Latest / Foundations of Geometry / 039 - The representation of algebraic numbers and of integral rational functions as sums of squares
Transcript
- 0:02Speaker 1: Chapter seven, Section thirty eight of Foundations of Geometry by
- 0:06Speaker 1: David Hilbert. This is a LibriVox recording. All LibriVox recordings
- 0:11Speaker 1: are in the public domain. For more information or to volunteer,
- 0:15Speaker 1: please visit LibriVox dot org. Recording by jem Nitzer with
- 0:21Speaker 1: slides and illustrations at YouTube dot com. Foundations of Geometry
- 0:27Speaker 1: by David Hilbert, translated by E. J. Townsend. Chapter seven
- 0:33Speaker 1: Geometrical constructions based upon axioms one through five, Section thirty
- 0:38Speaker 1: eight the representation of algebraic numbers and of integral rational
- 0:44Speaker 1: functions as sums of squares. The question of the possibility
- 0:49Speaker 1: of geometrical constructions by the aid of a straight edge
- 0:53Speaker 1: and a transfer of segments necessitates for its complete treatment
- 0:58Speaker 1: particular theory of an arithmetical and algebraic character, which it
- 1:04Speaker 1: appears to me are themselves of interest. Since the time
- 1:09Speaker 1: of Urma, it has been known that every positive integral
- 1:13Speaker 1: rational number can be represented as the sum of four squares.
- 1:18Speaker 1: This theorem of Verma permits the following remarkable generalization definition.
- 1:26Speaker 1: Let k be an arbitrary number field, and let m
- 1:29Speaker 1: be its degree. We will denote by k prime k
- 1:34Speaker 1: double prime dot dot dot k to the m minus
- 1:37Speaker 1: one prime. The m minus one number fields conjugate two k.
- 1:42Speaker 1: If among the m fields k k prime k double
- 1:46Speaker 1: prime dot dot dot k to the M minus one prime,
- 1:50Speaker 1: there is one or more formed entirely of real numbers
- 1:54Speaker 1: than we call these fields real. Suppose that the fields
- 1:58Speaker 1: k k prime dot dot dot k to the S
- 2:02Speaker 1: minus one prime are such a number. Alpha of the
- 2:06Speaker 1: field K is called in this case totally positive in
- 2:10Speaker 1: k whenever the s numbers conjugate to alpha, contained respectively
- 2:16Speaker 1: in k k prime k double prime, dot dot dot
- 2:20Speaker 1: k to the S minus one prime are all positive. However,
- 2:24Speaker 1: if in each of the m fields k k prime,
- 2:28Speaker 1: k double prime dot dot dot k to the M
- 2:31Speaker 1: minus one prime there are also imaginary numbers present, we
- 2:35Speaker 1: will call every number alpha n k totally positive. We
- 2:41Speaker 1: have then the following proposition theorem forty two. Every totally
- 2:47Speaker 1: positive number in k may be represented as the sum
- 2:51Speaker 1: of four squares whose bases are integral or fractional numbers
- 2:57Speaker 1: of the field K. The demonstration of this theorem presents
- 3:03Speaker 1: serious difficulty. It depends essentially upon the theory of relatively quadratic
- 3:09Speaker 1: number fields, which I have recently developed in several papers.
- 3:14Speaker 1: We will hear call attention only to that proposition in
- 3:18Speaker 1: this theory, which gives the condition that a ternary diophantine
- 3:23Speaker 1: equation of the form alpha zi squared plus beta ata
- 3:29Speaker 1: squared plus gamma zeta squared equals zero can be solved
- 3:35Speaker 1: when the coefficients alpha beta gamma are given numbers in
- 3:39Speaker 1: K and zi ata zeta are the required numbers in K.
- 3:46Speaker 1: The demonstration of theorem forty two is accomplished by the
- 3:50Speaker 1: repeated application of the proposition just mentioned. From theorem forty
- 3:56Speaker 1: two follow a series of propositions concerned earning the representation
- 4:01Speaker 1: of such rational functions of a variable with rational coefficients
- 4:06Speaker 1: as never have negative values. I will mention only the
- 4:10Speaker 1: following theorem, which will be of service in the following sections.
- 4:16Speaker 1: Theorem forty three. Let ffx be an integral rational function
- 4:21Speaker 1: of X whose coefficients are rational numbers and which never
- 4:25Speaker 1: become negative for any real value of X. Then ffx
- 4:30Speaker 1: can always be represented as the quotient of two sums
- 4:34Speaker 1: of squares, of which the bases are all integral rational
- 4:38Speaker 1: functions of X with rational coefficients. Proof we will denote
- 4:44Speaker 1: the degree of the function ffx by m, which in
- 4:48Speaker 1: any case must evidently be even when M equals zero,
- 4:54Speaker 1: that is to say, when ffx is a rational number.
- 4:58Speaker 1: The validity of theorem forty five three follows immediately from
- 5:02Speaker 1: Erma's theorem concerning the representation of a positive number as
- 5:07Speaker 1: the sum of four squares. We will assume that the
- 5:11Speaker 1: proposition is already established for functions of degree two, four, six,
- 5:17Speaker 1: dot dot dot two, and show in the following manner
- 5:21Speaker 1: its validity for the case of a function of the
- 5:24Speaker 1: MTh degree. Let us first of all consider briefly the
- 5:30Speaker 1: case where ffx breaks up into the product of two
- 5:34Speaker 1: or more integral functions of x with rational coefficients. Suppose
- 5:40Speaker 1: p of x to be one of those functions contained
- 5:43Speaker 1: in ffx, which itself cannot be further decomposed into a
- 5:48Speaker 1: product of integral functions having rational coefficients. It then follows
- 5:54Speaker 1: at once from the definite character which we have given
- 5:58Speaker 1: to the function ffx, that the factor pFx must either
- 6:02Speaker 1: appear in ffx to an even degree, or pFx must
- 6:07Speaker 1: be itself definite, that is to say, must be a
- 6:11Speaker 1: function which never has negative values for any real values
- 6:16Speaker 1: of x. In the first case, the quotient ffx divided
- 6:22Speaker 1: by the quantity pFx squared. And in the second case,
- 6:26Speaker 1: both pFx and ffx divided by pFx are definite, and
- 6:33Speaker 1: these functions have an even degree less than m. Hence,
- 6:39Speaker 1: according to our hypothesis, in the first case, ffx divided
- 6:43Speaker 1: by the quantity pFx squared and in the last case
- 6:48Speaker 1: pox and ffx divided by pFx may be represented as
- 6:53Speaker 1: the quotient of the sum of squares of the character
- 6:56Speaker 1: mentioned in theorem forty three. Consequently, in both of these cases,
- 7:03Speaker 1: the function fvx admits of the required representation. Let us
- 7:10Speaker 1: now consider the case where ffx cannot be broken up
- 7:13Speaker 1: into the product of two integral functions having rational coefficients.
- 7:19Speaker 1: The equation f of theta equals zero defines then a
- 7:23Speaker 1: field of algebraic numbers k of data of the m degree,
- 7:28Speaker 1: which together with all their conjugate fields, are imaginary. Since
- 7:34Speaker 1: according to the definition given just before the statement of
- 7:39Speaker 1: theorem forty two, each number given in k of theta
- 7:43Speaker 1: and hence also negative one, is totally positive in k
- 7:48Speaker 1: of theta, it follows from theorem forty two that the
- 7:53Speaker 1: number negative one can be represented as a sum of
- 7:58Speaker 1: the squares of four to definite numbers in k of data.
- 8:04Speaker 1: Let for example, negative one equal alpha squared plus beta
- 8:10Speaker 1: squared plus gamma squared plus delta squared equation one, where alpha, beta,
- 8:16Speaker 1: gamma delta are integral or fractional numbers in k of theta.
- 8:23Speaker 1: Let us put alpha equal a sub one theta to
- 8:27Speaker 1: the M minus one power plus a sub two theta
- 8:32Speaker 1: to the M minus two power plus dot dot dot
- 8:36Speaker 1: plus a sub m equals fee of theata. Beta equals
- 8:43Speaker 1: b sub one data to the M minus one power
- 8:47Speaker 1: plus b sub two data to the M minus two
- 8:51Speaker 1: power plus dot dot dot plus b sub m equal
- 8:55Speaker 1: to si of theta. Gamma equal c sub one data
- 9:02Speaker 1: to the M minus one power plus c sub two
- 9:06Speaker 1: data to the M minus two power plus dot dot
- 9:10Speaker 1: dot plus c sub m equals ki of data. Data
- 9:17Speaker 1: equals d sub one data to the M minus one
- 9:22Speaker 1: power plus d sub two data to the M minus
- 9:27Speaker 1: two power plus dot dot dot plus d sub m
- 9:32Speaker 1: equals row of data. Where a sub one a sub
- 9:37Speaker 1: two dot dot dot a sub m dot dot dot
- 9:41Speaker 1: d sub one d sub two dot dot dot d
- 9:44Speaker 1: sub m r. The rational numerical coefficients and v of
- 9:49Speaker 1: theata sigh of theta ki of data row of theata.
- 9:55Speaker 1: The integral rational functions in question having the degree m
- 10:00Speaker 1: minus one in theta from equation one, we have one
- 10:07Speaker 1: plus the quantity fee of Theta squared plus the quantity
- 10:13Speaker 1: si of Theta squared plus the quantity chi of Theta
- 10:18Speaker 1: squared plus the quantity row of Theta squared equals zero.
- 10:25Speaker 1: Because of the irreducibility of the equation, f of x
- 10:29Speaker 1: equals zero. The expression f of x equals one plus
- 10:35Speaker 1: the quantity fee of Theta squared plus the quantity si
- 10:41Speaker 1: of Theta squared plus the quantity chi of Theta squared
- 10:47Speaker 1: plus the quantity row of Theta squared represents necessarily an
- 10:54Speaker 1: integral rational function of x, which is divisible by f
- 10:58Speaker 1: of X. F of x is then a definite function
- 11:04Speaker 1: of the degree the quantity two m minus two or lower. Hence,
- 11:10Speaker 1: the quotient f of X divided by ffx is a
- 11:14Speaker 1: definite function of the degree m minus two or lower
- 11:18Speaker 1: in x having rational coefficients. Consequently, by the hypothesis, we
- 11:25Speaker 1: have made ffx divided by ffx can be represented as
- 11:30Speaker 1: the quotient of two sums of squares of the kind
- 11:34Speaker 1: mentioned in theorem forty three, and since ffx is itself
- 11:39Speaker 1: such a sum of squares, it follows that ffx must
- 11:44Speaker 1: also be a quotient of two sums of squares of
- 11:48Speaker 1: the required kind. The validity of theorem forty three is
- 11:52Speaker 1: accordingly established. It would perhaps be difficult to formulate and
- 11:59Speaker 1: to demonstrate the corresponding proposition for integral functions of two
- 12:04Speaker 1: or more variables. However, I will here merely remark that
- 12:10Speaker 1: I have demonstrated, in an entirely different manner, the possibility
- 12:15Speaker 1: of representing any definite integral rational function of two variables
- 12:21Speaker 1: as the quotient of sums of squares of integral functions,
- 12:25Speaker 1: upon the hypothesis that the functions represented may have as
- 12:30Speaker 1: coefficients not only rational but any real numbers. End of
- 12:37Speaker 1: Section thirty eight