Version:	1.5-3
Title:	Octonions and Quaternions
LazyData:	TRUE
Description:	Quaternions and Octonions are four- and eight- dimensional extensions of the complex numbers. They are normed division algebras over the real numbers and find applications in spatial rotations (quaternions), and string theory and relativity (octonions). The quaternions are noncommutative and the octonions nonassociative. See the package vignette for more details.
Maintainer:	Robin K. S. Hankin <hankin.robin@gmail.com>
License:	GPL-2
Depends:	methods, R (≥ 3.5.0)
Suggests:	testthat,knitr,rmarkdown,covr
VignetteBuilder:	knitr
Imports:	emulator, Matrix, freealg (≥ 1.0-4), mathjaxr
URL:	https://github.com/RobinHankin/onion
BugReports:	https://github.com/RobinHankin/onion/issues
RdMacros:	mathjaxr
NeedsCompilation:	yes
Packaged:	2024-03-29 10:01:31 UTC; rhankin
Author:	Robin K. S. Hankin [aut, cre]
Repository:	CRAN
Date/Publication:	2024-03-29 11:40:02 UTC

Octonions and Quaternions

Description

Quaternions and Octonions are four- and eight- dimensional extensions of the complex numbers. They are normed division algebras over the real numbers and find applications in spatial rotations (quaternions), and string theory and relativity (octonions). The quaternions are noncommutative and the octonions nonassociative. See the package vignette for more details.

Details

Index of help topics:

adjoint                 The adjoint map
Arith                   Methods for Function Arith in package Onion
biggest                 Returns the biggest type of a set of onions
bind                    Binding of onionmats
bunny                   The Stanford Bunny
c                       Concatenation
Compare-methods         Methods for compare S4 group
condense                Condense an onionic vector into a short form
cumsum                  Cumulative sums and products of onions
dot-class               Class "dot"
drop                    Drop zero imaginary parts of an onionic vector
i                       Extract or Replace Parts of onions or glubs
length                  Length of an octonionic vector
log                     Various logarithmic and circular functions for
                        onions
logic.onion             Logical operations on onions
names.onion             Names of an onionic vector
O1                      Unit onions
onion                   Basic onion functions
onion-class             Class "onion"
onion-package           Octonions and Quaternions
onionmat                Onionic matrices
orthogonal              Orthogonal matrix equivalents
p3d                     Three dimensional plotting
plot                    Plot onions
prods                   Various products of two onions
Re                      Complex functionality for onions
rep                     Replicate elements of onionic vectors
roct                    Random onionic vectors
rotate                  Rotates 3D vectors using quaternions
round                   Rounding of onions
seq                     seq method for onions
show                    Print method for onions
sum                     Various summary statistics for onions
threeform               Various non-field diagnostics
zapsmall                Concatenation

There are precisely four normed division algebras over the reals: the reals themselves, the complex numbers, the quaternions, and the octonions. The R system is well equipped to deal with the first two: the onion package provides some functionality for the third and fourth.

Author(s)

Robin K. S. Hankin [aut, cre] (<https://orcid.org/0000-0001-5982-0415>)

Maintainer: Robin K. S. Hankin <hankin.robin@gmail.com>

References

R. K. S. Hankin 2006. “Normed division algebras in R: introducing the onion package”. R News, Volume 6, number 2

Examples

rquat(10)   # random quaternions

Ok + (Oi + Ojl)/(Oj-Oil)  # basic octonions

x <- roct(10)
y <- roct(10)
z <- roct(10)

x*(y*z) - (x*y)*z   # nonassociative!

Methods for Function Arith in package Onion

Description

Methods for Arithmetic functions for onions: +, -, *, /, ^

Usage

onion_negative(z)
onion_inverse(z)
onion_arith_onion(e1,e2)
onion_arith_numeric(e1,e2)
numeric_arith_onion(e1,e2)
harmonize_oo(a,b)
harmonize_on(a,b)
onion_plus_onion(a,b)
onion_plus_numeric(a,b)
onion_prod_onion(e1,e2)
octonion_prod_octonion(o1,o2)
quaternion_prod_quaternion(q1,q2)
onion_prod_numeric(a,b)
onion_power_singleinteger(o,n)
onion_power_numeric(o,p)

Arguments

Details

The package implements the Arith group of S4 generics so that idiom like A + B*C works as expected with onions.

Functions like onion_inverse() and onion_plus_onion() are low-level helper functions. The only really interesting operation is multiplication; functions octonion_prod_octonion() and quaternion_prod_quaternion() dispatch to C.

