03_log_ratio_transformation Flashcards

1
Q

What does the euclidean dot product do

A

Tell us how to multiply two vectors,

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Hvad gør en norm?

A

Regner længden på en vector ud.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Hvad gør den euklidiske distance?

A

Regner længden af en vektor mellem to punkter ud.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

En grund til at euklidisk distance ikke fungerer for kompositioner:

A

Den regner den numeriske distance mellem parts ud lige og tager ikke højde for den relative stigning.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Hvad er meningen med Aitchison geometri?

A

En finde en metric på simplexen der gør det muligt at sammenligne som med euklidiske metrics.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Hvordan kan man definere det indre produkt mellem to vektorer på simplxen?

A

Aitchison inner product:

1/2D * sum_i_to_D(sum_j_to_D(ln(xi/xj)*ln(yi/yj)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Hvordan kan man regne normen ud?

A

sqrt(innerproduct(x,x)) =

1/2D * sum_i_to_D(sum_j_to_D(ln(xi/xj)^2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Hvordan kan man regne en aitchison distance ud?

A

norm(x-y)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is the Hilbert space?

A

The linear vector space of the simplex generated by thge aitchison metric.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is pertubation?

A

Like anormal vector space the Hilbert space needs the algebraic + and *. Pertubation is the +.

x + y = C(x1y1,…,xDyD)

Can be sued to center a composition.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is powering?

A

The simplex space equivalent of multiplying.

a * x = C(x1^a, … , xD^a)

can be used to scale a composition.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Hvad er ideen med transformation?

A

Fordi Aitchison er bøvlet at arbejde med kan man oversætte til det reele talrum.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Hvad er princippet bag den additive log-transformering?

A

Mapper fra S^D til R^D-1. Er sande koordinator (der er ikke en del af talrummet der ikkefyldes ud), men distance fungerer ikke, da nævneren kan ændres og det hele laves om. At tage log gør at + og * stadigvæk virker. Kan skrives op som:

ALR(x) = (ln(x1/xd), … , ln(xd-1/xd))

Affine transformationer kan gå an.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly