A STR.CT1RE THEORY 0F MODULES WITH A SPECIAL REFERENCE TO THOSE DEFINED BY THE CMtHY CONVERGENCE CRITERION by James Russell Brown A THESIS submitted to OREGON STATE CCLLEGE in partial fu1fillrnnt of the recwireitents for the degree of MASTER OF AHTS June l98 APPROVED Associate Professor of Mathematics in Charge of Major Chairman of Department of Mathematics Chairman of School of Science Graduate Committee Dean of Graduate School Date thesis is presented Typed by :--( ever1y Saralyn Brown IC I95- TABLE OF CONTTS Pa go INThODtXT ION SECTION DefinitIons 1. arid Preliminery Results 3 SECTION 2. Complete Normed Modules lo SECTION Rings and Ideals of Bounded Secuences 25 Other Ringe Related +n the Real Number Field 29 3. SECTION 4. BIBLIOGRAPHY 34 A STRtLTURE THEORY OF MODULES WITH A SPECIAL REFERENCE TO THOSE DEFINED BY THE CAIXHY CONVERGENCE CRITERION INTROD(XT ION familiar construction of the re?1 numbers as The ecuiva1ence classes of Cauchy sec'uences assumes particularly elegant a of rational numbers form when expressed In alge- A development of this type is given, for braic terms. instance, by Van der be the rational v:aerden [5, p. 211-2171. field, number let R the set of Cauchy se- the set of null secuences define addition and multiplication of ouences of elements of R, and of elements of R. R1 Thus We N elements in R1 by [a} + b} [a 4 = Then R1 is commutative ring with identity and a 1mal ideal in R1. field, called the real number field. numbers, an into R1/1. The principal results (i) R1/N a o field; subfield R' of is a max- Hence the residue class ring R1/N is sec'uences of rational is N (2) R is Rj/; (3) R' By defining "positive" ordering is concerning ring- a introduced R1,44 are: and order-isomorphic is dense in R14'; complete, that is every Cauchy secuence of (4) R114 to is elements of R1,4 converges; and (5) every bounded set of elements of Rj/ has a least upper bound. These results may be extended without any groat diffiCulty to the case wher (Archjmedean or not) sent R is an arbitrary ordered field [5, 211-2171. p. investigation is two-fold: The aim of the pre- first, to vary the above construction by considering rings of seruences of rational numbers other than R3 and kernels other than N and to in- vestigate the algebraic properties of the ouotient rings so obtained; and, secondly, to generalize the construction and the above-mentioned results to other metric-algebraic varieties, in such a way as to include some results which are known, as well, possibly, as some which are not known. In results section 1 we state and prove some preliminary later sections. Section 2 Is inpartial solution of the second problem which are used in tended as at least a posed above, while sections and intent, treat some 3 and 4, algebraic in nature very special cases of the first problem. Throughout this paper free use is made of such well- known results as the "fundamental theorem of homomorphism for groups with operators" 133 tions to rings [2, and to modules. p. and its specialize- In addition are used the following set-theoretic and algebraic symbols, with the usual meaning: Finally, X, c, , (L , we adopt the abbreviation jx,) (denoting "1ff" for "if a ard se6uence. on' if." ci SECTIJ 1. DEFINITIC*S AND PRELIMINARY RESULTS this section with some fundamental theorems from the thory of rings and ideals. be arbitrary rings such that Theorem 1. Let P and P and let N be an ideal in Q. Then the intersection N(\P is an ideal in P. Proof. NP is not empty, for C NfP. If N/P If x,y c N('tP and p e P, (o), then NIP is an ideal in P. then x - y P and x - y t N, so that x - y t Nfl?. Also We begin ( C X P, plies implies xp P, px e O, which implies xp c N(tP. p and px N, p P; and x p L P N, px Thus xp N. im- P NCIP Corollary 1.1. If N, P, Q are rings such that N P Q and N is an