Names are implemented and the rules are inherited (via harmonize_oo() and harmonize_on()) from rbind().

Value

generally return an onion

Note

Previous versions of the package included the option to use native R rather than the faster compiled C code used here. But this was very slow and is now discontinued.

Author(s)

Robin K. S. Hankin

Examples

a <- rquat()
b <- rquat()
a
Re(a)
j(a) <- 0.2
a*b
b*a  # quaternions are noncommutative


x <- as.octonion(matrix(rnorm(40),nrow=8))
y <- roct()
z <- roct()


x*(y*z) - (x*y)*z  # octonions are nonassociative [use associator()]

Methods for compare S4 group

Description

Methods for comparison (equal to, greater than, etc) of onions. Only equality makes sense.

Value

Return a boolean

Examples

# roct() > 0 # meaningless and returns an error



x <- as.octonion(matrix(sample(0:1,800,TRUE,p=c(9,1)),nrow=8))
y <- as.octonion(matrix(sample(0:1,800,TRUE,p=c(9,1)),nrow=8))
x==y

matrix(as.quaternion(100+1:12),3,4) == 102



  

Complex functionality for onions

Description

Functionality in the Complex group.

The norm Norm(O) of onion O is the product of O with its conjugate: |O|=OO^* but a more efficient numerical method is used (see dotprod()).

The Mod Mod(O) of onion O is the square root of its norm.

The sign of onion O is the onion with the same direction as O but with unit Norm: sign(O)=O/Mod(O).

Function Im() sets the real component of its argument to zero and returns that; Conj() flips the sign of its argument's non-real components. Function Re() returns the real component (first row) of its argument as a numeric vector. If x is an onion, then x == Re(x) + Im(x).

Usage

## S4 method for signature 'onion'
Re(z)
## S4 method for signature 'onion'
Im(z)
Re(z) <- value
Im(x) <- value
## S4 method for signature 'onion'
Conj(z)
## S4 method for signature 'onion'
Mod(z)
onion_abs(x)
onion_conjugate(z)
## S4 method for signature 'onion'
sign(x)

Arguments

Value

All functions documented here return a numeric vector or matrix of the same dimensions as their argument, apart from functions Im() and Conj(), which return an object of the same class as its argument.

Note

If x is a numeric vector and y an onion, one might expect typing x[1] <- y to result in x being a onion. This is impossible, according to John Chambers.

Extract and set methods for components such as i,j,k are documented at Extract.Rd

Compare clifford::Conj(), which is more complicated.

Author(s)

Robin K. S. Hankin

See Also

Extract

Examples

a <- rquat()
Re(a)
Re(a) <- j(a)

Im(a)

b <- romat()

A <- romat()
Im(A) <- Im(A)*10

Extract or Replace Parts of onions or glubs

Description

Methods for "[" and "[<-", i.e., extraction or subsetting of onions.

Usage

## S4 method for signature 'onion'
i(z)
## S4 method for signature 'onion'
j(z)
## S4 method for signature 'onion'
k(z)
## S4 method for signature 'octonion'
l(z)
## S4 method for signature 'octonion'
il(z)
## S4 method for signature 'octonion'
jl(z)
## S4 method for signature 'octonion'
kl(z)
## S4 method for signature 'onionmat'
i(z)
## S4 method for signature 'onionmat'
j(z)
## S4 method for signature 'onionmat'
k(z)
## S4 method for signature 'onionmat'
il(z)
## S4 method for signature 'onionmat'
jl(z)
## S4 method for signature 'onionmat'
kl(z)
i(x) <- value
j(x) <- value
k(x) <- value
l(x) <- value
il(x) <- value
jl(x) <- value
kl(x) <- value

Arguments

Value

Extraction and methods return an onion or onionmat. Replacement methods return an object of the same class as x.

Note

If x is a numeric vector and y a onion, one might expect typing x[1] <- y to result in x being a onion. This is impossible, according to John Chambers.

Author(s)

Robin K. S. Hankin

Examples

a <- roct(9)
il(a)
Re(a) <- 1:9

j(a) <- l(a)
a

Logical operations on onions

Description

Logical operations on onions are not supported

Usage

onion_logic(e1,e2)

Arguments

Value

none

Note

Carrying out logical operations in this group will report an error. Negation, “!”, is not part of this group.

