Lecture 4 : Gauss’ Theorema. Gareqiun Gauss applied work on Geodesy must have. motivated hin +0 address the Biot Question : To it possible to map any (open) region of Xhe sphere +o we plone uithowr distorting distance % TH you wank +p chow hat two +topologitad spaces Comnot [ne wapped Covrinuous | a ehueen each other Le. nok homeomerphic) , you would need 40 ute topological imvaviante Cie- Properbes War are wnecestanily preserved bay continuous maps 2.4. 4—}P camnot be homeomovpwic +0 a coss K iw RR? USE, Connectedness - a ropological invamiamte TE £ : Ceross XK) 26a) is @ homeomorphiem Ahen ©: (cross \{ cemker}) > Cab) \ LF lcemer)} iS Also @ homeomorphism, which mutt presene whe number 0% connected Components. This Qeverares a covivadicton ag: Ae oF cowmecked componenes in Sk = 4 He of Comnecked components in EO = 2 C4. —> Cannot be homtomoph, +0 OOD in R noteowpact: Compack, US2 Compackners — also Q +opdog'caR invariant,So +o answer Gauss’ question negatively , one needo an * isometric. invandant"” | fe. property / Guanting tha is preserved by Someries. Then show +hak any region of +he sphere Qnd any vegion oF +he plane would have, Afferent Cuch. properties / Quantities, Actually I dont knour if Gauss was the first son +O ask +hak Guestovy, but he was +he first +0 answer it: (Theorema Garegivm) the Gaussian curvature K of a& veguler surface in R° is ivwanant by local isometwes. Part of +he challenge of underckandng rwic “vemarkable theorem’ if +0 undersrand what (Ocal isometry means, ‘Def let $,5 be two requiar surfaces unr she Same _inavinsic dimensions (but Pocsioly digferent Quwbient dimensone.) A smooth wap €:VCS°AS oa neignbochovc V of peS is & local isometry ap if Riv > av)ic A AiPReoworphigm, Caretying : Km, Wag = CAGy OI, AL wv Deg), Vege. Note: Qi: VAQ(V) is @ loca isomerng at peS Qe aM rv is @ local isometry at eipreS + neve exists a local isomerry into S at evers PES , the curfae S ig Said +0 be, locally isometric xo 3. PY Note har it is nota Speometne velat ow, “6 is a isowemmic 40 3" So is locally isomeric +0 $” eg. Ser = SE > (=) S= sphere wth @ @ =x bottom Every, Gmatt enough) open vhbd of PE R Can be warped Sowe+vically xo an open Sek im the Flak bottom of $. So S is locahy isomemic to J. But +hee is no reason +O Oxpect rvak F is locally ispmewic to ¢. Therefve , we Say: Ree Ccomtinued) S and S are locally isomeric if S is local 4 isomeric 49 $ anh J is locally isometric +o S.€q. S = cylinder = [Ouy2): xreyteit S = pane = R Lek @: 92S, CHET > Ceose, sine, 21" For any (40,2) S ER™ lee Vit Hor Tr tot) eR Rly V7 aw) is a Af feoworphisw C smooth win & Swootn inverse) AQa|-snt oO ort oO o 4 Agtag = aes cose, ral ales e\"| ea 2 Cost. oO so. C wy wa % = CA wy Aw oe Petigy vee Nowe +hat 2 is Sujectve, fov any yes, choose Any PES ct. Qlp=y aud V aud Pst. aly: yrs atv) is a dA:kfeqmovphiswy, y So (Wy: giv) FV is @ looal isometry at y.v7 The example Above suggerie Hat $f we Can Sind a. function 2:598 at. oO Cuywh p= Cddpwh, du wa regi ,WreS @ ase S, ic. & is surjective, Ahern, S adh SC ave locally, 'Sometvic, . Nore : we dovit weed to check +nat P provides local dikfeomorphisms , ik is implied ba © : O =? dk : TS TeinS is a linear isometry =? dkp: TS > Tams is injective. =) dlp! TS > TeprS is bizeerve Ncammd dimension, on L =) & is a local Aiffemnorplism ote > TWS single , globally defined , map is net required tobe, And typically iS wok (as iw Aue Previous