### Steven De Keninck PRO

Mathematical Experimentalist

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

Steven De Keninck (enkimute)

follow live for smooth animations!

https://slides.com/d/0A4Ehfc/live

```
On Lie groups defining automorphisms that leave invariant fundamental subspaces of geometric algebra.
```

**Dmitry Shirokov**

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

L’application des mêmes idées de dualité peut s’étendre à la Mécanique. En effet, l’élément primitif des corps auquel on applique d’abord les premiers principes de cette science, est, comme dans la Géométrie ancienne, le point mathématique.

Ne sommes-nous pas autorisés à penser, maintenant, qu’en prenant le plan pour l’élément de l’étendue, et non plus le point, on sera conduit à d’autres doctrines, faisant pour ainsi dire une nouvelle science?

**- Michel Chasles, 1875**

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

The application of the same ideas of duality can be extended to Mechanics. Indeed, the primitive element of bodies to which the first principles of this science are applied is, as in ancient Geometry, the mathematical point.

Are we not permitted to think, now, that by taking the plane for the basic element, and no longer the point, we shall be led to other doctrines, making, as it were, a new science?

**- Michel Chasles, 1875**

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

Are we not permitted to think, now, that by taking the plane for the basic element, and no longer the point, we shall be led to other doctrines, making, as it were, a new science?

**- Michel Chasles, 1875**

1. Symmetry Groups

2. Geometry

3. Rigid Body Dynamics

- classic approach
- k-reflections
- geometric gauges
- Mozzi-Chasles

- coordinate agnostic
- dimension agnostic
- metric agnostic

- Forques
- Demo

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

Are we not permitted to think, now, that by taking the plane for the basic element, and no longer the point, we shall be led to other doctrines, making, as it were, a new science?

**- Michel Chasles, 1875**

1. Symmetry Groups

2. Geometry

3. Rigid Body Dynamics

- classic approach
- k-reflections
- geometric gauges
- Mozi-Chasles

- coordinate agnostic
- dimension agnostic
- metric agnostic

- Forques
- Demo

All our examples will be using the Euclidean groups in 2 and 3 dimensions.

Why?

- these are non-simple Lie groups
- the natural choice for intuition

Plane and Simple.

the universally accepted unwritten axiom

AGACSE2021

Are we not permitted to think, now, that by taking the plane for the basic element, and no longer the point, we shall be led to other doctrines, making, as it were, a new science?

**- Michel Chasles, 1875**

1. Symmetry Groups

2. Geometry

3. Rigid Body Dynamics

- classic approach
- k-reflections
- geometric gauges
- Mozi-Chasles

- coordinate agnostic
- dimension agnostic
- metric agnostic

- Forques
- Demo

All our examples will be using the Euclidean groups in 2 and 3 dimensions.

Why?

- these are non-simple Lie groups
- the natural choice for intuition

a vector space

transformations on a vector space

\begin{bmatrix}
x_1 \\ \vdots \\ x_n
\end{bmatrix}

\begin{bmatrix}
m_{11} & \dots & m_{1n} \\ \vdots & & \vdots \\ m_{n1} & \dots & m_{nn}
\end{bmatrix}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

4x4 matrix

6x6 matrix

4x4 matrix

AGACSE2021

1. Symmetry Groups

a vector space

transformations on a vector space

\begin{bmatrix}
x_1 \\ \vdots \\ x_n
\end{bmatrix}

\begin{bmatrix}
x_{11} & \dots & x_{1n} \\ \vdots & & \vdots \\ x_{n1} & \dots & x_{nn}
\end{bmatrix}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

4x4 matrix

6x6 matrix

4x4 matrix

In this model, perfectly reasonable starting point.

Also valid, but why would we?