Author(s)

Robin K. S. Hankin

Examples

# roct() & roct()   # reports an error

Various logarithmic and circular functions for onions

Description

Various elementary functions for onions

Usage

onion_log(x,base=exp(1))
onion_exp(x)
onion_sign(x)
onion_sqrt(x)
onion_cosh(x)
onion_sinh(x)
onion_acos(x)
onion_acosh(x)
onion_asin(x)
onion_asinh(x)
onion_atan(x)
onion_atanh(x)
onion_cos(x)
onion_sin(x)
onion_tan(x)
onion_tanh(x)
onion_cos(x)
onion_sin(x)
onion_tan(x)
onion_tanh(x)

Arguments

Details

Standard math stuff. I am not convinced that the trig functions (sin() etc) have any value.

Author(s)

Robin K. S. Hankin

Examples

x <- roct()
exp(x+x) - exp(x)*exp(x) # zero to numerical precision

jj <- exp(log(x)/2)      # use sqrt() here
jj*jj-x                  #  also small


y <- roct()
exp(x+y) - exp(x)*exp(y) # some rules do not operate for onions



max(Mod(c(sin(asin(x))-x,asin(sin(x))-x)))   # zero to numerical precision

Unit onions

Description

Each of the eight unit quaternions and octonions

Usage

H1
Hi
Hj
Hk
H0
Him
Hall
O1
Oi
Oj
Ok
Ol
Oil
Ojl
O0
Oim
Oall

Format

Each one is an onionic vector of length one.

Details

Try Hi (=quaternion(i=1)) to get the pattern for the first four. The next ones are the zero quaternion, the pure imaginary quaternion with all components 1, and the quaternion with all components 1. The ones beginning with “O” follow a similar pattern.

These are just variables that may be overwritten and thus resemble T and F whose value may be changed.

Value

A length-one onion, either a quaternion or an octonion

Examples

Oall
seq_onion(from=O1,to=Oil,len=6)

stopifnot(Hj*Hk ==  Hi)
stopifnot(Okl*Oil == -Oj )  # See tests/test_aaa.R for the full set

The adjoint map

Description

The adjoint \(\mathrm{ad}_X\) of \(X\) is a map from a Lie group \(G\) to the endomorphism group of \(G\) defined by

Usage

ad(x)

Arguments

Details

Here for completeness really.

Author(s)

Robin K. S. Hankin

Examples

x <- rquat()
y <- rquat()

f <- ad(x)
f(y)

f(f(y)) # [x,[x,y]]

Returns the biggest type of a set of onions

Description

Returns the biggest type of a set of onions; useful for “promoting” a set of onions to the most general type.

Usage

biggest(...)

Arguments

Details

If any argument passed to biggest() is an octonion, then return the string “octonion”. Failing that, if any argument is a quaternion, return the string “quaternion”, and failing that, return “scalar”.

Value

Character string representing the type

Author(s)

Robin K. S. Hankin

Examples

biggest(O1,rquat(100),1:4)

Binding of onionmats

Description

Methods for rbind() and cbind() of onionmats. These are implemented by specifying methods for rbind2() and cbind2().

Usage

bind_onion(x,bind,...)
bind_onion_onion(x,y,bind,...)
bind_onion_onionmat(x,y,bind,...)
bind_onionmat_onion(x,y,bind,...)

Arguments

Value

Return onionmats

Author(s)

Robin K. S. Hankin

Examples

rbind(rquat(3),rquat(3))

cbind(diag(5),roct(1))


cbind(matrix(Oil,4,2),matrix(roct(12),4,3))

The Stanford Bunny

Description

A set of 3D points in the shape of a rabbit (the Stanford Bunny)

Usage

data(bunny)

Format

A three column matrix with 35947 rows. Each row is the Cartesian coordinates of a point on the surface of the bunny.

Value

as for format

Source

https://graphics.stanford.edu/data/3Dscanrep/

Examples

data(bunny)
p3d(rotate(bunny,Hk))

Concatenation

Description

Combines its arguments to form a single onion.

Usage

c_onionpair(x,y)
## S4 method for signature 'onion'
c(x,...)

Arguments

Details

Returns an onion of the same type as its arguments. Names are inherited from the behaviour of cbind(), not c().

Value

An onion

Note

The method is not perfect; it will not, for example, coerce its arguments to the biggest() type, so c(rquat(),roct()) will fail. You will have to coerce the arguments by hand.

Dispatch is based on the class of the first argument, so c(1,rquat()) will return a list (not an onion), and c(rquat(),1) will fail.

