VECTOR ALGEBRAIC THEORY OF ARITHMETIC: Part 2 (March, 2006)

 

C.L. Greeno
The American Institute for the Improvement
of MAthematics LEarning and Instruction
(a.k.a. The MALEI Mathematics Institute)
P.O. Box 54845     Tulsa, OK  74154
greeno@mathsense.org

 

ABSTRACT:  Virtually unrecognized by curricular educators is that the theory of Arabic/base-number arithmetic relies on equivalence classes of whole-scaler vectors. Sooner or later the recognition will catalyze a major re-formation of scholastic curricular mathematics — and also of the mathematics education of all teachers of mathematics. Part 1 of this survey appears in the 2005 Proceedings of the MAA’s OK-AR Section Meeting. It sketches the vector-algebraic foundations of the Hindu-Arabic arithmetic for whole-number calculations — and the theory easily generalizes to include all base-number systems.  As promised, this Part 2 sequel outlines that theory’s extension to cover the whole-scaler vector arithmetic of fractions — which includes the arithmetics of finite decimal-points and of all other n-al points. Of course, the extension to infinite-dimensional whole-scaler vectors leads directly to the n-al arithmetics for the reals (forthcoming, as Part 3).

Background

Part 1 opened with an alphabetic construction of the Peano line of simple Arabic digit-strings — devoid of any reference to numbers — and extended that construction to achieve an Arabic numbers kind of whole-numbers system. Part 1 concluded by disclosing how the whole-scalers, vector-algebraic structure which underlies the Arabic arithmetic for whole numbers can serve as a mathematical basis for greatly improving the instructional effectiveness of curricular education in the whole numbers.

 

That development described how the geometric-vectors — along the alphabetic line of simple Arabic digit-strings — yield whole-number additions/subtractions, and scaler multiplications/divisions by repeated additions/subtractions. (The geometric-vector definitions construct those operations simply by counting spaces along the scale of any Peano line.) Imposing those operations onto that line suffices for the resulting structure to satisfy the axiomatic definition for systems of whole numbers.

 

Thereby, the line of simple Arabic digit-strings becomes its own system of whole numbers (as well as being serviceable as a “numerals” vocabulary for any other whole-numbers line). Accordingly, once that concept of Arabic numbers has been achieved, even the very young can freely use such numbers as the scalers for finite dimensional vector spaces. Even before knowing how to perform any of the arithmetic calculations, children can rely on hand-held calculators to count the spaces — thereby attaining the vector algebra on which Arabic arithmetic depends, as a rational means of learning the Arabic calculations theorems.

 

The most primitive whole-scalers vector spaces were labeled, “inventory spaces”. Those are finite-dimensional, and become whole-scaler “measurement spaces” after they are duly partitioned by “conversion equations” among their unit vectors. In turn, examination of measurement spaces reveals that the familiar arithmetic of Arabic numerals uses a far more complex whole-scaler vector-space, and relies on much higher developmental levels of mathematical sophistication within the learners. It means that early childhood, calculator-assisted, thorough learning about the vector-algebra of commonplace measurement systems should precede education in the far more sophisticated arithmetic of Arabic numerals.

 

While inventory and measurement spaces are finite-dimensional, the Arabic numerals are (singly-) infinite-dimensional vectors — normally truncated into finite simple forms. Generalizing from measurement-spaces, the Arabic numerals are but the proper vectors within equivalence classes of the base-ten partition of the singly-infinite dimensional, whole-scalers vector space. Within those cells occur the “carrying” and “borrowing” operations on which the Arabic arithmetic rely.

 

Thus Part 1 discloses that Arabic arithmetic is simply vector algebra, performed on the base-ten partition of the space of whole-scaler, singly infinite vectors — combined with intra-cell, vector-conversions to, and from their equivalent Arabic numerals. Beyond the vector operations, as such, is the operation of polynomial multiplication (with conversions) — whereby the arithmetic of the Arabic numerals is seen as taking place within the base-ten partition of the space of whole-scaler polynomials.