ideal in Q, then N is an ideal in P. are rings such that P Q and if Theorem 2. If P and N is a prime ideal in then N(\P is prime in P. Proof. Suppose x,y t P and xy e Nt'P. Then x,y L Q and xy e N. Since N is prime in O, either x e N or y e N. C? Hence either If A and x e B N/P or y e N,IP, and NR? is prime in P. are subsets of a group R, we denote by A + B the set of all elements a + b of R, where a e A and b e B. In general, A + B will not be a ring whenever A and B are. However, we have the following theorem. Iheor 3. If P and Q are rings such that P C and if Moreover, N is an ideal in Q, then P + N is a subring of C. it is the smallest subring of O containing P and N. 4 Proof. P + N is certainty not empty. Supr.ose n2 are arbitrary elements of P + N with P1,P2 + P2 P + n1, and Then N. (Pi p3. 1) = (p2 - (p - p2) + (n1 fl2) E + N P and (pj nj)(p Since N is an idoal + fltP2 + fljfl2 that + N Ing both p + P n c R is Q, a P ring. N. If -f N p L C R. fliP2 L N, so Since pjp2 L P, let Now P that P1fl2 this proves and Pifl2 belongs to N. end and = (p1P2) + (pin2 +fljp2 tfljfl2). R be a subring of and n s N, then p,n contain- Q E Hence R. This corrpletes the proof. If N is an ideal in R, we write R/N for the residueclass (ouotient) ring of R modulo N. Now In the above theorem we have, by corollary 1.1, that N is an ideal in p + N . Moreover, if o s P + N and we denote by and , respectIvely, the residue classes in Q and in P + N which However, since any . contain o, then ouite obviously n where n s N, may be written in the form element in . Thus ve arrive at the following inportant we have 4- isomorphism theorem. heorern 4. If P and O are rings such that P Q and N (P +N),k Q/. ideal in C, then P,'(PiN) s QA be the residue let X Proof. For ec class containing x and consider the mapning x -e- of P into (P+N),ì. Clearly this mapping is a homomorphism. Moreover, it is onto, for any element of P + N may be titten is an ' 5 in the form p + n with x P e and then , p e x e and n P N N, and hence kernel of the mapping is PCN and by and p + n = P\N. If . the the fundaienta1 theox e Thus rem of homomorphism P/(PrN) have already seen We that (P +N)A If and the theo- Q/N, and so is proved. rem 4.1. QLol1ary in such , that of subring is extension of O Q are rings and ideal in the ring such th3t P/IN if Q/ is N ideal an Q/. then P,4 If N is an P . P Q, P N Corollary 4.2. P (p +N),4. (0), then and only if P O4 , and if is an intersects every residue class of modulo N. Next we introduce the concept of a module and give a brief re'sum of the elementary theory. Definition 1. Let A be a ring. A set is an A-mod.LkQ 1ff R is a commutative group (operation +) and there C) exists a on A X R to R such that for all a,b function and all x,y 1) ii) e R, a(x (a +y) = a.x + a'y. ax f bx, +b).x = iii) (ab)x a'(b'x). the "productt' ax we shall write briefly For Definition 2. Let is a submodule of i) A L R R be an A-module. iff for all x,y e N A ax. subset and all. N a e A, of R ifl axN. Nw let x y be an A-module and R iff partitions x R y, y and u submodule of R. e define is an ecuivalence relation and into eruivalence classes . Furthermore, if - y x a N Then ' e N. have we x + u y + y and ax ay for all are thus led to make the fo1lowin definition. DefInition 3. Let R he an A-module and N a submodule of R. The A-module consisting of the set of ecmivalence classes defined by N and the compositions given by a A. We i)+x+ forallx,yER, for all x e R, a e i, where ; the eriulvalence class containing x, is called the Qtient moduLe of R modulo N, and is denoted by R/. That R/ is indeed an A-module is seen by the following eouations: a(x+y) a( f) = ii) ax = 3(T) (a+b) (ab) y x of P Xj ii) atbx) c;i;T; = Definition 4. mapping axbx (TTT x - a(b). single-valued homomorphism 1ff Let i-,C be A-modules. onto 'j, y a() X2 C is called Y2 implies ax a implies -+ ay for Xj A + X2 all + a e A. is called the homomorphic image of P. The kernel of the homomorphism is the set of all x e P such that x-0. If the mapping is l-1, lt is called an isomprphi and we C write P Q. 7 odu1e A concept of is only sreciei cese of the a more Qeneral group with operators (seo Jacobson [2, p. 128- a 186]). Thus the entire theory for the latter carries over to the former. In particular, we hve the fjindarnenta1 If Homomorphism. florern P is an I.