .,ta tt. Y.Act Ofetti-,. à This thus tith0 . Ots + ii6(r.r) * caftI + 141) purely topological considerations. Y- Apkt of the The alternative `` w..i Shea ?, !s . P, s, .. t Giran sox E > o < E r - -E md , nL4,,(:.r) < Tr xE Trr<n _ 3 . IS swat labial e tra k s* ink Sita sat bail= sash that n i(a.r) > -r-r. r, Corollary A. _ dmoklb& .-_ above s.v_ lama being excluded by lama 1. tJllEt Wait r.at ask E.- bTrrb2(1 + E) [7\1 0 . . . bre <ÉÇ re, Trt < nMAIL--- 41T; n 1(:o;t} <. Write IT t+ MO" r, re °: f AMAX) t t ro <_ C and by < o (nt + ,1Udt 3/41.0 r: Then Z+ (adt 'o lisimeowds fre adt o Meet < r2 E # rrtE for all . t. fre it dt > o contradicting (1). 3/471-re2E than (1) 4 Po Lemma A., 2 ha then 11-11, c And x(Po,ro) Eig.Q, rrro2(1 ±- E) C re . 4rrrz < [AtPGPt# < Trr2. + E4). 0 Let fAl(P0,t)dt . F(r). ay [1, lemma *] o r 11(P,=) F(r) ° iir t' (Y) hence 1=(r) F(r) and >ir integrating from r2 to r1 where F(r 1) log r2 < rl ro we obtain .r12 > log r22 (r2) and hence i) . 02= r12 and r22 Q so F(r) /r2 is a monotone not -decreasing function of r. toms 4] jAI(P0,t)dt s gm for - < A (Poor) 1/21-r(1 + E4) r2 r2 r2 Et, !. and the hypothesis il(r). sv all r <s*. jams it. P.,aE =r, o E 11 kTrra2 < A2K(Pv.ro) S 1/27 r`,o21 + Eg2) -, then .gps 11.10 ti 5 < 41112 soma 4 By for all r <ro. that 3/4 1-r <ro r2 < nR( there exists an Then n211(Po,rl) hence for all t < t 1+ > : so + E (2) < ljrrr2(1 + 4 E5) is false for .rt < ro such that 5) 4 4 E . r2 /4 E5 n2 (P..r rQ. . f f 4E) > rrrl 3/4 > . , 3/4T-rt2(1 + 3/4 E5) that for all t satisfying (3) n1(P > T7t(1 + k Es) and hence r rl r, fAl(P0,t)dt o . , ffAflP0t)dt + « g2) > + If exists if 3/4r-r'r E <3 E 1R-P and which contradicts (3) then. there / rz Al + - tTr(1 + E5)(r22 - r12) o k1rr22(i 6. / r, . > [/1(P0,t)dt and using (2) Le ®a . o " ` 'IT r - r12) our assumption is false. + k -rr(1 + k E5) (r22 and hence some kalif.-plans ü(S(Po t))_ C i some " that <rl ( i+ A2x(i rr2(1 k 1/2rrr12(1 + 4 such 4 E 5) . . fAi(P0,t)dt < } r2 Assume 21(q.e.4- IL.7r2(1 + _ A _ <- .Trt(1 + E62) r*(PQ.t through, ro, such that (3) . t 6 Awl C (f *(Pc, 1 Let n ) . E6t). be the plane through Po orthogonal to BS(P0,t) BS(P0,t) r(P0,t) be any plane through D1 *(Po,t) r(P0,t) f Ì' *(P0,t) o. . BS(P0,t) extended and be that half -plane such that r *(P0,t) ' 1*(Po,t) O. By use of elementary spherical trigonometry it can be shown that for . any point x E 1*(P0,t), then xE(r *(P0,t), E6t). r Henceforth, the symbols . (Po,t) and r *(Po,t) will denote this plane and half -plane, respectively, as constructed in t¡°'r..y . a a the q<' t* E ,T . '.. - class t* ca S-V,, :5-/ Pß o s. rrr 2 rM < _T-rtsi.+ < .( I, - l U62 < E ,numbers t=r Al(Po.t) Let r E If 2 T a fauna. . Lemma Z. . let such that (P0,t) is divided into two half -planes by let and + _ and E7. such that (27b. G7) E) 2 MILL .. tAen l*(PA t) C (r *(P,, t )125 yt*) S (4) faiT_ less shan . (Po,t). _ *(Po t) a(P t) and let C be any finite set components of (Po,t) + 1(P0,t). We can find an open set Gt such .; :i ,;, ,,,tt :,.. ..>y C C Gt ( -1((P0 ) + 1(Po,t)) CO, E 72t) ,ana' nG. ( (Pp,t + l(Po,t)) ¿ AC E7 2t. and Let (' . y, of that ; , By an application . of the Heine Sorel Theorem we can find two closed disjoint sets !* and F . .,". , such that . . !. J. t. 0, r' t ; - 7. s .S. il. 7 C C We + + 1(a (5) +-S(P0,t) E72t). + 7S-(P0,t)Ch#*C(1*NNN,t) 1 *00,t) and #C t. ( .6(P0,t) 1100 By (4) and lemma 6 we see that E72t lies within of -4f (6) (Po, t). The elementary domain consisting of the points of s(P0,t) lying within 2 E72t of -S(Po,t) will contain F* and moreover this domain is con- tractible to a point so that by lemma 2A there will exist a [1] surface J *(Po,t) with boundary containing p*. A2J *00, t) lama By < Hence 21T E 72t2, (7) 2 and (4) nl*(PQ, t) Tit E and hence by (5) and (6) /17 *< TT t E74. By E1, lemma 8] there will convex hull of F exist with boundary A2J *(P0,t) < 4 ,/3 J H,(F (n f*) 2 5:1 ) Ó(Post) and S(Po,t *) planes - K(P0,t) + - S(P0,t). ('(Po,t*) and t*) 00,t) plus Let (I) the be Then by virtue of is: a surface [1, with boundary *(P,., t*) - + J(P0,t) J is a surface with , angle of intersection of the . K(Po,t) + J(Po,t*) 11-702 - s + IInt2 (mediate by projecting this surface ing lemma 5 to the hypothesis we obtain 4. (8) two intervals of I lying in Then rl*a,t). . is r r, ?(P0,t*) + /\_ This such that in the + 84(P0, t) . Now K(Po,t *) boundary K(P0,t) 11A], J(P0,t) + lying err 2t2 (E78. 4 Let J *(P0,t) + J*(io,t) - J(P0,t). lemmas 13A J *(Po,t) 0 a surface ' , o,t) cos 0, onto r(Po,t *). :.. (9) :. WY. a; 8 <Tr t2(ï A1/410, t) + 4E 2' for all t< r0. Together with (7), (8) and (9) and after manipulation obtain we sin (t) < 25 E 7 t <t* as long as and t >E7 t*. follows from elementary if While t Thus sin geometry. some E7ht* the result < (N 25 E 7 for all k 8<t*' lama .8. 11 CB > A2x(P 112-1102 r <ra,L4 h =. 7 and (+ , 2± CrY E s4 and by lemma 4 Tfte2 < AG be E the set of < hilt for all to < ro /1l(Po, t) dt 3/411-t02 + all r_ to< ro/2. Then to2, there ( exists such that for any rE(=-fg) fr r < /11{tw, r) < grit + 4 (10) Eta wo Let ). E . juz,#rLÇ*ito.a Let . < Tr o- _then .fax <E E8 Por, if not, a set GCE, 114 2113/4r. (12) r' then 4. nl( t) at . which contradicts ' ,l > A (11). Hence nl(P°,t)dt .. $ 4 4,t)dt + IAAl(Po,t)dt - AG) + nGOirrto + 3/4Trt l0 ' , Al(Po,t)dt - t f Oto Al(Po,t)dt. 9 , Therefore nl(P®,t)at < %Tr% + htoA . < to2 + ht 3/4n to2(h + 37 the corollary of and lemma 2 f Et4 + Tf Trtdt ho Trtdt - Tf h3/4). (13) 4 and 3 lemmas JAL*(P.,*)4s * 8G J nl(py,t)àt f Al(Po,t)dt - c . ? 02(3/4-rr Let tEt340 and assume there igtriottr by leers 7, for all know 2h' ). such (14) that r<S*, and rEts.4) r) ,. (13) 0. that n2[(A, er "TT 1141 ur, a point nl*ts .t)dt ( 0) , 26 E t0).. ti(X. 25 We is h - J 1R-c . by 2 E .t0) ? =2tp2-rr . lemma 3 and (15) ;Trto + E . 