 

If the constructions in Part 1 have previously been published, they are not easy to find. Probably, that much of the vector-algebra of arithmetic is of strictly academic interest only within the foundations of mathematics. Therein, the above cited constructions have their places among the classical (cardinal and ordinal) constructions of the wholes, the difference-pairs construction of the integers, the planer constructions of complexes, the quotient-pairs construction of the fractions, the planer constructions of the rationals, and the n-al constructions of the reals. The same is true of what follows.

 

However, originality is far less important than is significance for curriculum reform. Each year, millions of humans are being educated in Arabic arithmetic — often in troublesome ways that mathematically make no common sense — while the learners are quite unaware that they actually are using a humanly natural version of vector algebra. Not only are partitioned, whole-scalers vector spaces not currently being used as mathematical foundations for prevailing school curricula — that context is not presently seen even in the teacher-education textbooks in mathematics. So, the mathematical basis for curricular instruction in arithmetic is quite shabby, weak, and confused — resulting in woefully ineffective learning and instruction, which undermines students’ mathematical growth during the years that follow.

 

Moreover, schools now are being subjected to growing pressures — and also to strong skepticism — about pressing “grade 9” algebra into lower grades. So, this mathematical news that Arabic arithmetic actually is based on a mathematically rigorous, “kindergarten” version of vector algebra opens the way for a major algebraic re‑formation of the early school curriculum through whole-number arithmetic. That pre‑destined reformation surely will begin within a decade, if not sooner.

 

So arise two questions: one mathematical, and the other, instructological.  First, does that vector-algebraic development of Arabic arithmetic extend, in an equally natural way, to include the usual arithmetics for the fractions?  Yes — and that is what the rest of Part 2 is about.

 

Then, what does such a mathematical extension say, if anything, about a beneficial, vector-algebraic re-formation of the core-curriculum through fractions? The pragmatic need is to make the arithmetic of fractions more common-sensible to the learners. For sure, part of the vector theory can be advantageously invoked. At the very least, the vector-algebra of fractions must become a compulsory topic in the education of all teachers of mathematics. Of course, there is always the danger that some over-zealous missionaries could worsen the situation by pressing the formal theory beyond the limits of the learner’s own level of mathematical maturity.

 

Introduction

 

Classically, the term, “fractions” refers to all number-systems that meet the axiomatic definition for systems of fractions. It makes no sense to call them “non‑negative rationals” without first having access to the concepts of “negative” and of “rational-ness”. The “new math” reformers of the 1960’s chose to use the term, “fractions” for all quotient-formulas that are expressed in the “over” format. But the purpose of this paper is to provide mathematical clarification of the arithmetic for the number system — for which the label, “fractions” suffices.

 

For the vector-algebraic theory of fractions, the mathematics, itself, can be formally (and inconsequentially) presented in a simple and straightforward manner. But as a conceptual re‑view of school mathematics, useful insights from vector algebra flow through a more constructive development. Much of the following discourse is aimed at facilitating the theory’s interfacing with a duly re‑formed curriculum. So, investing a bit of patience while tediously re‑viewing some kiddy notions will return some insights that, while seeming to be mathematically minor, are conceptually critical to the vector-algebraic reform of the core curriculum.

 

The vector theory of fraction arithmetic is a quantitative theory.  It conceptually contrasts to the traditionally confusing “denominator” theory — which is not likely to be replaced in public use or in the curriculum. Rather, the vector theory can be used to greatly increase the learners’ conceptual understanding of traditional notions — and this is a mathematical sketch of how.

 

As with the (finite dimensional) whole-measurement spaces and with the (infinite dimensional) base-ten numerals, the fractions are here constructed by partitioning a whole-scalers vector space into cells of equivalent vectors. But this is not merely an extension of the classical 2-dimensional construction of fractions — which views fractions as being (essentially slopes of …) the proportion-lines whose binary-ratios are ordered pairs of whole-numbers, within a coordinate plane. Rather — as already suggested by the decimal construction of the reals — the vector space underlying the fractions is infinite-dimensional, and the equivalence classes are far from being geometric lines in that space.  