Author(s)

Robin K. S. Hankin

Examples

a <- roct(3)
b <- seq_onion(from=Oil,to=Oj,len=6)
c(a,b)

c(rquat(3),H1,H0,Him)

Condense an onionic vector into a short form

Description

Condense an onion into a string vector showing whether the elements are positive, zero or negative.

Usage

condense(x,as.vector=FALSE)

Arguments

Value

If as.vector is TRUE, return a string vector of the same length as x whose elements are length 4 or 8 strings for quaternions or octonions respectively. If FALSE, return a matrix with these columns.

The characters are “+” for a positive, “-” for a negative, and “0” for a zero, element.

Author(s)

Robin K. S. Hankin

Examples

condense(roct(3))
condense(roct(3),as.vector=TRUE)

Cumulative sums and products of onions

Description

Cumulative sums and products of onions

Usage

onion_cumsum(x)
onion_cumprod(x)

Arguments

Value

An onion

Note

The octonions are nonassociative but cumprod() operates left-associatively, as in ((a[1]*a[2])*a[3])*a[4] etc.

Author(s)

Robin K. S. Hankin

Examples

cumsum(as.quaternion(matrix(runif(20),4,5)))
cumsum(roct(5))

cumprod(rquat(7))

Class “dot”

Description

The dot object is defined so that idiom like .[x,y] returns the commutator, that is, xy-yx or the Lie bracket \([x,y]\). It would have been nice to use [x,y] (that is, without the dot) but although this is syntactically consistent, it cannot be done in R.

The “meat” of the package is:

setClass("dot", slots = c(ignore='numeric'))
`.` <- new("dot")
setMethod("[",signature(x="dot",i="ANY",j="ANY"),function(x,i,j,drop){i*j-j*i})

The package code includes other bits and pieces such as informative error messages for idiom such as .[]. The package defines a matrix method for the dot object. This is because “*” returns (incorrectly, in my view) the elementwise product, not the matrix product.

The Jacobi identity, satisfied by any associative algebra, is

Function ad() returns the adjoint operator. The adjoint vignette provides details and examples of the adjoint operator.

The dot object is generated by running script inst/dot.Rmd, which includes some further discussion and technical documentation, and creates file dot.rda which resides in the data/ directory.

Value

Always returns an object of the same class as xy

Slots

ignore:

Object of class "numeric", just a formal placeholder

Methods

[

signature(x = "dot", i = "ANY", j = "ANY"): ...

[

signature(x = "dot", i = "ANY", j = "missing"): ...

[

signature(x = "dot", i = "function", j = "function"): ...

[

signature(x = "dot", i = "matrix", j = "matrix"): ...

[

signature(x = "dot", i = "missing", j = "ANY"): ...

[

signature(x = "dot", i = "missing", j = "missing"): ...

Author(s)

Robin K. S. Hankin

See Also

adjoint

Examples

x <- rquat()
y <- rquat()
z <- rquat()
.[x,y]


.[x,.[y,z]] + .[y,.[z,x]] + .[z,.[x,y]]  # Jacobi, expanded

Drop zero imaginary parts of an onionic vector

Description

If an onion has zero imaginary part, drop it

Usage

## S4 method for signature 'onion'
drop(x)

Arguments

Details

Generally, “drop” means coercion of an object to a less general type without loss of information. In many contexts, function drop() means to lose redundant information. This is not done by default (doing so would result in unexpected coercions).

Methods are given for onion and onionmat objects.

Author(s)

Robin K. S. Hankin

Examples

a <- rsoct()
a
a-Im(a)
drop(a-Im(a))

Length of an octonionic vector

Description

Get or set the length of onions

Usage

## S4 method for signature 'onion'
length(x)

Arguments

Details

Operates on the columns of the matrix as expected.

Value

integer

Author(s)

Robin K. S. Hankin

Examples

a <- roct(5)
length(a)

Names of an onionic vector

Description

Functions to get or set the names of an onion

Usage

## S4 method for signature 'onion'
names(x)
## S4 method for signature 'onionmat'
rownames(x)
## S4 method for signature 'onionmat'
colnames(x)
## S4 method for signature 'onionmat'
dimnames(x)
## S4 method for signature 'onionmat'
dim(x)

Arguments

Details

Names attributes refers to colnames of the internal matrix, which are retrieved or set using colnames() or colnames<-().

Author(s)

Robin K. S. Hankin

Examples

a <- roct(5)
names(a) <- letters[1:5]

b <- romat()
dimnames(b) <- list(month = month.abb[1:5], location=names(islands)[1:6])

Basic onion functions

Description

Construct, coerce to, test for, and print onions

Usage

octonion(length.out = NULL, Re = 0, i = 0, j = 0, 
    k = 0, l = 0, il = 0, jl = 0, kl = 0)
as.octonion(x, single = FALSE)
is.octonion(x)
quaternion(length.out = NULL,Re = 0, i = 0, j = 0, k = 0)
as.quaternion(x, single = FALSE)
is.quaternion(x)
is.onion(x)
as.onion(x,type,single=FALSE)
quaternion_to_octonion(from)
octonion_to_quaternion(from)
## S4 method for signature 'onion'
as.matrix(x)
## S4 method for signature 'onion'
as.numeric(x)

Arguments

Details

Functions quaternion() and octonion() use standard recycling where possible; rbind() is used.

Functions as.quaternion() and as.octonion() coerce to quaternions and octonions respectively. If given a complex vector, the real and imaginary components are interpreted as Re and i respectively.

The output of type() is accepted as the type argument of function as.onion(); thus as.onion(out,type=type(x)) works as expected.

Value

Generally return onions

Note

An onion is any algebra (over the reals) created by an iterated Cayley-Dickson process. Examples include quaternions, octonions, and sedenions. There does not appear to be a standard generic term for such objects (I have seen n-ion, anion and others. But “onion” is pronouncable and a bona fide English word).

Creating further onions—such as the sedenions—is intended to be straightforward.

There is a nice example of the onion package in use in the permutations package, under cayley.Rd. This also shows the quaternion group Q8, but from a different perspective.

Author(s)

Robin K. S. Hankin

Examples

x <- octonion(Re=1,il=1:3)
x
kl(x) <- 100
x

as.quaternion(diag(4))


# Cayley table for the quaternion group Q8:
a <- c(H1,-H1,Hi,-Hi,Hj,-Hj,Hk,-Hk)
names(a) <- c("+1","-1","+i","-i","+j","-j","+k","-k")

f <- Vectorize(function(x,y){names(a)[a==a[x]*a[y]]})
X <- noquote(outer(1:8,1:8, f))
rownames(X) <- names(a)
colnames(X) <- names(a)
X

Class “onion”

Description

The formal S4 class for onion and onionmat objects

Objects from the Class

Class onion is a virtual S4 class extending classes quaternion and octonion. In package documentation, “onion” means an R object that behaves as a vector of quaternions or octonions, stored as a four- or eight- row numeric matrix.

Class onionmat is the S4 class for matrices whose elements are quaternions or octonions. An onionmat is stored as a two-element list, the first being an onion and the second an integer matrix which holds structural matrix attributes such as dimensions and dimnames. Most standard arithmetic R idiom for matrices should work for onionmats.

Class index is taken from the excellent Matrix package and is a setClassUnion() of classes numeric, logical, and character, which mean that it is an arity-one matrix index.

Author(s)

Robin K. S. Hankin

Examples

as.octonion(1:8,single=TRUE)
as.quaternion(matrix(runif(20),nrow=4))

H <- matrix(rquat(21),3,7)
dimnames(H) <- list(foo=letters[1:3],bar=state.abb[1:7])

i(H) <- 0.1

I <- matrix(rquat(14),7,2)
dimnames(I) <- list(foo=state.abb[1:7],baz=LETTERS[1:2])
H %*% I

Onionic matrices

Description

Simple functionality for quaternionic and octonionic matrices, intended for use in the jordan package. Use idiom like matrix(Him,4,5) or matrix(roct(6),2,3) to create an onionmat object, a matrix of onions.

The package is intended to match base R's matrix functionality in the sense that standard R idiom just goes through for onionic matrices. Determinants are not well-defined for quaternionic or octonionic matrices, and matrix inverses are not implemented.

Usage

newonionmat(d, M)
onionmat(data = NA, nrow = 1, ncol = 1, byrow = FALSE, dimnames = NULL)
as.onionmat(x)
is.onionmat(x)
onionmat_negative(e1)
onionmat_inverse(e1)
onionmat_prod_onionmat(e1,e2)
onionmat_power_onionmat(...)
onionmat_prod_single(x,y)
onionmat_power_single(e1,e2)
onionmat_plus_onionmat(e1,e2)
matrix_arith_onion(e1,e2)
onion_arith_matrix(e1,e2)
matrix_plus_onion(e1,e2)
matrix_prod_onion(e1,e2)
## S4 method for signature 'onionmat,onionmat'
cprod(x,y)
## S4 method for signature 'onionmat,missing'
cprod(x,y)
## S4 method for signature 'onionmat,ANY'
cprod(x,y)
## S4 method for signature 'ANY,ANY'
cprod(x,y)
## S4 method for signature 'onion,missing'
cprod(x,y)
## S4 method for signature 'onion,onion'
cprod(x,y)
## S4 method for signature 'onion,onionmat'
cprod(x,y)
## S4 method for signature 'onionmat,onion'
cprod(x,y)
## S4 method for signature 'onionmat,onionmat'
tcprod(x,y)
## S4 method for signature 'onionmat,missing'
tcprod(x,y)
## S4 method for signature 'onionmat,ANY'
tcprod(x,y)
## S4 method for signature 'ANY,ANY'
tcprod(x,y)
## S4 method for signature 'onion,missing'
cprod(x,y)
## S4 method for signature 'onion,onion'
cprod(x,y)
## S4 method for signature 'onion,onionmat'
cprod(x,y)
## S4 method for signature 'onionmat,onion'
cprod(x,y)
## S4 method for signature 'onionmat'
t(x)
## S4 method for signature 'onion'
t(x)
## S4 method for signature 'onionmat'
ht(x)
## S4 method for signature 'onion'
ht(x)
nrow(x)
ncol(x)
herm_onion_mat(real_diagonal, onions)
onionmat_complex(z)
onionmat_conjugate(z)
onionmat_imag(z)
onionmat_re(z)
onionmat_mod(z)
onionmat_matrixprod_onionmat(x,y)
onion_matrixprod_onionmat(x,y)
onionmat_matrixprod_numeric(x,y)
onionmat_matrixprod_onion(x,y)

Arguments

Details

An object of class onionmat is a two-element list, the first of which is an onion, and the second an index matrix of integers used for tracking attributes such as dimensions and dimnames. This device makes the extraction and replacement methods easy. Use getM() to access the index matrix and getd() to access the onionic vector.

The S4 method for matrix() simply dispatches to onionmat(), which is a drop-in replacement for matrix().

Function drop() has a method for onionmat objects.

Function newonionmat() is lower-level: it also creates onionmat objects, but takes two arguments: an onion and a matrix; the matrix argument can be used to specify additional attributes via attr(), but this ability is not currently used in the package.

Functions such as onionmat_plus_onionmat() are low-level helper functions, not really designed for the end-user.

Vignette onionmat shows some use-cases.

The print method for onionmat objects is sensitive to option show_onionmats_in_place. If TRUE, it prints the matrix elements in-place, using onion_to_string(). It works best when option show_onions_compactly is effective.

Author(s)

Robin K. S. Hankin

Examples

matrix(rquat(28),4,7)

M <- onionmat(rquat(10),2,5)
cprod(M) 

Re(M)
Re(M) <- 0.3

romat() %*% rquat(6)

a <- rsomat()
a            # default
options("show_onionmats_in_place" = TRUE)
a
options("show_onionmats_in_place" = FALSE) # restore default

Orthogonal matrix equivalents

Description

Convert a quaternion to and from an equivalent orthogonal matrix

Usage

matrix2quaternion(M)
as.orthogonal(Q)

Arguments

Value

Function matrix2quaternion() returns a quaternion.

Function as.orthogonal() returns either a 3\times 3 matrix or a 3\times3\times n array of orthogonal matrices

Note

Function matrix2quaternion() is low-level; use as.quaternion() to convert arrays.

Author(s)

Robin K. S. Hankin

See Also

rotate

Examples

as.orthogonal(rquat(1))

o <- function(w){diag(3)-2*outer(w,w)/sum(w^2)}  # Householder
matrix2quaternion(o(1:3))   # Booorrrriiinnnggg
matrix2quaternion(o(1:3) %*% o(3:1))

Q <- rquat(7)
Q <- Q/abs(Q)
as.quaternion(as.orthogonal(Q))   # +/- Q


A <- replicate(7,o(rnorm(3)) %*% o(rnorm(3)))
max(abs(as.orthogonal(as.quaternion(A))-A))

Three dimensional plotting

Description

Three dimensional plotting of points. Produces a nice-looking 3D scatterplot with greying out of further points giving a visual depth cue

Usage

p3d(x, y, z, xlim = NULL, ylim = NULL, zlim = NULL, d0 = 0.2, h = 1, ...)

Arguments

Value

Value returned is that given by function trans3d().

