# Metareal numbers

1. Aug 5, 2009

### Phrak

The hyperreals consists of elements having the property that

h > 0
h < 1/2
h < 1/3
h < 1/4
...
and their negatives.

I'm curious to know if this continues within the definition of the hyperreals. That is, for all h,

g > 0
g < h/2
g < h/3
g < h/4
...

2. Aug 6, 2009

### g_edgar

The original hyperreals will still work for that. For example, if $h$ is infinitesimal (that is, $h<1/n$ for all positive integers $n$ but $h > 0$), then $g = h^2$ satisfies $g < h/n$ for all positive integers $n$ and $g > 0$.

3. Aug 6, 2009

### Phrak

I think you're telling me that numbers that satisfy g and h are all hyperreal numbers, right.

You know, I thought I had a ranking system worked out for this--magnitude classes of hyperreals: g0 would be a real number, then g1, g2... Now, I'm not so sure I haven't used a circular argument. I think I need something better than wikipedia...

4. Aug 7, 2009

### g_edgar

There is also $\sqrt{g}$ and $g |\log g|$ and many more...

5. Aug 7, 2009

### Hurkyl

Staff Emeritus
Have you read Keisler's calculus text?

6. Aug 9, 2009

### Phrak

Thanks for the reference! I've just now completed the relevant sections in Ch1 (1.4-1.6) concerning the algebra. The calculus is scattered throughout the text, so more difficult to pick-up.

The scheme I've been developing appears to be equivalent to hyperreal numbers--oh well, that's OK. I simply introduced an additional structure so that indeterminate forms such as

$$e \cdot H$$

are determined by the structure, where H is an infinite, and e is an infinitessimal.
(btw, recall that indeterminate means that the number could be a) finite and real, b) finite and infinitessimal, c) infinite.

Last edited: Aug 9, 2009
7. Aug 10, 2009

### g_edgar

Last edited by a moderator: May 4, 2017
8. Aug 10, 2009

### Phrak

Very nice, g_edgar. I like your writing style, as well. I've bookmarked it for reading.

I didn't understand this, "There is also $\sqrt{g}$ and $g |\log g|$ and many more..."

Rather than talk around things, what I have done is set up a hiarchy of infinitessimals, and infinities by defining a unit infinitessimal. I believe I am dealing with hyperreal numbers with the addition of subscripts, to indicate magnitude scales:

Real numbers have no subscripts (or a subscript of zero). There is an infinitessimal element that acts similarily to 1 that is represented as 1-1.

1/1-1 = 1+1 defines infinite unity.

1-12 = 1-2

In general 1-1n = 1-n

Multiplication: for c = a b, cp = am bn , where p = n + m

Addition: an + bn = cn and (... + a-2 + a-1 + a0 + a1 + a2 + ...) + (... + b-2 + b-1 + b0 + b1 + b2 + ...) = (... + c-2 + c-1 + c0 + c1 + c2 + ...)

an > bm for all a and b, where n > m amd a ≠ 0

I could add more, but I'm sure you get the idea.​
Although, √1-1 could defeat this scheme or require irregular forms with funny subscipts. an = √(1-1), n = ?

Were you suggesting there is more than one way to define a scaling hierarchy?

Last edited: Aug 10, 2009
9. Aug 11, 2009

### g_edgar

No, actually I was mistakenly assuming that by "hyperreals" you meant the already-existing system with that name (also called nonstandard analysis) ... http://mathforum.org/dr.math/faq/analysis_hyperreals.html", for example.

Last edited by a moderator: Apr 24, 2017
10. Aug 11, 2009

### Hurkyl

Staff Emeritus
Isn't that just the real function field (i.e. the fraction field of the ring of real polynomials) along with the ordering that makes the indeterminate variable infinite?

i.e. the thing you call 1-1 is just the rational function x-1 (where x is the indeterminate)

Other interesting fields are the field of real formal Laurent series, and the field of real Puiseaux series.

But if you want to define something that has the same analytical behavior as the reals, then a nonstandard model of real analysis is your only option.

Last edited: Aug 11, 2009
11. Aug 12, 2009

### Phrak

According to Keisler's calculus text, the square root of an infinitesimal is still an infinitesimal. I was afraid it would lie half-way in-between reals and infinitesimals.

With 1-1 replaced with the symbol 'u', we can dispense with the cumbersome subscripts for infinitesimals infinities.

I'm afraid I don't know enough mathematics to follow you quit, but you bring up something I hadn't noticed. With this scaling scheme, a hyperreal, a* can be expressed as

a* = ... + a-2 u-2 + a-1 u-1 + a0 u + a1 u1 + a1 u2 + ... ,

where the subscripts resume their usual role of distinguishing variables, and the a's are real valued. Other than the a's being real valued rather than quotients, is this what you were talking about?

that seems to be the case, though comming at it from a different direction, I'm trying to add to it. After scanning Keisler I didn't find any referral to infinitesimals with numerical values. That is, where are the equivalent counting numbers 1, 2, 3, ... in the hyperreal infinitesimals?

More, I wonder if u1/u can stand in for the identity element of addition without inconsistency.

Last edited: Aug 13, 2009