 

The most comprehensive vector-construction of fractions uses a generalization of the standard decimal-point vocabulary for the decimal fractions — and of every other n-al vocabulary for the n-al fractions. The fraction vectors are the finitely non‑zero vectors of a singly-infinite, whole-scalers vector space — clustered into equivalence classes. The resulting vector-algebraic structure does provide all of the usual fraction arithmetic. But absorbing all those ingredients into single vector-algebra takes too much sophistication for it to clearly be applicable for purposes of improving curricular instruction. A more pertinent re‑view is through incrementally developing the fractions from measurement spaces.

 

The key to the construction is that the phonics for the classical “over” symbols for fractions comes directly from the quantities used in inventory-spaces —whereby “3/4” is often pronounced as “3 fourths”, which could just as well be written as “3(4ths)” or as “3F”.  The “3(j)” expression, introduced by the UICSM curriculum designers, likewise speaks of quantities — and it can effectively assist the transition from “3(4ths)” to “3/4”.

 

As always, each quantity consists of a denomination (for all things of its kind), with a (coefficient) numerator depicting an amount of that kind — nD, as with “3F(rogs)”. The phonic vocabulary presents the fraction denominations essentially as nths — which presently are not included in the read/write language of the core curriculum. The vector theory focuses directly on the nths denominations, rather than on only on their identifying denominators. Its vernacular includes the harmless, but helpful frivolity of occasionally speaking in terms of 1ths, 2ths, 3ths, and 5ths.

 

Empirically, nths are things, and each numerator speaks of an amount of things of that kind. When fractions are so viewed as being whole-scaler quantities, serial combinations in various denominations constitute an infinite-dimensional whole-scalers vector space. When those vectors are viewed simply as strings of coefficients, the vector space is seen to be the same as the one that underlies the base-n arithmetics for the whole numbers. The essential difference comes in how that vector space is partitioned by the conversion equations among the unit-vectors — which serve as denominations.

 

The stage for the arithmetic of fractions is set by the fractional partition of the vector space — which is where the conceptual difficulties arise in calculating with the over-numerals. That partition can glibly be achieved through formally imposing the conversion equations — thereby equating various infinite-dimensional vectors with each other. The vector operations being class-consistent, they nicely abstract to provide encompassing operations among the cells.  But mathematical elegance of that kind would be of little help for purposes of curriculum reform.

 

The troublesome mathematical complications arise in connection with the intra-class conversions. The ultimate need is to linearly order the cells into lines of fractions — in such a way that each cell can be located with reference to other cells.  But unlike with whole-measurements spaces or base-n spaces, not all fraction cells can be aligned through reference to a single 1-denomination scale.

 

Instead, the alignment takes recourse to an in‑common denomination for each finite set of fraction denominations. But a conceptual grasp of how that mechanism forces all fraction-cells into a dense line requires an inductive generalization from how it works in successive finite-dimensional cases. So, the focus, herein, is on how such finite-dimensional constructions are successively extended, ad infinitum.

 

Along the way, the role of in-common denominations cannot be substantively detached from construction of a “multiplication” among the unit vectors — which is not always done on vector spaces.  (Interestingly, fraction-multiplication of fraction units is seen to be surprisingly parallel to polynomial multiplication of polynomial units.) Of course, it is easy to formally and arbitrarily define the multiplication table for the fraction units. But a more plausible construction must be invoked — at least, if the common-sensibility of the vector operations is to persevere onto the unit-multiplications.

 

So arises the operation of fractioning things, by (natural) cut numbers — notions that are essential in the traditional curriculum, but traditionally are so thoroughly buried that they presently are of almost no instructional value. For mathematical nicety, what gets fractioned are the spaces of the Peano scales for each of the fraction denominations — spaces in which might be placed some fractionible things of any kind. The case of US coinage illustrates that fractioning is not necessarily a matter of geometric congruence. But geometric congruence becomes increasingly accessible with progress toward the continuum.

 

In what follows, the usual vector algebra, as such, will be re-viewed only where an unusual perspective might provide some additional perspective on the algebraic re‑formation of the core curriculum.  Instead, the major focus is on progressive construction of the fraction-cells of whole-scaler vectors.