- is a submodu1e then is homomorphic to P/a; and, conversely, if F is homomorphic to R and N is the kernel of the homornor}hisn, then Is a submodule of P nd module and G I P4\i. R L2, p. (uite obviously if is submodule of R, then state the analogue of Theore suhrnodules of Ci. a ¿are A-modules and P submodule of L. We is a may now theorem 4 for modules. Let Ç 5. P c Q c R Thon let be an A-nodil.e and P(N and P + N I and N be are submodules of L and P/(PiIN) Proof. P(\N. ax e Also Pt\N. Supìose x,y x x e a submodule. are arbitrary members of + Y2 Y1.Y2 e arid N let (xi +yi) - a A. Then (x2 y2) (xj + N -X2) Then x - y a e A. N implies ax e P, PIN is Thus e (p+i)/! PN and N ax Suppose with + (yj + x,x2 e and a(xj +yl) Thus P - N is a = submodule. ax1 ay1 e P + and hence Xj y2) N. e y e P + N and arid LJ Again let be the e.uiva1enc. class of x and consider the mapping x ping is a £ A. a - e of P onto P + N. module homomorphism since ax = -+ Q modulo N This mapfor all a The proof follows then exactly as for theorem 4. Corollary 5.1. N Ç P Ç p, then If N,? Ç P4': Corollary 5.2. Q/1'. If Q is an A-module and if N and P are submodules of C such that sion of P. Ç4 P Q are A-modules such that arid (o), NiP then ÇA' is an exten- iff P intersects every eruivalence class of C modulo N. Theorems 4 and 5 may both be deduced from the "second isomorphism theorem" for groups with operators as stated and proved in Jacobson L2, p. 136]. following theorem is phism theorem" [2, Theorem 6. if M Ç 135]. p. If i! and N are ideals in the ring P, and For each x e P, and P,41 p/. (P4)/(N4) Proof. the special case of the "first isomor- a then N/M is an ideal in N, In a similar way, let L P4QI and L P/ be the ecuivalence classes modulo M and N respectively which con- Consider the ma.ing tain x. maping is hence - x X x e N. Thus ; = X x.y so that the map::ing is a Moreover, . ' P4 onto Ç, then single-valued, for if y of +y xy -+ +, -xyxy, homomorphism. Now x P/Ì'J. - y The arid 1ff zeN 1ff so that N,' is the kernel of the homomorphism. By the fundamental theorem of homomorphism then, N4 is en ideal P/N. and (P/M)/(NA1) if rings This theorem obviously remains velid and ideals are replaced throughout by modules end sub-modules. However, we shall riot make use of this fact in the secuel. Theorem 7. Let Q be a commutative ring with identity, r and let MO be the set of all elements of the form E m.o. i=1 ' O, nj e O (11,2,...,r). Then MO is the where m1 M smallest ideal in Q which contains M. s r If Proof. Z m11 e MC and E mId e MC, and 1f 1=1 1=1 t h en r r+s s in o. E 1=1 m!c 1 rn"o'.' 1 * 1=1 and r r E i L i=l is an ideal In O, Since r o E Thus MO for in e M Implies me ideal in C and M P, then MO, in Mfl E 1=1 Q m.o 1 L identity e, M Finally, if P is an has an e MC. PC = e MO. P. o 2. SkCTION As before hi w' COMPLETE NORMED by a me&r ODTJLES complete system or space one in which every Cuchy secuence converges. shown Metric that It my be any metric spece can be imbedded In a complete SpaCe, re eouiv1ence classes of whose eiements seuences [4, p. 84-881. The space which we shBll Consider is fl sorne ways 1es genri, but it includes as speciel czses two im:ortant applicatlons of this construction integral domains with valuation and norrned linear spaces. toreover, the present theory has the adventage of preserving the algebraic structure of these special cases. Thus an integral domain will be imbedded in an integr'l domain and a linear space in a linear space. We begin this section wits some preliminary definitions. Definition 5. A valuation for an integral domain A is A, a function ç on A such that for all a,h i) p(a) P, a completo ordered field, ii) ç(a) > o for a O; (o) = o, Cauchy : iii) iv) p(ab) c(a +b) < Since any ordered field tp(a) + may be com1.eted in the manner indi- cated in the introduction, it is no restriction to suppose that P is complete. We note that, if A has an identity e, p(-e) = 1. In any case then by li) and iii) cp(e) = 11 Recalling the definition of an A-module from the previous section we note that by i) and ii) of definition 1, 1f o E A and O C R are the respective "zero" elements, then and all ox = aO = O and (-a)x a(-x) -ax for all a R. In case A has an identity e, we shall for the sake x of simplicity insist on the additional restriction iv) ex = x for ali x c R. We are now justified in making the folloing definition. Definition 6. Let A be an integral domain with valuation p to P an ordered field. Let R be an A-module satisfying the condition O o or x o) ax = O implies a for R is a function i on R for all a E A, x R. A IA E such that for all x,y i) (x) ii) i(x) iii) = i(x) and ìl A, a P, > O O: for x O, ii(o) .t(ax) (x+y) iv) Since R t(3x) = < i.t(x) q-a)p(-x) by condition iii). ciled briefly + cp(a)i(-x), A module with a we hcve (-x) norm defined is ring A itself is a normed module if R is taken to be the additive group of A, multiplication !s ring multIplication and .i = p. If A is the real (complex) number field and (a) a1, then R is a real (complex) normed linear space. In any case, since a normed module. The 12 t(x), R is * metric space, with the metric a (x,y; y). We are now ready to define what is meant by a Cauchy se'uertca in a norrned modulo and thus to introduce the nues- tion of completeness. flnitiçm values in [x} and let Then of R, Let R he an 7. i) having be an infinite senuence of elements is convergent 1ff there exists an each positive ¶ n > N ii) Amodule with norm implies P an E x R and for integer N such that < .L(x - Cauchy 1ff for each positive P 'r there exists an integer N such that n.m > N iii) irplios (xx) null 1ff for each positive an integer n > N In case i) we call x 1.1 'r < 'r; e P and there exists such that implies (x) the Limtt of {x} < 'g. tt. x 11* (or 11m xe). n Let R be an arbitrary norred A-module (norm in ?); N be the set of null senuence of elements of R; R0 the set of convergent senuences of elements of R; and R Cauchy se'uences of elements of R. R1. the set of If R is complete, R Otherwise, we shall see that R may be imbedded in complete A-module, let Let us defino addition in R and a 13 multiplication of eiecrents of R1 by elements of A by + [yI = {x a[x) = Íax). We shalt show that N, R0 and R1 are Amoduies and that N R0 R1. The associative and commutative laws of addition follow immediately from the corresponding laws in R. [x} [X1 + + oreover, [xh [01 [-X fol, where [01 is the secuence in which each element is 0. addition we have (a+b)(x) '4ax3 aUx} + Eye)) = atx1 a(x+y} = [ax+ay) identity, Now if (x),[y) - a[x) = a[bx t [i.X) m Rl and - (xm c a A, then since A(x - Xm) y y1) + (y we have L( x + then l'[xfl} Ux + (aya) f(ab)x} r [a(bx)} (ab)h) + + (t,xI z[ax) If A has [ax+bx} [(a+b)x) - xm) < 'r/2, ( < i/2 - in 14 inì pl le s ((x-y n )(x m ymfl<; and (xnxm) < Implies p(a)(x_x) (axnaxm) provided a[x} R1. - This shows that O. a Thus R1 is an A-module. < 't:, R1 and [y} Similarly, R0 and N are A-modules, Suppose [x} each positive R0. E P a -r n > N Then there exists R and for x E positive Integer N such that i(x-x) implies < -/2. Thus m,n and > N [X) { x} t then 11m 8 Proof. Let fines a R R oA' 4x - xm) submodule of R3. a t R0. Thus N is 1 that R014 < 'r If a sub- and R1/14 are also a/N. be the A-module of secuences = X for all n, Then R l-1 mapping of R onto R and all a + They are related in the fo11o;;ing manner. ThtcL x - R,, We have seen in section which (x = O and Ext) module of R0 and hence of A-modules. < xm) Therefore, R0 is R1. N, i(x - implies e A, ? Ro, for [xfl} x - [x for de- under which, for all x,y t 15 +y ax Since [x} converges to Now(ÌN X for each [o), and Thus R L Proof. 