example), injective. R only, Fumishes a local diffeomorplisw Wear Cach poive PES, Q ikserp doeo vot need be 2 (gloloal) diffeomorphism. Typically , S awk S are vot even hhomeowmorpic , agadn Qs in ne prenims Acamae.Twis is Almost gerting off +opic (if all +wak You Care is Gauss’ vemarkable +heoven), but eb's quickly see wi ane Cylinder and. he plane re vor worphic. , leis Ame the topological Mvawiawt we ute is called: Simple Connectivity | s Every loop in (R* is homotopic = +o 0 Aviviak loop, but $s @ loop +thak Goee around a cylwder can be chorm +o & be yok homotopic to a Avivial loop. eR’: i oo teractioe 1S Simply connected, 5B pre but cyliwer is not, Oo -+hey Camnck be howtoworphic . &. $=R* Sa { Ceoru, su, cosv,suvi] tek _ u,v € Co,27) RS 4S CuviT Ceosu, simu, cosy, Sinv 1", is clearly gurjectve . Tk is also a loca isometry : “su ° Tv Oo buy = Coou 0 |, de ae=|5 ‘I. oO -sw Oo tsw Bub note thak S, F are not homeomorphic. CS is compact but S is not.) We covclude xhatSand F ao locally isomeric, S$ is a Flak rorus called v 4 sv 6g. S = cakenoid wee = ([coshvcos uw) v€lo,27) 4 coshy” Cosh v sinw|* ve v GS = helicord = Cte ce i) | tone > 6 R PE ° <> Digresstore = Careyaid weniwmizes area. These Gre example? 0F wininal Surfaces, With whe Property they have, Bern wean Curvakure , 14. W=0, evernwhere . Whar Aees winimeal area hove Ang sing ko With, +he WSO Covoltion % Th kung out snat, if a Surface S$ is pecuebed in the normal olivectione : XCuy i= (i Vt) = Xtyyd + pee Netw) vocal a vocal Q Sma Some sear field Por se, pave of perameter There Be AealSe] [yey = —fabeH dA From +Ws you see tak HEO is a necesrany condition fpr Grea minwization,.On *op of being & minima Sun back) ne we. Oe - . ough it ted keg on ane furbace Note (82]-(s)-e J con you ve Helicotk More can be Said about corxewik Mai te jAnese *wo minimal curfaces; Leading im particular 40 An Lxplaivation, of why they should be locally +vo far. isometric. But ts will +ake ue Tustead | TL wil chow you how +o reparametente Ane +vO Suxfaus to vereah a& local isometry : Heleord : PERU, PErcmhy = biel —_ a cos wh KxCatewnd . use the ovigival garamererizerion Cut) 9 C cork Cosa, Cosh Sinu, Uv) ¥ Let's Com pute. Zé <2F = [Sharer Srur? Paw] Dax] = [88 Lx, ~% “a : _2e of = ss =f mr Q iso local 'SDvnedry OR ~~ & E€=E€, F=F, 4=a & RE v z Proo$s Noke xhok iN ‘ { #89 %0)= Ku a 886 (w= Xv Q is @ local ‘somerng ARe (WHS, < rp) (Ker, < Yap) is Olinead isometry wees Le CARy wy, Akp We Peipy = Sr 2, VP, WesES? (Cpolonzerion iAenk+4) 5 Cwwre Cad tt &] Cl aod ~ = CAsty wt, at wr cov [2 ICE] use, &) So: _ _ BD is a local isomerry < E=E, FoF, a=G. Note : knowing 6, —, GQ is just te Sane af Knouwng © ,% on TeS, Vp (coord, nhbd) Also Tn wamifold *heory, £-,-%p is Speche +ype of -+ewor Led on S, and 6,F,& Ore +he Component, Pumorions wvder the Gordinete system, Knowing, €, FQ * (coor. whbdcS 2 R also weang we nour their denvatives Eu, Gv, Guu, uv, Bw, Fu, Fu, 7° ete. This wilh be exploited in +he proof of Gauss’ theorem belowProof 0% aust’ Tneorema Earegiven : The ‘dea is +o show +heb K cam be expressed wre. in enue 0f GF, Q avd hey (ist anh 2nd) partials. v re {CO RN ND EL =) ON xu? = — On re $ ZN ww PEO DON, ku? = — CW, xv? 7% DON, xw7 = — embed , cay , ane alswack Sphere of Xorus into R So nok +e vegulted cC* veguior surface. hag vanishing Gauss curvature everywhere . Fack . A C* veguiar compact surface SC a hag ab least one e\ieric