 

Fractional Measurement Spaces

 

Measurement spaces are partitions of finite dimensional vector spaces — by use of specific kinds of conversion equations among the unit vectors. Some such spaces have several “whole” denominations — commonly called “5’s” or “10’s” or “2’s”, etc. — each of which is equated with natural multiple of one particular denomination.  Other measurement spaces have several “fraction” denominations — and still others have both.  

 

Herein, the discourse is greatly simplified by neglecting the possibility of multiple whole denominations — since those were attended in Part 1. Accordingly, we begin with a singly-infinite, whole-scaler vector space with one dimension, called “W”, for “wholes”. Following the classical language of vector algebra, all denominations are expressed either as units that classically are indexed by whole or natural numbers, as with “xn”— or are expressed as successive places,  [   |   |   | , into which the corresponding numerators are posted. (The standard use of parentheses does not speak of places nearly as well as do the data-boxes.) Those two vocabularies are interfaced by equating each xn to a numerators-string of 0’s and a 1 — with the 1 occurring in whatever place is numbered, n.

 

The set of all whole multiples of a unit vector is a dimension or axis of the space. Each is a Peano line for that place — discretely ordered, with a min and no max. Along each such axis, adjacent points are separated by 2-point intervals — the (line‑) spaces of that line. The alternating succession of points and spaces is a Peano scale — whereon the multiples of the unit vector are the marks of the scale. So, the family of all such axes is a singly-infinite dimensional Arabian counter — portions of which are viable empirical laboratories for representing the vectors.

 

In harmony with the western convention for measurements and for polynomials, the unit-vectors’ indices decrease from left to right. But for our abstract discourse, it will be convenient to have the indices also reveal which nths are identified with what places. The compromise is to number the places with negative integers, starting with ­­‑­1. The succession of unit vectors thus is  x-1, x-2, x-3, … — with each x-n being called, “nths”. To be sure, so labeling the unit vectors cause the indices to serve as denominators. But as seen below, their formal use for numbering the places does not imply their visible use within the vector language.

 

Within the singly-infinite, whole-scalers vector space, the vectors to be used as fraction vectors are the “finite” ones — those having at most a finite number of non‑0 numerators. Those constitute a singly-infinite dimensional subspace. But for convenience, as with the decimal-point vocabulary, the fraction vectors normally are truncated after the lowest denomination that has a non‑0 numerator.

 

That entire subspace of fraction-vectors is partitioned by the conversion equations, x‑1 = n(x-n).  Use of the phrase, “fraction vectors” implies that the whole-scaler vectors are clustered into equivalence classes, through those conversion equations.  The arithmetic of fractions is concerned not so much with the usual (and simplistic) vector operations with the vectors, as such, as with determining which vectors are equivalent to each other.  That is what motivates this approach of developing the arithmetic through the context of measurement spaces.

 

Within the so partitioned subspace of fraction vectors, the fractional measurement spaces use only finite-dimensional subspaces — always including x-1 as the largest denomination — and always, there also is another one that a smallest denomination for that space.  As with the whole-measurements spaces, each denomination is a natural-number multiple of that space’s smallest denomination.  For the novice, the latter condition greatly aids progressive construction of the number system — by gradually easing toward fluency with in‑common denominations.

 

As before, if the selected dimensions constitute a regular system, each non-smallest unit is a scaler multiple of the next smaller one. If a regular system also is consistent, all of those multipliers are the same — as with the commonplace “yardstick”, whose denominations make it actually an “inchstick”. Of course, the total family of all successive fraction denominations is not even regular — and that is the mathematical source of the curricular dilemma with fractions.

 

Simple Fraction Scales

 

The simplest fractional measurement spaces have only two denominations — x‑1 and, for some natural n>1, x-n.  Our development begins with partitioning those 2‑dimensional vector spaces.

 

When the largest (x‑1) denomination is called W (for wholes), the x‑2 denomination may be called H (for halves). That puts the vector-algebra language within easy reach even of those too young to spell the words. Notice that along each axis, the additions/subtractions of quantities, and their scaler multiplications/divisions are fully natural — as with 5H+4H , and with 3·7H.  For sure, that much of the theory is quite helpful to childhood students — because using that “phonic” vector-algebraic quantity-format initially circumvents the confusions that are induced by the appearance of numerical denominators.

