Can the orientation of a body be defined by Euler angles?

  • A
  • Thread starter Trying2Learn
  • Start date
In summary: If you have two sets of orthogonal axes (x,y,z) and (x',y',z') then if z and z' are coincident then you can make all the axes coincident by a rotation about z.
  • #1
Trying2Learn
377
57
TL;DR Summary
Euler angles and orientations
[Mentor Note -- OP has been reminded that AI chatbots are not valid references in the technical forums]

Good morning.​
Every now and then, I ask this question to colleagues, friends and professors (and a few times, here).​
Eventually, I give up. For I have never been given a satisfactory explanation. This time, I asked chatGPT.​
I hope I can share the chat and maybe then, someone can help me.​
-------------------------------------------------------------------------------------------------------------------------​
Here is the question I asked chatGPT:​
how does one mathematically prove that a body's orientation can be determined by three Euler angles (where the rotations are intrinsic and the last axis is repeated)?
After taking me through the three rotations (obvious to me, and not the issue; and no need to repeat for it just defines the standard rotation matrices), it closed with this​

5. To prove that any arbitrary orientation can be determined by these three Euler angles, we need to show that any rotation matrix R can be expressed in terms of the angles α, β, and γ.
By expanding the multiplication in step 4, you will obtain a general expression for R that consists of trigonometric functions involving α, β, and γ.
R = [[R11, R12, R13],[R21, R22, R23],[R31, R32, R33]]
Each element of R can be expressed as a combination of trigonometric functions involving α, β, and γ.6. Therefore, we have mathematically proven that any arbitrary orientation can be represented by the three Euler angles α, β, and γ, where the rotations are intrinsic and the last axis is repeated.


So then I asked:

I do not understand your summative step 6. For that is EXACTLY where I am confused. How do you prove that a matrix that is a member of SO(3) formed from the three intrinsic angles (with the last one repeated) can describe any orientation?
And here is the answer​
The fact that a matrix in SO(3) (the special orthogonal group in three dimensions) can be represented by three Euler angles (with the last one repeated) is not a straightforward mathematical proof. Instead, it is a convention and a choice of representation that has been widely adopted due to its practicality and intuitive interpretation...
Whoa! So then I asked​
Where can I find this complicated proof?
And here is the answer​
The proof you are seeking, which demonstrates that a matrix in SO(3) can be uniquely represented by three Euler angles with the last one repeated, is not a straightforward or simple proof. In fact, such a proof does not exist.
Whoa! And now I understand... not the answer... but my question... (and that I am not an idiot, I hope).​
Now I would like to ask all of you.​

Where can I find such a mathematical proof? I can already intuit or imagine the validity of the result (I have long ago accepted that, visually).​
But I have never seen a proof. I want to see it in my mind, not in my eyes.​
In other words.​
You give me a body and an orientation in 3D. Prove to me that three intrinsic euler angles (with the last one repeated) can define that orientation.​
This is exactly the issue that has bothered me for several years.​
 
Last edited by a moderator:
  • Like
Likes dextercioby
Physics news on Phys.org
  • #3
vanhees71 said:
I don't read texts generated from ChatGPT. If you like an answer from a human, here's my try to explain it:

https://itp.uni-frankfurt.de/~hees/pf-faq/spinning-top.pdf

See Sect. 2 for a pretty intuitive argument and the Appendix for a continuum-mechanical proof of Euler's theorem.

You refer to Euler's theorem.

Euler's rotation theorem states that, in 3D, any displacement of a rigid body such that a point on the rigid body remains fixed, is equivalent to a single rotation about some axis that runs through the fixed point.

I get that. But I am not interested in that.

I have only one focus: how do you KNOW (in the mind by math, not the eye by observation), that if I give you an orientation of a body from some fixed inertial frame...

You can always find a member of SO(3) that defines that orientation when that SO(3) is made of three intrinsic rotations (where the third one repeats the first).

I already know how to convert a member of SO(3) into the three unique angles (so that is not the issue).
The issue is the bold highlight, above.
 
  • #4
Trying2Learn said:
You can always find a member of SO(3) that defines that orientation when that SO(3) is made of three intrinsic rotations (where the third one repeats the first).
I'm not sure I understand what you want.

If I have two sets of orthogonal axes (x,y,z) and (x',y',z') then if z and z' are coincident then I can make all the axes coincident by a rotation about z. If z and z' are not coincident I can make them coincident by a rotation around an axis orthogonal to both z and z'. And if I do a rotation around z I can place x (or y) to be that axis. So I have a rotation around z to make x orthogonal to z and z', a rotation around x to make z and z' coincident, and a rotation around z to make the remaining axes coincident.

Is that what you're looking for?
 
  • Like
Likes Trying2Learn and topsquark
  • #5
Trying2Learn said:
You refer to Euler's theorem.

Euler's rotation theorem states that, in 3D, any displacement of a rigid body such that a point on the rigid body remains fixed, is equivalent to a single rotation about some axis that runs through the fixed point.

I get that. But I am not interested in that.

I have only one focus: how do you KNOW (in the mind by math, not the eye by observation), that if I give you an orientation of a body from some fixed inertial frame...

You can always find a member of SO(3) that defines that orientation when that SO(3) is made of three intrinsic rotations (where the third one repeats the first).

I already know how to convert a member of SO(3) into the three unique angles (so that is not the issue).
The issue is the bold highlight, above.
If you keep one point fixed you just need to give the three body-fixed right-handed cartesian basis vectors as linear combinations of the three space-fixed right-handed cartesian basis vectors. It's easy to show that the corresponding transformation matrices are SO(3) matrices. That's shown in Sect. 2 in the above quoted manuscript (Eqs. (1-7)).
 
  • Like
Likes Trying2Learn
  • #6
vanhees71 said:
If you keep one point fixed you just need to give the three body-fixed right-handed cartesian basis vectors as linear combinations of the three space-fixed right-handed cartesian basis vectors. It's easy to show that the corresponding transformation matrices are SO(3) matrices. That's shown in Sect. 2 in the above quoted manuscript (Eqs. (1-7)).
Apparently, the moderators did not like my use of chatGPT to ASK the question (I disagree with them, but I will accept it).

However, this exchange has finally helped me to ask the question (and, now, answer it in the way I wanted).

To that end, please let me continue.

Given a member of SO(3), I know I can decompose that into three intrinsic angles in one of two cases:
  • Where the last intrinsic rotation is not either of the first two,
  • Where the last intrinsic rotation is a repeat of the first.

That is not my issue.

My issue is this:

How do I know there exists a member of SO(3) than can take a body into all possible orientations?

And, now, I think I have the answer (please understand that while the other answers helped, this was the point i was trying to understand -- if this point below is correct -- haha)

SO(3) is three dimensional and continuous, and those are the two critical issues that address my question.
  1. The 3D nature means we have 3 unique elements of any rotation matrix the other six elements determined by orthogonality)
  2. Continuity means that there is a continuous path through rotations that can smoothly transition between any two orientations of the body.

Is this true?
 
  • #7
Thread closed for Moderation...
 
  • #8
berkeman said:
Thread closed for Moderation...
Thread reopened. OP has been warned that AI's are not valid sources.
Greg Bernhardt said:
ChatGPT and AI-generated text
  • Posting AI-generated text without attribution is categorically disallowed and will lead to a warning and an eventual permanent ban with continued use.
  • Answering a science or math question with AI-generated text, even with attribution, is not allowed. AI-generated text apps like ChatGPT are not valid sources.
  • Threads about the technology and cultural impact behind AI like ChatGPT are allowed
  • Usage of AI-generated text output in entertainment threads in General Discussion with attribution is allowed.

Please focus the discussion on Euler angles.
 
  • #9
Trying2Learn said:
Apparently, the moderators did not like my use of chatGPT to ASK the question (I disagree with them, but I will accept it).

However, this exchange has finally helped me to ask the question (and, now, answer it in the way I wanted).

To that end, please let me continue.

Given a member of SO(3), I know I can decompose that into three intrinsic angles in one of two cases:
  • Where the last intrinsic rotation is not either of the first two,
  • Where the last intrinsic rotation is a repeat of the first.

That is not my issue.

My issue is this:
Maybe I'm just not understanding, what you are asking.
Trying2Learn said:
How do I know there exists a member of SO(3) than can take a body into all possible orientations?
That's what's shown in my manuscript. https://itp.uni-frankfurt.de/~hees/pf-faq/spinning-top.pdf

What is not answered in the beginning of Sect. 2 (up to including Eq. (7))?

Trying2Learn said:
And, now, I think I have the answer (please understand that while the other answers helped, this was the point i was trying to understand -- if this point below is correct -- haha)

SO(3) is three dimensional and continuous, and those are the two critical issues that address my question.
  1. The 3D nature means we have 3 unique elements of any rotation matrix the other six elements determined by orthogonality)
  2. Continuity means that there is a continuous path through rotations that can smoothly transition between any two orientations of the body.

Is this true?
Exactly!
 
  • Love
  • Like
Likes Trying2Learn and malawi_glenn
  • #10
vanhees71 said:
Maybe I'm just not understanding, what you are asking.

That's what's shown in my manuscript.https://itp.uni-frankfurt.de/~hees/pf-faq/spinning-top.pdf

What is not answered in the beginning of Sect. 2 (up to including Eq. (7))?Exactly!
thank you thank you thank you to all

I am sorry to have included Chat. It was not my intention to circumvent the rules, but that chat helped me ask the question.

Then I had to read the document that was posted; and while I understood, it was not what I wanted.

Thank you vanhees71 (and all others) for your infinite patience!
 
  • Like
Likes berkeman
  • #11
Have you thought about the contradiction, that there at least one orientation in 3d that can not be expressed in euler angles?
 
  • Like
Likes Trying2Learn
  • #12
malawi_glenn said:
Have you thought about the contradiction, that there at least one orientation in 3d that can not be expressed in euler angles?

Oh yes, I am aware of the gimbal lock, but that was not the issue.

But thank you anyway!
 
  • #13
Trying2Learn said:
thank you thank you thank you to all

I am sorry to have included Chat. It was not my intention to circumvent the rules, but that chat helped me ask the question.

