Está en la página 1de 3

On a comprehensive mathematical notational system

Michael Yu
January 31, 2014
Abstract
Missing.

Foreword

Mathematics notation has come a long way since the days we wrote theorems strictly in the words of our
natural language. I do not believe mathematics should suffer now, as it had when we used words instead of
equations, because of our complacency with the sufficiency of present day notation.

2
2.1

What is wrong with current notation


Function application

The f (x) notation as a standard for function application introduces the hazard of unnecessary parentheses.
As a result we have ugly things like f (g(h(x), k(x, h(y)))) note the four layers of consecutive close parenthesis at the end! If we are concerned about wasteful use of (), we can unambiguously write it instead as
f ghxkxhy, assuming the arity of each of the functions are fixed, via the (Polish) prefix notation. Unfortunately this is much more confusing for us; the parentheses, though ugly and semantically unnecessary, do
help us parse the expression.
After the above considerations, I am inspired to ask, is there a notation where the most economical unambiguous expression also happens to be highly readable? The answer is infix notation, i.e. a + b. We like infix
notation and when it looks right, we want to use it. But it is not visually appealing too often, making us
choose instead the f (x, y) notation. In the example f (g(h(x), k(x, h(y)))) from before, if we use the binary
functions g and k in an infix way like x y xgy and x + y xky and we use the unary functions
f and h in a more economical way like x f x and 1/x hx, we can rewrite the expression as
((1/x) (x + (1/y))) f ((hx)g(xk(hy)). Certainly we never write ((1/(x), +(x, 1/(y)))), so why
do we write f (g(h(x), k(x, h(y))))? We truly want to be able to write things in infix form, but our functions
being letters instead of symbols make us slightly uncomfortable (I will discuss the subtle reason why it makes
us uncomfortable in the next section). A notation where we have the choice of writing function evaluations
in infix form in a natural way, that is, without discomfort, can definitely be an improvement.
We can recognize the source of our inclination prefer infix notation when it is generalized to be used for
n-ary functions. For example, if I have a tertiary function each of whose parameters have distinct meaning, I
might name it tag with and use it as x tag y with z. In general, infix notation generalizes to n-ary functions
like so:
hoptional-name-parti arg 1 hname-parti hname-parti arg n hoptional-name-parti .

Consider the similarities between this way of writing a function evaluation with the way our natural language
is formed, e.g. use x to fill y with z (this kind of syntax has been explored in the programming language
Objective-C).
Also, you probably know f x and f : X Y are mutual contradictions. The reversed order is a problem,
but really it is a symptom of a weak notation system. What if we have the choice of writing either f x or
xf , and either f : X Y or f : Y X or even Y X : f , and the choice of being explicit about which
order to go?

2.2

Choice of operator representation

The problems do not disappear when we only have operators represented by infix symbols. Here are some
choices of notation for infix operators that I think are awkward.
1. Using left-right symmetrical symbols for noncommutative operators.
2. Operators that need to change the representation of its arguments.
3. Operator whose intrinsic representation is not a concrete mark on paper (i.e. is invisible).
4. The same (possibly invisible) symbol being used for distinct operators.
5. Operators have lower precedence than they visually appear, or vice versa.
A particular example I have in mind for the second property is exponentiation, where the power argument
needs to be shrunk and raised when writing the exponentiation operation. In fact, the first situation with
nonsymmetric symbols representing noncommutative operators is a specific instance of a generalization of
this second situation, because the location of the arguments, as opposed to the symbol of the noncommutative
operator, have to be swapped if we want the operator to evaluate the other way.
The invisible operator situation leads to frustration when one wants to show something being done to the
operator itself, e.g. repeated application, taking inverse. The same symbol, e.g. the invisible symbol, being
used for different operations leads to the same problem as operators having visually contradictory precedence
in that one cannot easily tell just by visual cues which operator binds tighter, hence requiring the use of
more parentheses for clarity. For example, does sin xy mean sin(xy) or sin(x)y? What about sin x y? Which
binds tighter in f g h? What happens if we rewrite that as f gh? What if we throw in other operations
like +, , in there?

2.3

Discrete collections and indices

The notation we have for discrete collection objects like sets, arrays, matrices seems natural until we start to
want to do complicated things with multidimensional versions of them. We run out of places to put indices.
We nest indices so deep that a line of equation becomes columns of nested indices. Permutations of ordered
collections get religiously marked in each index explicitly, bloating otherwise short expressions. Complicated
filtering logic used in sums and products take up more vertical space than the sum or product content, all
the while being in tiny subscript font.

A new notation: principles, ideals, standards

3.1

Principles in visual structure

Notation is ultimately meant to be consumed by the eyes, so let us first gather the basic but fundamental
visual structure principles that the mathematical notational system we all currently used have.
1. Semantics is formed on a surface of one background colour by drawings composed of lines of a thickness
that is thin enough that we never confuse width for length in one fixed foreground colour.
We really are starting from the most basic of the basic principles here. Note that I do consider
black and white to be distinct colours, as mathematician would reasonably consider the set of
shades to be a subset of the set of colours.

También podría gustarte