1, If N. By By corollary 5.1, R/ If R Is complete, e then R14 ' R. is complete. R1, then [R(x)) converges. e {xr,) [x} Ç Next we show that R114 Lemma R0 there exists that Is (xe) = x, R0/t4 Qr.pi1ary 8.1. [y), R0. corollary 5.2, we have R1,43. + [ax) - x, R such that 11m x E [x} y) [x From the properties of < i(x xm) + .t(X) < a norm, i(x..x). Also t(xm) L(X-X) < - (x Xm) Thus - Therefore, [.(x)) is a (x)I A(xn_Xm)i Cauchy secuence of elements in Since P is assumed to be complete, the lemma follows. Let *([)) = 11m where [;;) is the ecuivalence class modulo N which contains [x}. P. 16 Lemma 2. is a norm for R/ and agrees with ii on the image R0,M of R. Proof, First we show that R1/N satisfies condition o) of definition 6. either a O or a[) Thus if {x) e N. then (ax,) O, Therefore, either a s N and O or = O. By definition and lemma a N, t*({1) 1, s P. If (xe) then um M(x-y) O. But for all n, so that lia (L(x) lia it(x) and 1.L* is lia singlevalued, ((r}) a O *((1) O ii(y)) (y) o, a Moreover, iff [c) since s N iff (x) O end j*(ax lia a a J) a L(ax) (a)lim L(x) t;;i o, for ill n; 17 Finally y) t(x i(x) < (y). + (n 1,2,...), {x) - so that *([x +y}) + t;}) This proves that Now if *([}) + L*([) + ; is a norm, (1 is the image of x t R, then e N and as above litri um (i(x) - (x)) O, *U'}) (x) This completes the proof of lemma 2. Lemma 3. is dense in Ro,4 R4' with respect topology imposed by the metric T«x,y) Proof, [} Let y). be any element of R1,4 Then there exists tive element of P, to the a and 't positive integer such that > no n implies i(x n -x ) < r/2. o Thus 1f [yl is the constant seruence defined by y for all n, then ri or, > o (x-y) implies < passing to the limit *((x_;)) *([} .. ty}) ¶72 < < 'V, any posi- 1, c/2, o p'.] -[}) ;i But : h} R044, since r. converges to x0. Thus every [} 'p-neighborhood of contains an element of R0ft. eierrent of R14< is the limit of Every Lemma 4. < a con- vergent seouence of elements of R0/N. Proof. Let h;} be a monotone, null seouence of positive elements of P. By lemma 3, there exists for each a t R1,4, < a ¶ with seouence Thus, for each ' integer rio such n > > O that P*(a _ r) there exists a positive RoAI such in P, that implies no .L*(aj3) < < ¶, and hence u = 11m Theorem 9. is complote with respect to the a114i norm null se-uence of elements of P. ( In the Archimedean case we may take, for instance, ' be a Cauchy secuence of elements of R1/N 1/n.) Let Proof. Let be secuence of elements of Ro/N such that j*( < and let Now for any positive and fl2 be such a ¶ t P there exist positive integers that n,rn and a > t implies i*(a_a) < /3 n1 19 T Then for n,in > n0 .*(p n . m) max(nj, na), we have *(_u) < + + + + + ¶13 + ¶/3 and hence Let R. {} is be By lemma under this isomorph.sr. eouonce. [xL Then By theorem 8, R0A1 e1ments sectaence of a T Cauchy soouence. a 1L*(am_3m) ' of R such that 2, xJ is a Cauchy R1,4 be the eouivalence class containing Let for each n ji*(a,) L*(_a) 1(Xk -x) + p.*(c_,) + hin k Now there exists an integer fo such that irpies r,k > r. (xk -x) < i/2 and fl > no implies T1. < ¶/2. Thus implies n > no 11m L(Xk -xe) k implies *(a_a) <T. Thus i:al converges. We have shown that any normed module can be imbedded in a complete normed module. pi.x) In particular, any real (corn- normed linear space can be imbedded in a real 20 this section will devoted to applying this result to integr1 domains. Theore 10. Any integral domain with valuation can he (complex) Banach space. be imbedded in a complete Proof. Let R be having values in P, The remainder of integral domain with valuation. an integral domain with valuation Then we have seen that R is an R-module the above construction we obtain the cornplete normed R-module R1A. Now since R R04, we have, in an obvious way, that R1/ is an RoA-moduie, in the sense * is a valuation for R0,41 of isomorphism. Moreover, p* and p* j$ a norm for RjA'4 with respect to this valuation. Let OE,B he arbitrary elements of R1/. By lemma 4, there exists a seruence of elements of R0/N such that with norm p. By 11m a The secuence ta} given any positivo is r a L for, if there exists an Integer Cauchy senuence, P, f 0, then such tha t rn,n > o implies implies If ß = O, then _arn) < ç*(an_aîn1) [01 is a nm)P*3) < Cauchy se-uence. define two compositions + and in the set R1AT in the following way. + is the group addition in the module R1,4. For any a, e R1,4\T We a2 wh er e = 11m 21 1ima. aRo/. [Note that if a t R0/4, then a!3 11m a$ n This composition Is single-valued, for ges to a, then (a null senuence. -a} if [a} also conse'uently [a and shall show that gral domain. is - a 11m a3. with these compositions is an into- R1/W We theory of double numbers seruences of real 247-292] is easily carried over to R1/N. In p. E3, particular, if are convergent seruonces, then 1%) end +) 11m um a nfl where a conver- Thus um aB The afl.] 11m OEA. B = 11m a 11m n a n 11m k The = 11m + 11m first Bk of these is a con- seruonce of To prove the other note that for wo Q*(a) some M C < M. Thus ,*(a),*(.) = so that (u -aB} is a 11m nul]. seouence. (a3_aP) = O < M Thus P and all n 22 Now let 11m um 1a, {} be Cauchy secuences of elements of = 11m a0 R04 converging respectively of R1/, 11m to the arbitrary elements and let y he any other element of R111. x and Then = 11m a0(iim = 11m 11m n k = 11m 11m n = flkh' k ),Im -(u')'y. Also + y) = lin! = 11m cx( + y) (cL +ay) + 11m = '3 11m + and = 11m (-e)y = 11m (!1y+By) 11m y + um = a.y + B'y. Fina11y suppose a'B and for each positive 'r O and P O. Then 11m iÇ3 there exists an integer fl O such 23 tha t > no n *() Since O, *() implies < 'r this implies = arid hence um a = O. is automatically a commutative group with respect to addition by the definition of +. Thus R1/Ï is an integral domain and theorem 10 is proved. If R has an identity e, then RoA has an identity and Now R14 = = jim so is that an ae = a, um identity for R1/I. If R is commutative, t h en = a3. um um is also commutative. Finally, if R is a division ring, then Thus R1/ (11m 80 ¿1),a = 11m Çc um a1k um tÇ1a that a_l and 11m w = um have proved the foitowing field with valuation can be imcomplete field with valuation; any division ring Cpro1lar bedded in a corollary. 10.1. Any 24 with valuation can be imbedded in with valuation. a complete division ring 25 3, SECTION RINGS AND IDEALS OF BOUNDED SECUENCES to the investigation of sequences of rational numbers. Let R be the rational number field, let N, R0, R1 have the same meanings as in the revious section, and let R2 be the set of all bounded seouences of rational numbers. Thus N CR0 CR1 CR2. We define addition and multiplication in R2 just as we did for R1 in the introducW return now tion; thus, [a} + [b} [a)'[b} {a + b} [ab}. is a commutative ring with identity. Proof. The associative, commutative and distributive laws follow directly from the corresponding laws for the rational number field. Let [a} be any element of R2. Then {a} + [o) Theorem 11. R2 [aa) + [.m.a} = [O), [a)'[i} is the additive identity, [-a} is the additive inverse or "negative" of [a}. and [i) is the ring identity. Suppose now that [ar), [b} e R2. Then there exist rational numbers A and B such that Thus (o) IaI Then < A, fbJ < B for all n. 26 Ia +bI < [a} Thus JbI < A + L3 IaItbI <AB IabI for all n, + Ia1.I a}(b} [b}. + R2. This completes the proof. N is an ideal in R2. Theo;em 12. Suppose Proof, exists a By the definition of > O a t {b}, [c} R2; N. Then there rational number A such that Ial 'r c [ar' < A null secuence, there exists for each a such that positive integer n > no Likewise, there exists n > n1 Theorem 13. Proof. implies lbI implies IabI a < 'v/A 'r. < positive integer n1 such that implies IbI implies Ib-cI Ib t is an fb} Thus 1a}'[bn' for all n. {c1} t/2, < N, nd N ¡cRI n I < 'r/2 + Ic n I < ideal in R9. N is not prime in R9. Consider the secuences a (o,i,o,i,...} b Now a,b R2 - N and ab [o) N, integral domain. Corollary 13.1. R9/ Ço1lar.y 13.2. N is not maximal in R9, is not an 'r. 27 conclusion that We are thus led to the there exists at least one ring P such that NCPCR i) P is an ideal in R2. ii) Let be the class of all rings P satisfying i) and ii). JI By theorem p PuR1 is an ideal PuR1, and since over, N Pr\R1 1, {i} in R1 £ R1 for all P and [11 E ' P, II. More- we have Since N is maximal in R1, this proves the fol- R1. lowing theorem. Theorem 14. PIR1 Theorem 15. R24' N for all P is a proper extension of the real number field R1/M. By corolliry 4.1, R114! Proof. /. Theorem 16. For each P R1/(PuR1) = R1,44 Obviously, R2/P will be is £ U, R2/P is an extension of By theorem 4 and theorem 14, Proof. vidsd P By corollary R2/N. 13.1, R1/N i R2,4. not maximal in R2. (R1 a P)/P Ç R2/P. proper extension of R1/1 pro- However, we are interested in ths ces. where P is maximal in R2, for then, and only then, R2/P is any P £ First of all, we must determine whether a field. 11 is maximal in R2. In order to do this we make use of the Maximum rjncip],e. Let be a class of sets such that the union of any linear subclass of Q is an element of Then Q h By . maximal elements. linear subclass A of a that, if A,B we mean then either A A, The9rj 17. Proof. By the definition of p L II IT, Q B cless A a such A. or B contains maximal ideals. p p, :: II we have O implies II. Thus we need only to show that there exist maximal elements in Let A be any linear subclass of TI. U(A) is an element of H. Then a A, b e either A hence U(A) a B or E ideal in R2. also N cU(A). Since N Moreover, U(A) Then a,b B. Thus a - b, p' A for any A and U(A) II. A, and B, U(A) and ac A for each A we have A, N, for if it were, we would have A su(A) = N contrary to the definition of I} Ra. c Since A is linear, A. Suppose A ac belong to B. - b and is an ,b e U(A) and For suppose A. Then L1[AZAEA) B for some A, B e il. II Finally, Thus U(A) so that [i) % U(A). By the maximum principle II. R2 contains maximal elements. Thus there exists an ideal tension field of R1/. such that R2/ Moreover, by theorem homomorphic image of R24\i, 6, is an R2/ ex- is a 29 OTHER RINGS RELATED TO THE REAL NUMBER FIELD SECTION 4 One might ask whether something similar to what we have done for R1 and R2 might riot ouences of rational numbers, S be the set of all be done for other rings of se- We see secuences of rational numbers. immediately that S is Let This is indeed possible. commutative ring with identity, for a the sum or product of two secuences (as defined in the in- troduction) is certainly a We shall be concerned, seouence. in general, with subrings of S, ideals in these subrings and the corresponding nuotint rings. In order to make our results meaningful, we should like to be able to compare the resulting quotient rings with the real number field R/L Thus the following theorem is not without interest. Theorem 18. R is t)e largest suing of S in which N is an ideal. Proof, Let a} for an infinite number of subscripts n. exists a Thus be an unbounded secuence. monotone, unbounded subsequence a O Moreover, there [a } of [a}. with k Let b O. b 'k k n (k1,2,3,...) and let = 1/a k (kl,2,3,...). = [ab1 Then (be) is has a a = O for n null sequence, but subsequence converging to i and hence is not null. In addition, we may show that there is no proper ideal in S which contains N, for, by theorem 7, NS is the smallest ideal with this property. But NS S, for (l/n} N and [ni t S. Thus, if fa,.