 

That [W | H ] space consists of binomial vectors — as with 7W+5H — among which the vector operations are naturally inherited from inventory spaces. But the H-dimension truly becomes “halves” only after the conversion equation is imposed: 2H = 1W. It forces the equivalence of each W-quantity with exactly one H-quantity — whereby, the family of all aW+bH vectors is partitioned into equivalence classes.

 

Each cell has exactly one “H-vector”, 0W+bH. So, the Peano ordering of the Halves-axis induces a corresponding order of the equivalence-classes. Each W-cell has exactly one aW+0H vector — but between successive W-cells is exactly one non-W cell.  In effect, that line of equivalence classes overlays the H-axis, onto the W-axis.  That process inserts 1H into the [0W, 1W] space on the W-scale (actually, by inserting 0W+1H in between 0W+0H and 1W+0H — which has been equated with 0W+2H). That same construct likewise injects exactly one H-quantity into every W-space — thereby fracturing each whole-space in(to) 2 (spaces).

 

For that passage from the W-scale, to the W&H‑scale, the conversion factor, 2, is the cut number for that construction.  Although essential for converting fraction quantities, cut numbers are not overtly attended in the traditional curriculum — much to the conceptual loss of the students.

 

The resulting Peano line of cells — the W&H fraction‑scale — has its own line-spaces, and thus its own scale of alternating marks and spaces. Along that scale, each cell is a mark — and each comprises a single proper vector which locates that cell as being a W‑cell, or as being between successive W‑cells. Thereby, each of the W&H‑scale’s spaces is numbered by the proper vector for its lower end-point.

 

Each cell also has a unique reduced vector whose numerators-sum is the minimum among all numerator-sums for that cell — which establishes the intra-cellular conversions of “carrying” and “borrowing”, as with 7W+5H “carrying” to 9W+1H. Notice that the reduction is strictly about converting to smaller coefficients — not about the “location” of a vector, or of a cell. But for 2-place systems of this kind, each cell’s reduced form is also its proper form.

 

That construction of the W&H-scale illustrates the general nature of “fractioning” any Peano scale, by any natural number, done by overlaying another scale onto the original. But the simple-fractions scales are those in which the W scale has been fractioned by the nths scale — resulting in a discrete line of equivalence-class cells of proper or improper “mixed” binomials.  For more scales of the simple kind. BROY G BIV offers a colorful rainbow of 8‑dimensions, for generating seven simple scales. For even more learning-space, English offers 26 alphabetic dimensions — notably including the (daily) 24ths-scale.Such families of 2-dimensional spaces give ample room for developing the needed vector operations and conversions — even for clarifying the denominators-vocabulary for fractions. Of course, the cross-combinations of non-W denominations do not yet surface.

 

The immediate instructional significance of the construction of the family of simple fraction scales is that — even before introducing numerical denominators — much headway can be made into the arithmetic of fractions, by exploring the vector algebra of simple scales. Their constituent binomial vectors already are widely used in the core-curriculum. But at present, the visual display of numerical denominators obscures the equivalence classes of binomial vectors — and conceals the fact that “doing fractions” is an arena of “doing algebra”. Evidently, premature curricular injection of numerical denominators (in the standard “over”‑vocabulary for fractions) has long hidden the underlying, more conceptual vector algebra of fractions.

 

Fractioning Fractions

 

As a portal into the general theory, the simple-scales provide the conceptually crucial notion of fractioning any Peano scales and its spaces — by any natural cut number. In the simple cases, each W&nths scale is got by overlaying the nths-scale, onto the W‑scale — done by partitioning the [W|nths] space, relative to the conversion equation. But complications can arise when trying to extend such constructs by overlaying, onto the W&nths scale, a third scale — to achieve a W&nths&mths‑scale of cells of equivalent trinomials.

 

So, from the entry through simple fraction scales, the development must somehow explore the arena of whatever nths‑scales can be overlaid onto which others.