AGACSE2021

1. Symmetry Groups

a vector space

transformations on a vector space

\begin{bmatrix}
x_1 \\ \vdots \\ x_n
\end{bmatrix}

\begin{bmatrix}
x_{11} & \dots & x_{1n} \\ \vdots & & \vdots \\ x_{n1} & \dots & x_{nn}
\end{bmatrix}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

4x4 matrix

6x6 matrix

4x4 matrix

a graded vector space

x \in \mathbb R_{p,q,r}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

versors/rotors

R \in \mathbb R_{p,q,r}^+

AGACSE2021

1. Symmetry Groups

a vector space

transformations on a vector space

\begin{bmatrix}
x_1 \\ \vdots \\ x_n
\end{bmatrix}

\begin{bmatrix}
x_{11} & \dots & x_{1n} \\ \vdots & & \vdots \\ x_{n1} & \dots & x_{nn}
\end{bmatrix}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

4x4 matrix

6x6 matrix

4x4 matrix

a graded vector space

x \in \mathbb R_{p,q,r}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

versors/rotors

R \in \mathbb R_{p,q,r}^+

AGACSE2021

1. Symmetry Groups | k-reflections

a graded vector space

x \in \mathbb R_{p,q,r}

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

versors/rotors

R \in \mathbb R_{p,q,r}^+

start from Hamilton's observation that all isometries can be generated by composing reflections. Then stick with it.

\(E(2)\) - The Euclidean Group of the plane

# reflections | Isometry |
---|---|

0 | identity |

1 | reflection |

2 | rotation/translation |

3 | glide reflection |

4 ? | → 2 |

AGACSE2021

1. Symmetry Groups | k-reflections

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

\(E(2)\) - The Euclidean Group of the plane

# reflections | Isometry |
---|---|

0 | identity |

1 | reflection |

2 | rotation/translation |

3 | glide reflection |

4 ? | → 2 |

Invariant |
---|

plane |

line |

point / ∞point |

line + ∞point |

→ 2 |

AGACSE2021

1. Symmetry Groups | k-reflections

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

\(E(2)\) - The Euclidean Group of the plane

# reflections | Isometry |
---|---|

0 | identity |

1 | reflection |

2 | rotation/translation |

3 | glide reflection |

Invariant |
---|

plane |

line |

point / ∞point |

line + ∞point |

GA |
---|

scalar |

vector |

bivector |

trivector |

Using Chasles' hyperplanes as vectors in a GA gives us exactly the structure we need.

the reflections give us everything.

Geometric Algebra Sandwich

\( -a b a^{-1}\)

Group Theory Conjugation

\( a b a^{-1}\)

AGACSE2021

1. Symmetry Groups | k-reflections

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

\(E(2)\) - The Euclidean Group of the plane

# reflections | Isometry |
---|---|

0 | identity |

1 | reflection |

2 | rotation/translation |

3 | glide reflection |

Invariant |
---|

plane |

line |

point / ∞point |

line + ∞point |

Geometric Algebra Sandwich

\( -a b a^{-1}\)

GA |
---|

scalar |

vector |

bivector |

trivector |

Using Chasles' hyperplanes as vectors in a GA gives us exactly the structure we need.

the reflections give us everything.

AGACSE2021

1. Symmetry Groups | geometric gauges

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

AGACSE2021

1. Symmetry Groups | geometric gauges

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

AGACSE2021

1. Symmetry Groups | geometric gauges

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

AGACSE2021

1. Symmetry Groups | Mozzi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

AGACSE2021

1. Symmetry Groups | Mozzi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

The Mozzi-Chasles Theorem :

Every rigid body motion can be decomposed into a rotation around followed or preceded by a a translation along a single line.

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

The Mozzi-Chasles Theorem :

**rigid body motion** can be decomposed into a rotation around followed or preceded by a a translation along a single line.

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

The Mozzi-Chasles Theorem :

Every *quadreflection* can be decomposed into a rotation around **followed or preceded by** a a translation along a single line.

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

The Mozzi-Chasles Theorem :

Every *quadreflection* can be decomposed into a *commuting ***rotation and translation**

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

The Mozzi-Chasles Theorem :

Every *quadreflection* can be decomposed into *commuting bireflections*

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

Our final theorem :

Every k-reflection can be decomposed into \(\lceil \frac k 2 \rceil\) commuting parts.

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

Our final theorem :

Every k-reflection can be decomposed into \(\lceil \frac k 2 \rceil\) commuting parts.

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

Our final version :

Every k-reflection can be decomposed into \(\lceil \frac k 2 \rceil\) commuting parts.

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

Martin's talk follows up on this and goes into all the algebraic details!

Today, lets envision how this could make a difference in applications and intuition.