} is any element of S, then [ari = [i/n}[na1 e NS. On the other hand, S is not simple, for if R is the set of all seouences having a finite number of non-zero terms, is ari ideal in S. By a trivial application of the maximum principle S must contain maximal ideals. If M is a maximal ideal in S then S4 is a field and (by theorem 2) M(\R1 is a prime ideal in R1. Then S/M is an extension field of the integral domain R1/(MfR1). The latter may or may not be comparable to the real number field. A ring which is in many ways more interesting is the of all seouences [as) such that [(n + l)a -na_1} set then e R R1, Now I n+l that, sums [i. so -kaki] [(k+l)ak E k O by a well-known theorem on p. 101], we that have CC = a n the regularity of Cesaro R1. Moreover, we have the following theorem, is Iheorem 19. Proof. [i} e C, and we need only show - nan_li [(n+i)a identity. [(n+l).l - n.1} t R1. Thu! that C1 is a subring of S. commutative ring with Firs t: of all, [i} Let [agi and [bei be [(n+1)a a arbitrary elements of Cr1. e R1 and ((n+i)b -nb.1} e - na n-1 } - [(n+l)bn - nb n- Then R1. 1 = Thus 31 n(a1 b1 b) belongs to R1, and hence 1aI C. belongs to [(n+1)a X [b} [a, - Also _ na1}[b} {(n+1)anbrì na_1b} R1 and [a_1)[(n+l)b - nb_1} y = belongs to R1. + X {(n+1)a1b na_1b_1} Thus [(n+1)ab y = = * ((n+1)sb na1b1 + nanibni} + [a1}[b} t R1, Finally, thai X [an_i}[bn} + y [abÌ and hence [a)[b1 [(n+1)ab - C. theorem 2, N1\C1 is Now by a n_ibn'-i} R1 This completes the proof. prime ideal In calling upon the regularity of Cesaro surnmability, we N(C1 if and only if [a,.) converges to o see that [a} na_1) converges, which in turn is possible and [(n+i)a if and oniy if [(n+l)ar nani1 converges to O. Thus Again 1ff [(ri+i)a - n-l1 t N. is a field. Theorem 20. Proof. It is certainly a commutative ring with the identity, e, which is the eouivalenco class containing [i). [a1 Lt [a) e e Nr%C C, and let 32 if 1. otherwise. = O (rn) £ Thus O for n greater than sorno N, then r If (a.j = O, a «1 N(\C1, Let (b C. -nb) O for any n and Then b (l/b). - tan) - (l/b1 s and [(n+l)b Thus 1b) [bflØ1), all belong to R1. Hence [i/b)[l/b1)C[b} + [b11 n/b1} = [(n+l)/b and (l/b} t CZ1. nb1}J [(n+i)b £ R1 Now = [11 - since a(1/b) Since fr} 13y c O O if i otherwise, a N(\C1, this completes the proof. theorems 20 and 4, we have the following theorem. Theorem 21. CZ1/(NCCZ1) is a subfield of the real number field. Finally, we remark, thzt the set of sequences sum- mable by Cosaro means of the first order is not It is, however, an R-module (as is th any "proper" method of summability). a subring. summability field of We might thus construct 33 modules comparable to the real number field (conceived as module over the rational number field). However, case, the theorem corresponding to our lemma hold, as a i in this fails to so that we would be hard put to redefine the result ring. a 1. 2. :3. Hardy, G. H. Press, 1949. Divergent series. Oxford, Clarendon 396 p. Lectures in abstract algebra. Jacobson, Nathan. 217 p. voll 1. Princeton, D, Van Nostrand, l9l, und Pringsheim, Alfred, Vorlesungen Uber Zahlen Functionenlebre. Rand 1, Abt, 1. Leipzig, B. G. Toubner, 1916. 292 p, 4. Thielman, Henry P. Theory of functions of real variables. New York, Prentice-Hall, l95. 209 p, 5. Van der Waerden, B. L. Modern algebra. Vol, 1, Tr. by 258 p. Fred Blum. Now York, Frederick Ungar, 1949.