Then I had to read the document that was posted; and while I understood, it was not what I wanted.

Thank you vanhees71 (and all others) for your infinite patience!
I still don't know, what's unanswered in my manuscript. Please formulate your question precisely! Maybe then it can be answered (or even then you'll be able to answer it yourself)!
 
  • Like
Likes Trying2Learn
  • #14
vanhees71 said:
I still don't know, what's unanswered in my manuscript. Please formulate your question precisely! Maybe then it can be answered (or even then you'll be able to answer it yourself)!
Let me put it this way

Psalms, 23:4, edited:
Yea, though I walk through the valley of the shadow of death, I will fear no evil: for I am the most stubborn ass on the planet.

Deep in my head, I had MY answer like I wrote it, and I had to dig it out on my terms. I read your document, and that begin to bang around and loosened the thought.

Yes, now in hindsight, I see it, but I was so stubborn to want the answer like I did said on my terms.

I wish I could explain how I get when I start stumbling. But now, right now, your answer in your document is much better -- I just had to see it my way, first.
 
  • #15
Trying2Learn said:
Oh yes, I am aware of the gimbal lock, but that was not the issue.

But thank you anyway!
Note that Euler angles are only coordinates for SO(3) matrices. You cannot cover SO(3) with a single coordinate chart, you need at least 2. For the Euler angles, indeed the "gimbal lock" shows just that. Note that in my manuscript in this first few paragraphs, I don't use any coordinates to parametrize SO(3) at all but just SO(3) itself. So in this approach there are no such problems of coordinate singularities at all. The disadvantage for mechanics is that you don't have holonomous coordinates, which are provided by the Euler angles. Note, however, that I also show that by using the action principle directly for the parametrization with SO(3) matrices you can still derive Euler's equations, which is a lot already.
 
  • Like
Likes Trying2Learn
  • #16
vanhees71 said:
Note that Euler angles are only coordinates for SO(3) matrices. You cannot cover SO(3) with a single coordinate chart, you need at least 2. For the Euler angles, indeed the "gimbal lock" shows just that. Note that in my manuscript in this first few paragraphs, I don't use any coordinates to parametrize SO(3) at all but just SO(3) itself. So in this approach there are no such problems of coordinate singularities at all. The disadvantage for mechanics is that you don't have holonomous coordinates, which are provided by the Euler angles. Note, however, that I also show that by using the action principle directly for the parametrization with SO(3) matrices you can still derive Euler's equations, which is a lot already.
Absolutely, now I see your document is the best.

Here, this is what I wrote to the moderator when he censored me.

I was not using chat to answer, I was using chat to ask.

On her deathbead, the American librettist, Gertrude Stein, was asked, "Gertrude, what is the answer?"
She said, "What is the question?"
But she was not being facetious. Her response was the answer to the question.

Knowing how to ask the question has always been the most difficult for me. I find that once I ask the question on my terms (and that is so difficult to define), I know the answer. I am not particularly adept at theory, and this takes time, and my head is often clouded by how I WANT to see the answer.

I don't know if that makes sense, but it is how I have always learned: a massive struggle.

(I am also an engineer, not a mathematician or physicist, so this stuff comes slowly for me.)
 
  • #17
I've the impression, you still use some bot to create this prose ;-). SCNR. I don't see, how this leads to anything useful...
 
  • #18
vanhees71 said:
I've the impression, you still use some bot to create this prose ;-). SCNR. I don't see, how this leads to anything useful...
Oh now, this is me. I am immensely stubborn. That is a huge problem for me.

What does SCNR mean?
 
  • #19
SCNR = "sorry, couldn't resist". It's age-old Usenet slang.
 

FAQ: Can the orientation of a body be defined by Euler angles?

What are Euler angles?

Euler angles are a set of three angles that describe the orientation of a rigid body with respect to a fixed coordinate system. They represent rotations about the axes of the coordinate system and are commonly denoted as φ (phi), θ (theta), and ψ (psi).

Can the orientation of a body be uniquely defined by Euler angles?

Yes, the orientation of a body can be uniquely defined by Euler angles, but this is subject to certain conditions. Euler angles can suffer from gimbal lock, a situation where two of the three rotational axes become aligned, leading to a loss of one degree of freedom. Despite this, they are still widely used for many practical applications.

What are the common conventions for Euler angles?

There are several conventions for Euler angles, depending on the order and axes of rotation. The most common conventions are ZYX, ZYZ, and XYZ. For example, in the ZYX convention, the rotations are performed in the order of yaw (ψ), pitch (θ), and roll (φ).

What are the advantages and disadvantages of using Euler angles?

The primary advantage of using Euler angles is their simplicity and intuitive understanding of rotations. However, the disadvantages include the potential for gimbal lock and the complexity of converting between Euler angles and other representations like rotation matrices or quaternions.

How do Euler angles compare to quaternions for representing orientation?

Quaternions are often preferred over Euler angles for representing orientation because they do not suffer from gimbal lock and provide a more compact and efficient representation for interpolation and numerical stability. However, Euler angles are still useful for their simplicity and ease of understanding.

Similar threads

Back
Top