port. ieK70 Sketch of grab Compact, =) boundedyr Ory ball *wat comane S im ix¢ ivterior , centered Qb O, et , 72 infimum of rhe ods of @l\ such alle S wuse shave @ pome wi Awis “hissing ball" aud S is ow ove side of tne Aangent plane at p Te ie evident «het, = Kg lp) 2 Kysccmg (P) = v2 7O se as wished. fa ( Gea Do Carmo Ch fov wore dletai\s,) So Accovdv4g +0 Gans , ie iS impossiole +o cousruch a Ch wap 7. AR” 2 R™ st O Twrmw= TH vy Tl uvteow) = Tt), * aud. wage (T) is C*-vequar tovus @ Tis a \ocal isomerry. For 1 Such A V eniste, then according +0 Gauss’ theorem , S= Image (T) wutt hae ceroGauss curvarwe everywhere which cowrvadscto Ae Lack we juck ecrablished Tn +e language of Riemaman wanifold Clecure 10) , we Say snat i iS impossible +o C*™ embed rhe ~flak rorug’ to IR. The Nagh- Kuiper we ver owed swab 6f we velax C2 +0 C' , it is possible! ANNALS OF ManiExATICS Vol. 60, No. 8, Navember, 154 Printed in O.5.A. (* ISOMETRIC IMBEDDINGS \454 By Jonny Nasu (Received February 26, 1954) (Revised June 21, 1954) Introduction ‘The question of whether or not in general a Riemannian manifold can be isometrically imbedded in Euclidean space has been open for some time. The local problem was discussed by Schlaefli [1] in 1873 and treated by Janet [2] and Cartan [3] in 1926 and 1927. This question comes up in connection with the alternative extrinsic and in- trinsie approaches to differential geometry. The historically older extrinsic atti- tude sees a manifold as imbedded in Euclidean space and its metric as derived from the metric of the surrounding space. The metric is considered to be given abstractly from the intrinsie viewpoint. ‘This intrinsic approach has seemed the more general, so long as there was no contravening evidence. Now it develops that the two attitudes are equally general, and any (positive) metric on a manifold can be realized by an appropriate imbedding in Euclidean space. This paper is limited to the construction of C’ isometric imbeddings. It turns out that the C’ case is easier to treat and that surprisingly low dimensional Euclidean spaces can be used. A closed n-manifold always has C' isometric im- beddings in £"", But to get a C* imbedding of an n-manifold with C* metrie T have (as of this writing) needed 14n* + 54n dimensions. One expecis this num- ber to be reduced, but it is clear that there will always be a sharp transition be- tween the C’ case and more differentiable imbeddings. At least (n? + n)/2 dimensions will be required boyond the C' case. This many dimensions were used in the analytic local theory.Tohn Nash Suman : c sige, tome e& 0 a Followup work of : Gromov (19208) 1980s) Borrelli, Tabvane , Lazarus, TWibert (2012) : a qj GW % 2am aw 5 PRI tO Q Pring. a (RATT S TI a Fig. 3. The image of a square flat torus by a C! isometric map. Views are from the outside and from the inside.The two big vesuns in Surface sheony : 41 Perhaps You should ask: why do geameters Care Abouvk shis Slat torus? Ts there a “flak sphere", “ fat a-hole +orus" 7% Tk uns out there isn't, ane reason is xopolgical and +he fack +hat +xhe torus is ane only topology (among all Compact Surfaces) with Q Lak metric dha: imberestivg Covsegvence? , Acoumularng into a big veswk called. rhe Lundormiaativre heovewry 2, The Gauss - Bonnet, sneprem says SK oA = Am (IM) for cluted. surfaces. M Guler chawactensac of M, & topological Invariant, E.g. ,™ io - a (Kaa =a Orff {kK dA =O eS) u 1