You are on page 1of 4

The Vortrix Product is a Complex Isomorphism

A. Pseudonym
June 13, 2018

Abstract
Robert Distinti claims to have created a ’new’ algebraic structure, different from the com-
plex numbers, in which -1 has a square root (two, in fact). I show that his system is isomorphic
to the complex numbers.

1 Introduction
Given two rings (R, +, ×) and (S, ◦, ∗), a ring isomorphism from R to S is a bijection f : R → S
such that
f (x + y) = f (x) ◦ f (y)
f (x × y) = f (x) ∗ f (y)
Isomorphic rings are essentially identical except for how the elements are named. That is, their
structure is identical, but with the elements in one structure re-labeled as the other’s.

2 Proof
2.1 The Vortrix Product
Given two vectors ~a, ~b ∈ R2 , where ~a = (a1 , a2 ) and ~b = (b1 , b2 ), the vortrix product is defined
by
 
a b + a2 b2 a2 b1 − a1 b2
~a ~b = 1 1
a1 b2 − a2 b1 a1 b1 + a2 b2
Thus : R2 × R2 → M2 (R).
Here we only consider the 2-d vortrix product. Distinti uses the standard matrix product
definition to compute products involving elements of M2 (R) and elements of R2 (and the transpose
of those elements). While products of elements in M2 (R) are well-defined, products between (row-
or column-) vectors and elements of M2 (R) are not always defined. For example,

~a ~b ~c
Note that these vectors are all of dimension 2 × 1, i.e. column vectors.
If we carry out ~b ~c first, we get a 2 × 2 matrix. But now we cannot use standard matrix
multiplication to carry out ~a(~b ~c), since a (2 × 1)(2 × 2) product is undefined.
If we carry out (~a ~b)~c, we end up with a (2 × 2)(2 × 1) product, which yields a 2 × 1 vector.
Thus appears to be left-associative. Distinti appears to be unconcerned with this issue, and
is comfortable writing down expressions of the form C(AB), in which the various operations are
unclear (and perhaps incompatible or undefined).
An important fact is that the range of this operation contains only invertible matrices  (except

~ ~ ~ u −v
when ~a or b are 0). To see this, let u = a1 b1 + a2 b2 and v = a1 b2 − a2 b1 . Then ~a b = .
v u
Observe that det(~a ~b) = u2 + v 2 , which is equal to 0 if and only if u = v = 0. This condition
is met when the following equations are simultaneously satisfied:

a1 b1 = −a2 b2

1
a1 b2 = a2 b1
Suppose ~a 6= ~0. We will show ~b = ~0.
Since ~a 6= ~0, either a1 6= 0 or a2 6= 0. Suppose a1 6= 0. Then b2 = aa12 b1 and b1 = − aa12 b2 . Letting
c = aa12 , we have b2 = cb1 and b1 = −cb2 .
So b2 = −c2 b2 , and thus b2 + c2 b2 = b2 (1 + c2 ) = 0, so b2 = 0 and therefore b1 = −cb2 =
−c(0) = 0. QED.

2.2 C andM2 (R)


The preceding proof was long and boring, with a lot of notational annoyance for a somewhat
simple result. After establishing the existence of an isomorphism, the preceding proof becomes
much simpler.
Thus motivated, let’s do the heavy lifting to demonstrate an isomorphism. Consider the map
ϕ : C → M2 (R) given by
 
a −b
ϕ : a + bi 7→
b a
This is an injective ring homomorphism, which means C is isomorphic to Im ϕ.
To show injectivity, we must show ϕ(z)  ⇒
 = ϕ(w)  z = w.
 Suppose therefore that ϕ(z) = ϕ(w)
a −b c −d
and let z = a + bi, w = c + di. Then = . This shows a = c, b = d, and thus
b a d c
z = w.
To show that ϕ is a homomorphism, we must show it meets two criteria: one, that ϕ(z + w) =
ϕ(z) + ϕ(w); two, that ϕ(zw) = ϕ(z)ϕ(w). Observe that the operations on the right-hand side of
these equations are taking place in M2 (R).    
a + c −(b + d) a + c −b + −d
First, ϕ(z+w) = ϕ(a+bi+c+di) = ϕ ((a + c) + (b + d)i) = = =
b+d a+c b+d a+c
   
a −b c −d
+ = ϕ(z) + ϕ(w).
b a d c
 
ac − bd −(ad + bc)
Second, ϕ(zw) = ϕ((a + bi)(c + di)) = ϕ((ac − bd) + (ad + bc)i) = .
ad + bc ac − bd
    
a −b c −d ac − bd −ad + −bc
Now consider ϕ(z)ϕ(w) = = . By inspection, this is
b a d c bc + ad −bd + ac
equal to ϕ(zw).
Thus ϕ is an isomorphism, and C ∼ = Im ϕ.

2.2.1 Ring isomorphism Lemmas


In this section, I’ll restate some basic isomorphism facts. They’re mostly intuitive, and their proofs
are simple. Many of them are direct extensions of similar lemmas about group isomorphisms.
I’ll omit multiplication symbols for the sake of brevity. Which multiplication operation is being
used should be clear from the context (i.e., the rings in which the elements reside).

Lemma 1: If φ : R → S is a ring isomorphism, then φ−1 : S → R is a ring isomorphism.

Proof: φ−1 is clearly a bijection. We must show φ−1 (xy) = φ−1 (x)φ−1 (y) for all x, y ∈ S.
Let a = φ−1 (x) ∈ R, b = φ−1 (y) ∈ R. Then φ(ab) = φ(a)φ(b) = xy, i.e. ab = φ−1 (xy). But
ab = φ−1 (x)φ−1 (y). Thus φ−1 is an isomorphism. QED.

Lemma 2: If R ∼
= S are rings, then R commutative ⇒ S commutative.

Proof: Let a, b ∈ S. Let ψ : R → S be an isomorphism (at least one exists, by hypothesis). Let
x = φ−1 (a) ∈ R, y = φ−1 (b) ∈ R. Since R is commutative, xy = yx. Thus φ(xy) = φ(yx). But
φ(xy) = φ(x)φ(y) = ab and φ(yx) = φ(y)φ(x) = ba and thus ab = ba. QED.

Note that in the above proof, we only used the fact that x and y commuted. We can generalize
the commutativity-sharing in the following lemma:

2
Lemmma 4: If R ∼
= S, then xy = yx ⇐⇒ φ(x)φ(y) = φ(y)φ(x).

The forward direction follows directly from the definition of a ring isomorphism. The backwards
direction can be shown:

Suppose φ(x)φ(y) = φ(y)φ(x). Then φ(xy) = φ(yx). Since φ is a bijection, it is one-to-one;


thus xy = yx.

Lemma 3: If R ∼
= S, then the units of R are units in S.

Proof: Let φ : R → S be an isomorphism and let r ∈ R be a unit. Then there exists an


r−1 ∈ R with rr−1 = r−1 r = 1R . Let s = φr and s−1 = φ(r−1 ). Since r and r−1 commute, their
(ismorphic) images s and s−1 also commute. QED.

2.3 Vortrix Isomorphism


To show that Distinti’s vortrix product is isomorphic to the complex numbers, we’ll use the
isomorphism ϕ : C → M2 (R) defined  above.  
0 −1 r 0
Distinti claims that = i and, for any r ∈ R, r = . We would expect a complex
1 0 0 r
     
a 0 0 −b a −b
number a + bi to be given as + = . This is precisely the embedding of C
0 a b 0 b a
into M2 (R) we’d get by applying ϕ.
The vortrix product was originally defined as a function from R2 × R2 → M2 (R), but if we
identify a + bi with (a, b), it can just as easily be viewed as a function from C × C → M2 (R)

a + bi c + di = (a, b) (c, d)
Viewing a + bi as a vector (a, b) in the complex plane, we can make sense of Distini’s claims
about vector rotations and so forth.
Viewed in this fashion, we can decompose the vortrix product as follows:
 
~ a1 b1 + a2 b2 a2 b1 − a1 b2
~a b =
a1 b2 − a2 b1 a1 b1 + a2 b2
  
a1 a2 b1 −b2
=
−a2 a1 b2 b1
 T  
a −a2 b1 −b2
= 1
a2 a1 b2 b1
= ϕ(a1 + a2 i)ϕ(b1 + b2 i)
= ϕ(z)ϕ(w)
= ϕ(zw)
Note that ϕ(z)T = ϕ(z). This invalidates Distinti’s claim that conjugation doesn’t result in a
transposition for matrices encoding complex numbers.
The vortrix product is the projection into M2 (R) of conjugate multiplication in C. The fact
that ~a ~a = ||~a||2 follows from zz = ||z||2 .
The fact that vortrix products rotate vectors comes from the fact that the angle of zw, also
called the argument of zw and written arg(zw), is given by arg(z) + arg(w), along with the fact
that arg(z) = − arg(z). This is mostly easily seen by writing z and w in polar form, z = reiθ and
w = seiω , where θ = arg(z). There’s a bit of added complexity here due to the fact that there isn’t
a unique value for arg(z), since (e.g.) i is at angle π, 2π + π, 2π ∗ 2 + π. Adding 2π to an angle
yields another angle with the same terminus. Strictly speaking, arg(z) is a set, with a unique value
Arg(z) such that 0 ≤ Arg(z) < 2π. This is called the principal argument of z.
The fact that the transpose of ~a ~b reverses the angles follows from the fact that ϕ(zw)T =
ϕ(zw) = ϕ(zw).
We can now prove that every non-zero matrix in the image of is invertible by observing that
every non-zero complex number is invertible.

3
2.4 An Alternative Vortrix Product
Distinti has offered an alternative vortrix product definition (which he describes as a different "sign
convention"), given by:
 
~ a1 b1 − a2 b2 −a1 b2 − a2 b1
~a ⊗ b =
a1 b2 + a2 b1 a1 b1 − a2 b2
This is an even more straightforward embedding of C into M2 (R). Viewing ~a as a1 + a2 i:
 
~ a1 b1 − a2 b2 −a1 b2 − a2 b1
~a ⊗ b =
a1 b2 + a2 b1 a1 b1 − a2 b2
  
a1 −a2 b1 −b2
=
a2 a1 b2 b1

= ϕ(~a)ϕ(~b) = ϕ(~a~b)

3 Commutative Diagrams
The diagrams in this section are commutative, meaning that every path with the same starting
point and ending point are identical. Specifically, these diagrams show that vortrix products are
projects of standard products in C.
Denoting φ : (z, w) 7→ zw, what I refer to somewhat idiosyncratically as the conjugate product,
we have the following commutative diagram:
φ
C2 C
ϕ

M2 (R)

This expresses = ϕ◦φ. Note that φ is surjective, since φ : (1, z) 7→ z. Thus Im = Im ϕ ∼


= C.
Denoting η : (z, w) 7→ zw, we have the following commutative diagram:
η
C2 C
ϕ

M2 (R)

Again, η is surjective, since η : (1, z) 7→ z, so Im ⊗ = Im ϕ ∼


= C.

4 Conclusions
Since Distini’s vortrix products can both be expressed as compositions of well-known functions
that can be expressed in "legacy" linear algebra, his claims to have created a superior algebraic
structure are unconvincing.
Interestingly, in GL2 (R) (and therefore also in M2 (R)), there are actually infinitely many square
 2  
a b −1 0
roots of −1. For any a, b ∈ R with b 6= 0, −a2 −1 = . This matrix is invertible
b −a 0 −1
 
a b 2
since det −a2 −1 = −a2 − b −ab −1 = −a2 + a2 + 1 = 1. If we decide to restrict ourselves to
b −a
the subspace of M2 (R) defined by the range of his vector products in order to avoid this problem,
we have only two square roots of −1. But this subspace is isomorphic to C, since it is precisely
Im ϕ.

You might also like