In order to achieve that, the overlaid mths scale must fraction the nths scale by some cut number — meaning that m must be a natural multiple of k. Accordingly, the natural refinements of the nths‑scale are those having denominations (2n)ths, (3n)ths, etc.  Such refinements are tacitly used when cutting  3(4ths) into 6(8ths), 9(12ths), etc.

 

A serial overlayment of several scales is most easily done when each newly appended scale is a natural refinement of the preceding one — and therefore of all that occur earlier in that series. Of course, the existence of infinite series of that kind guarantees the existence of dense lines of fractions.

 

For the finite-dimensional series, such a refinement chain of invoked dimensions constitute a regular system fractional-measurement. Those constitute a nice next-step in the development, because they are structurally identical to the regular whole-measurements spaces — both of which are naturally convertible measurement spaces. The only conceptual difference lies in which denomination is to be called “W” — with all larger denomination being “whole” denominations, and with smaller denominations being fraction-denominations.

 

It means that, through vector algebra, there is a natural transition from regular whole-measurement spaces to regular fractional-measurement spaces. In the latter, every vector equates to a quantity in the smallest denomination. Using those, the regular measurement spaces accommodate much learning about ordering, addition, and subtraction with multiple-denominations.

 

The regular spaces also are natural arenas for generating the “multiplication” of denominations by successive refinements — as mths of nths give (mn)ths, by using m as a cut number for fractioning the nths-scale. In that context, the refinements can be used for constructing the regular system and for converting the vectors — without yet broaching the search for in‑common denominations. Of course, the special case is the consistent systems, including the finite decimals: aW+b(10ths)+c(100ths)+….

 

Greater complications arise when trying to join two nths-scales when neither index is a natural multiple of the other. The search arises when trying to construct non-regular systems of fractional measurement.  The [W|2ths|3ths] space cannot be partitioned into a measurement space, since the smallest denomination is not a natural multiple of all others. The minimal encompassing space is [W|2ths|3ths|6ths] — which does yield a scale of equivalence classes. The [W|3ths|4ths] space is completed by appending 12ths, or 24ths, etc. But the minimal full system of measurements from [W|3ths|4ths] includes not only the join of the proposed scales, but each factor of that join:  [W|2ths|3ths|4ths|6ths|12ths].

 

The [W|2ths|3ths|4ths|6ths|12ths] full-system completion of the [W|3ths|4ths] subspace enjoys a realism that cannot be matched by merely calculating the LCM. Plus, the process tacitly requires conceptual identification also of the cut numbers that are needed for the various conversions. So, the vector algebra enables even the most complicated aspects of fraction arithmetic to be resolved within the limited confines of finite-dimensional measurement spaces.

 

[This development further discloses that the search for in-common denominations is not strictly a fraction-theoretic problem. Consider the non-regular, [50’s|20’s|10’s|5’s|W] subspace of the US currency system. It does not qualify as a complete whole-measurements system without an in‑common denomination — for which 100’s suffices. So, exploring in‑common denominations for non‑regular, whole-measurements can set the conceptual stage for subsequent explorations with the fractions.]

 

Of course, the multiplication of fraction-vectors with each other entails the crossing of each quantity of one, with all quantities of the other — by using denominations of the first to fraction denominations of the second.  But access to in-common denominations allows that to be more simply done by simply crossing two single quantities.  Likewise, the division of one fraction vector by another could be unduly complicated. Instead, converting both to an in-common denomination reveals that the quotient is a scaler — the quotient of the numerators — and that the division of fractions actually amounts to finding a per-unit ratio.

 

Ad Infinitum

 

While all of the above can be achieved through the vector-based, finite-dimensional measurement spaces, no one space of that kind can accommodate all of the fraction-vectors. Finite-dimensionality must be outgrown — and with it, the presence of a smallest denomination that is in-common for all others.

 

However, even in the infinite-dimensional space of all “finite” fraction vectors, operations on any finite set of them can be done in a finite-dimensional subspace.  Moreover, each finite family of successive fraction denominations — W, 2ths, 3ths, et al — has its own join, which is a refinement of all of those scales. Since every fraction denomination eventually occurs within one such finite-dimensional space — and in all subsequent spaces, the scale-alignment is preserved, thereafter. It means that the full line of fraction-vector cells is dense.