210E I( . 3(Esto) (so'to) A . >it.a AMP ,r) > tostcTr - h which gives a ' ItrQ, This contractictioo concludes r0/4) lies if E< 1. that whenever in a narrow Lamas 1 TT 210I) h3/4 J. holds for Stra, than slab centered about a constructed plane 10 through We It is will TO., will row muddles below contained .._.4t, Auk% . 2 be (I) l the mapping 111 km IL imi is 2 a reducing. D E v-Tr2 < i; x [stro.ro) - s(rl.rl) , j- 0 . ?-1S ± E10) iki8 1.2.40) tit Z 1, -_ r 'Pit 9 with whg230.3 isral °C+s O. mmtlit Iv,/ :LIAO tturall t Eu g nllcri.ar) Tt! The !!voi` oasis . ( Place every terms < Ala 4 TT - ... _ follows Tt _ t fi _ 111** incidentally that Pi baling* of I. Isms 11. Vasa E",Q Shia El);, . lama 9 with from to the as the origin surface and ease it . I . ( Ail 1/212+312,z'0. direct application of laws s1Ase. L is a igres) - D ,n?itc 11. looms ... r. area 14t1,rl)i harif proof I flowis ihe) The r mete' 8' [We lift,E AH (yl,e T' reducing and thereat& which links c .7Gs ,_ + Lifer.) ._ k to 616011011 LAM .40*1 ,,..ii -dgiastts, w..i..riCr 111.040) S TlY hikes= ,n UMW* ,. _ *113 distant* the surface interior of :f far boa thus in stro,te). MONA Let !o note that ì -.1.2->.-112. suf . :. re!°::'C,, . 11 Mg far every S(P',r) C S(Po.ro), P'Ç10, < n K(P' ,r) < Tfx2 T1" 2(1 + á plane f (P$, then L21 each r<1(16 ro there ,exists 4 E2-)' g CI r'(yo.r) . E *r) f(Po ,,r) a.r) E (X4PQ. r) . K(P8,r) is This .-...; E *r) . proved in [1] towards the end of Chapter 3. ,. .. 12. Lemma such that P-- 4Tf(16Ro)2 EBv' < E13< Rua A21G(P,16R0) E5-' E 1Ef ¡' k--8 and E135 IrTI'S1610)2(1 + then far each R<R and each PEK(Po. ©) there exists á plane through the segment the boundary through such that .:,. f (P,R) ,f Ç(¡(Px. . RICA The proof Case is "R, ; Ei,R S P it . by using cases of P and R: ; I. PEBY; . R <Ro. By lemma 10 II tr ( 81to) 2 < A211:0,00 < 1-111010) 2( + 4 . E. Thus by lemma 5 ' 3/4fR2 < n2K(P,R) and hence by II. rE(So + 16 E134) (16) 8 C K(P,2R) Case < 1ïR2(1 ( *(P,2R), (17) 28 E13R). - B)1100,110) and IP P1 each Pj corresponding to each PE14P0,R0) < R <Ro, By (17) for O . K(P , 2R) C (f (P :2&) > Let ('(P,R) be the plane through K(P,R) Case IZI. C PE(S©- ( r'(P,R), 29 28 P E11» and repro), parallel to Ella). B) s(Po,R0) and LP then (18) PI > R >1/16 I P PI. a. " », r V From (18) . 12 . 4 , Irirl ) C R(P, P,pI), !E13 ) ( then 1C(P,R) P jP I irr. PE(so r I IP"P'I 2 E . all ) for . 