2. Geometry | coordinate, dimension and metric agnostic

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

Our final version :

Every k-reflection can be decomposed into \(\lceil \frac k 2 \rceil\) commuting parts.

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

Martin's talk follows up on this and goes into all the algebraic details!

Today, lets envision how this could make a difference in applications and intuition.

2. Geometry | coordinate, dimension and metric agnostic

this is not a vector.

ceci n'est pas un vecteur

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

Martin's talk follows up on this and goes into all the algebraic details!

Today, lets envision how this could make a difference in applications and intuition.

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

ceci n'est pas un vecteur

how many dimensions ?

does it have an orientation ?

AGACSE2021

1. Symmetry Groups | Mozi-Chasles Theorem

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

Martin's talk follows up on this and goes into all the algebraic details!

Today, lets envision how this could make a difference in applications and intuition.

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

ceci n'est pas un vecteur

\( x^2 = \lVert x \rVert^2 \)

only its length is well defined.

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

\( x^2 = \lVert x \rVert^2 \)

When a second non colinear vector is added

- we can measure two lengths

- we can measure one angle

- we know our space is >= 2D

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

\( x^2 = \lVert x \rVert^2 \)

When a second non colinear vector is added

- we can measure two lengths

- we can measure one angle

- we know our space is >= 2D

We can now repeat this ad nauseam. This is of course exactly the structure of geometric algebra. It allows us to postpone the choice of basis till the latest possible time. Such an approach is commonly called 'coordinate free'

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

\( x^2 = \lVert x \rVert^2 \)

When a second non colinear vector is added

- we can measure two lengths

- we can measure one angle

- we know our space is >= 2D

Independent of the choice of basis, our vectors, bivectors and n-vectors are still attached to the origin. To break free from the origin in our space of interest we projectivize and move it outside.

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

- we know our space is >= 2D

This works for spaces of arbitrary (constant) curvature.

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

- we know our space is >= 2D

This works for spaces of arbitrary (constant) curvature.

AGACSE2021

a 4D homogeneous point

a 6D Plucker line

a 4D homogeneous plane

any \(k\), even and odd

any # dimensions,

any metric

closed form exp and log. \(e^{a+b} = e^ae^b\) iff \(ab = ba\)

2. Geometry | coordinate, dimension and metric agnostic

this is a vector.

- we know our space is >= 2D

This works for spaces of arbitrary (constant) curvature.

AGACSE2021

2. Geometry | coordinate, dimension and metric agnostic

- we know our space is >= 2D

This works for spaces of arbitrary (constant) curvature.

Similarly, geometric elements can be defined in a coordinate, metric and dimension agnostic way, and only impose a lower limit on the dimensionality of the space.

AGACSE2021

2. Geometry | coordinate, dimension and metric agnostic

- we know our space is >= 2D

However, formulating a problem independent of the dimensionality of the space does require some extra care...

AGACSE2021

2. Geometry | coordinate, dimension and metric agnostic

- we know our space is >= 2D

robot arm

floor plan

robot arm

Both 2D drawings have points and lines. What happens if we go from two to three dimensions?

AGACSE2021

2. Geometry | coordinate, dimension and metric agnostic

- we know our space is >= 2D

robot arm

floor plan

robot arm

Both 2D drawings have points and lines. What happens if we go from two to three dimensions?

When going to 3D, the lines on the left stay lines. Those on the right become planes .. working dimension independent means embracing duality!

AGACSE2021

2. Geometry | coordinate, dimension and metric agnostic

- we know our space is >= 2D

robot arm

floor plan

robot arm

points and lines defined dually

points and lines defined directly

AGACSE2021

2. Geometry | Forques !

robot arm

floor plan

robot arm

points and lines defined dually

points and lines defined directly

AGACSE2021

2. Geometry | Forques !

In the body frame :

push at origin = translation

push at infinity = rotation

Force and Torque are both lines.

in any # dimensions.

AGACSE2021

2. Geometry | Forques !

In the body frame :

push at origin = translation

push at infinity = rotation

Force and Torque are both lines.

in any # dimensions.

nobody in their right mind does live programming during a talk.

Unfortunately for you all, that leaves me qualified.

By Steven De Keninck

- 306