Author(s)

Robin K. S. Hankin

See Also

bunny

Examples

data(bunny)
p3d(bunny,theta=3,phi=104,box=FALSE)

Plot onions

Description

Plotting method for onionic vectors

Usage

## S4 method for signature 'onion'
plot(x,y, ...)

Arguments

Details

The function is plot(Re(x), Mod(Im(x)), ...), and thus behaves similarly to plot() when called with a complex vector.

Value

Called for its side-effect of plotting a diagram

Author(s)

Robin K. S. Hankin

Examples

plot(roct(30))

Various products of two onions

Description

Returns various inner and outer products of two onionic vectors.

Usage

x %<*>% y
x %>*<% y
x %<.>% y
x %>.<% y
x %.% y
onion_g_even(x,y)
onion_g_odd (x,y)
onion_e_even(x,y)
onion_e_odd (x,y)
dotprod(x,y)

Arguments

Details

This page documents an attempt at a consistent notation for onionic products. The default product for onions (viz “*”) is sometimes known as the “Grassman product”. There is another product known as the Euclidean product defined by E(p,q)=p'q where x' is the conjugate of x.

Each of these products separates into an “even” and an “odd” part, here denoted by functions g_even() and g_odd() for the Grassman product, and e_even() and e_odd() for the Euclidean product. These are defined as follows:

• g_even(x,y)=(xy+yx)/2

• g_odd(x,y)=(xy-yx)/2

• e_even(x,y)=(x'y+y'x)/2

• e_odd(x,y)=(x'y-y'x)/2

These functions have an equivalent binary operator.

The Grassman operators have a “*”; they are “%<*>%” for the even Grassman product and “%>*<%” for the odd product.

The Euclidean operators have a “.”; they are “%<.>%” for the even Euclidean product and “%>.<%” for the odd product.

Function dotprod() returns the Euclidean even product of two onionic vectors. That is, if x and y are eight-element vectors of the components of two onions, return sum(x*y).

Note that the returned value is a numeric vector (compare %<.>%, e.even(), which return onionic vectors with zero imaginary part).

There is no binary operator for the ordinary Euclidean product (it seems to be rarely needed in practice). For Conj(x)*x, Norm(x) is much more efficient and accurate.

Function prod() is documented at Summary.Rd.

Note

Frankly if you find yourself using these operators you might be better off using the clifford package, which has an extensive and consistent suite of product operators.

Author(s)

Robin K. S. Hankin

Examples

Oj %<.>% Oall

Replicate elements of onionic vectors

Description

Replicate elements of onionic vectors

Usage

## S4 method for signature 'onion'
rep(x, ...)

Arguments

Author(s)

Robin K. S. Hankin

Examples

a <- roct(3)
rep(a,2) + a[1]
rep(a,each=2)
rep(a,length.out=5)

Random onionic vectors

Description

Random quaternion or octonion vectors and matrices

Usage

rquat(n=5)
roct(n=5)
rsquat(n=11,s=12)
rsoct(n=11,s=12)
romat(type="quaternion", nrow=5, ncol=6, ...)
rsomat(type="quaternion", nrow=5, ncol=6, ...)

Arguments

Details

Function rquat() returns a quaternionic vector, roct() returns an octonionic vector, and romat() a quaternionic matrix.

Functions rquat() and roct() give a quick “get you going” random onion to play with. Function romat() gives a simple onionmat, although arguably matrix(roct(4),2,2) is as convenient.

The “sparse” functions rsquat() and rsoct() and rsomat() return onions that have many zero entries; non-zero entries are small integers. They showcase the print method for the case when show_onions_compactly is set.

Author(s)

Robin K. S. Hankin

References

K. Shoemake 1992. “Uniform random rotations”. In D. Kirk, editor, Graphics Gems III pages 129-130. Academic, New York.

Examples

rquat(3)
roct(3)
plot(roct(30))

romat()


rsquat()
rsoct()

Rotates 3D vectors using quaternions

Description

Rotates a three-column matrix whose rows are vectors in 3D space, using quaternions

Usage

rotate(x, H)

Arguments

Value

Returns a matrix of the same size as x

Author(s)

Robin K. S. Hankin

See Also

orthogonal

Examples

data(bunny)
par(mfrow=c(2,2))
par(mai=rep(0,4))
p3d(rotate(bunny,Hi),box=FALSE)
p3d(rotate(bunny,H1-Hi+Hj),box=FALSE)
p3d(rotate(bunny,Hk),box=FALSE)
p3d(rotate(bunny,Hall),box=FALSE)

o <- function(w){diag(3)-2*outer(w,w)/sum(w^2)}  # Householder
O <- o(1:3) %*% o(3:1)

rotate(bunny,as.quaternion(O))
bunny %*% t(O)    # should be the same; note transpose

Rounding of onions

Description

Round elements of an onion

Usage

   ## S4 method for signature 'onion'
round(x,digits=0)
   ## S4 method for signature 'onionmat'
round(x,digits=0)

Arguments

Details

For onions, coerce to a matrix, round, then coerce back to an onion. For onionmats, coerce to an onion, round, then coerce back to an onionmat.