011(l0a0 lP I R in consideration. R<1/16 and } P'E TAP, corresponding to every 2 213 ( r(P,R) where cage C r(P.), I By IPPI. 2Tr IPi1101I (; + 24 < <_ j r E13 . t by lemma 11 rrIPP'I for ) 7 then (16) 2 < A ' < rr I P'I (1 + 25 E 13 4; and by lemma 5 - TIR2 AZx(P',R) Apply lemma 12 with - r - flit R and (P, x) (P,U C (r(P,R), sl G (x(P,R), E13* so small that *< r(P,R)S(P,a) , Taking m + 25 E132). to obtain Eft) , and ro (1 Eft). min , (2'13 (1:*) the desired result is obtained. "., . We now ,... quote a theorem from [1] with the . notation established I in this paper. Theorem: - point of G such that to each corresponds a plane r(X,R) , GS(x,R) and t g and f r set of points If G is a bounded in B3 and Po is a - C itt<ii o th and each =EGS(P0,80) there 0 X such that (r(x,R),ER)s(X,R) S(X,R) (X,R) s(x,R) C (G, ER) S(X,R) is a plane through Po such that ' (A) (s) , - 13 (c,Ex0)». Then if E < (C) there will exist 2"1 a topological 2 -disk G such that /16 gp) N C IIC GS(Po,Ro We will now construct from a set G and then we will : show that G is a set which satisfies (A), (B) and (C) above. a cylindrical coordinate system P(r,0,a) with P(P0,10) be the plane z w 0 and 1400, 10) o Define , w P(0,0,0). Let be the line 9 w 0 sr. e. Let F be a function whose domain is K(P0,R0) and such that F(P) r (r,w9,.e) for all PEX(P0,R,). Let G be the set of image points of F together with K(P0,R0). First we need another lemma. Lem pa j. Suppose xEX( Stu /2) ham ,.t Xt Filltcruto by virtue of lemma 13, through B. it and f2. B will divide f Cella) t R(X, into two halves both in ;Road. r which f1S(X has X near 19/10 hail) or is the projection of some For elementary topological reasons we could else point of 1(P0,ro) in F(S(X,R)), if such exist. f. . In particular there form a closed torus linking B but not meeting K(Po,ro). ' Then IXXI) will lie near a plane ('2S(X), 19/10 IXXI) point of K(Po,ro). 161,. - s. Now either every point of every point of ç I woad. F(S(X,R)) will cOntain no points, will bn na pair of distinct .4 i Let Y be a both 14 / 2 1's(x f and n2 , 19/10 I1' 1) + A (TI + /2x(, r2s(xj, 19/10 will substantially exceed I %1-rI Xxjl ,%I I xxja I zcI2`4 and one of them gives a con- tradiction, depending on which is fully covered by the projection of K(P0,ro). Now suppose that XEK(Po,ro) and that s(X,R) aider first the case where 1xjI < C S(Po,ro). Then by cases I -III of 1/16 R. lemma 13, x(X,*) will lie within 213 E13R of f (Xj, this plane passes through B the same will be true of 14 Thus all points of GS(X,R) lie within 214 X parallel to (l(Xj, I Xxjl Con* Ewa ixZjI). Since mi., -1(7.8)%' of the plane through ). On the other hand consider the case where IxXji > 1 /16R. By lemma 14 and lemma 13, case IV, our condition is then satisfied. Thus by the theorem quoted, G is a disc near Po. ' Now again quoting lemma 14, this disc is divided into two closed congruent sets whose common part is B, namely =00,ro) and F(X(Pe,ro)), each of which must therefore be half discs. This completes the proof of the following theorem: Theorem: For all non boundary points of So as defined 0 in the introduction of this paper, the surface is locally Euclidean. BIBLIOGRAPHY 1. Reifenberg, R. R. Solution of the Plateau problem. Mathematica 104:1 -92. 1960. .. . . ' , of .l ..171 i' Acta " .nt ; g ' , w a `