Value

Return an onion

Author(s)

Robin K. S. Hankin

Examples

round(rquat()*100)
round(rquat()*100,3)

seq method for onions

Description

Rough equivalent of seq() for onions.

Usage

seq_onion(from=1,to=1,by=((to-from)/(length.out-1)),length.out=NULL,slerp=FALSE, ...)

Arguments

Author(s)

Robin K. S. Hankin

Examples

seq(from=O1,to=Oil,length.out=6)
seq(from=H1,to=(Hi+Hj)/2,len=10,slerp=TRUE)

Print method for onions

Description

Show methods for onions

Usage

## S4 method for signature 'onion'
show(object)
onion_show(x,
     comp = getOption("show_onions_compactly"),
     h    = getOption("show_onions_horizontally")
)
comp_names(x)

Arguments

Details

Default behaviour is to print by rows. To print by columns, set option show_onions_horizontally to TRUE:

options("show_onions_horizontally" = TRUE)

Any non-TRUE value (including NULL and its being unset) will restore the default.

Similarly, to show onions compactly, set option show_onions_compactly to TRUE:

options("show_onions_compactly" = TRUE)

This option works best for simple onions with integer entries (or at least values with few decimal places), and especially if there are many zero entries.

Function onion_show() is a helper function, not really intended for the end-user.

The “names” of the components of an onion (viz Re, i, j, k for quaternions and Re, i, j, k, l,il,jl,kl for octonions) are given by function comp_names() which takes either a character string or an onion.

Note

The print method for onionmat objects is also sensitive to these options.

Author(s)

Robin K. S. Hankin

Examples

x <- roct(15)
x  #default

options("show_onions_horizontally" = TRUE)
roct(4) 

options("show_onions_horizontally" = FALSE)  # restore default

options("show_onions_compactly" = TRUE)
x <- as.quaternion(matrix(sample(c(0,0,0,-1,1),80,replace=TRUE),4,20))
options("show_onions_compactly" = FALSE) # restore default

Various summary statistics for onions

Description

Various summary statistics for onions

Usage

onion_allsum(x)
## S4 method for signature 'onion'
sum(x)
## S4 method for signature 'quaternion'
prod(x)
## S4 method for signature 'octonion'
sum(x)
## S4 method for signature 'onionmat'
sum(x)
## S4 method for signature 'octonion'
prod(x)
## S4 method for signature 'onion'
str(object, ...)
str_onion(object, vec.len = 4, ...)
onion_allsum(x)
onionmat_allsum(x)
quaternion_allprod(x)

Arguments

Details

For a onion object, return the sum or product accordingly

Value

Return an onion

Note

Function str() uses functionality from condense().

Author(s)

Robin K. S. Hankin

Examples

sum(roct())
str(roct())

Various non-field diagnostics

Description

Diagnostics of non-field behaviour: threeform, associator, commutator

Usage

threeform(x1, x2, x3)
associator(x1, x2, x3)
commutator(x1, x2)

Arguments

Details

The threeform is defined as Re(x1 * (Conj(x2) * x3) - x3 * (Conj(x2) * x1))/2;

the associator is (x1 * x2) * x3 - x1 * (x2 * x3);

the commutator is x1 * x2 - x2 * x1.

Value

Returns an octonionic vector

Author(s)

Robin K. S. Hankin

See Also

dot

Examples

x <- roct(7) ; y <- roct(7) ; z <- roct(7)
associator(x,y,z)

Concatenation

Description

Zapping small components to zero

Usage

## S4 method for signature 'onion'
zapsmall(x,digits=getOption("digits"))
## S4 method for signature 'onionmat'
zapsmall(x,digits=getOption("digits"))

Arguments

Details

Uses base::zapsmall() to zap small elements to zero.

Value

An onion

Author(s)

Robin K. S. Hankin

Examples

zapsmall(as.octonion(0.01^(1:8),single=TRUE))


a <- roct(7)
x <- a^1/a
x
zapsmall(x)