# What happens when circular polarization meets a diagonal polarizer?

• I
• fluidfcs
In summary: I don't know. I just assumed that since it's a basic concept that everyone would know. Thanks for clarification.
fluidfcs
TL;DR Summary
When circularly polarized light encounters a 45 degree (diagonal) polarizer does it get fully blocked (or inherently blocked by having it's intensity reduced to 0) or does it get partially through?
I'm finding what seems to be conflicting information on this question and could really use some help. It's my understanding that circularly polarized light is composed of two perpendicular linearly polarized components with a 90 degree phase shift between them. When considered individually, horizontally or vertically polarized light encountering a 45 degree oriented linear polarizer will be absorbed 50% of the time and pass through 50% while of the time (while being converted to 45 degree linear polarization) - assuming a theoretically 100% efficiency of our optical components. It seems logical to me that circularly polarized light would therefore be able to pass through a 45 degree oriented linear filter similarly 50% of the time while being converted to 45 degree linear polarization.

However, I've been posing some questions to various AI models, which I know can't be trusted to always give reliable information, and am getting some responses that make me question my conclusion.

Two such example responses are included below.

1. When circularly polarized light encounters a 45-degree oriented linear polarizer, the light is split into two perpendicular linearly polarized components with a 90-degree phase shift between them. These two components are at 45-degree angles to the axis of the filter and therefore cannot pass through. So, in this case, the entire beam of circularly polarized light is blocked, whereas with horizontally or vertically polarized light, only about half of the photons are absorbed by the filter.2. When circularly polarized light meets a 45-degree oriented linear polarizer, the components of the electric field that are parallel and perpendicular to the axis of the polarizer will be equal in magnitude and out of phase by 90°, resulting in destructive interference and no transmission.

These responses are wrong, aren't they? Thanks in advance.

fluidfcs said:
TL;DR Summary: When circularly polarized light encounters a 45 degree (diagonal) polarizer does it get fully blocked (or inherently blocked by having it's intensity reduced to 0) or does it get partially through?

I've been posing some questions to various AI models, which I know can't be trusted to always give reliable information
Please do not ask an AI language model any physics questions. A language model knows language, it does not know physics. All it knows is that those words generally go together. It has no knowledge of their meaning. It is a language model not a physics model.

Yes, these statements are wrong.

Lord Jestocost, fluidfcs, gentzen and 1 other person
Dale said:
Please do not ask an AI language model any physics questions. A language model knows language, it does not know physics. All it knows is that those words generally go together. It has no knowledge of their meaning. It is a language model not a physics model.

Yes, these statements are wrong.
Thank you for confirming. So, we would expect the circularly polarized light to get past the diagonal linear polarizer roughly 50% of the time and become diagonally polarized, right?

The problem is noit asking some AI. You can ask whomever or whatever you want whatever you want. The problem is when you then ask us to explain the answer to you. As you yourself said, they can't be trusted,

vanhees71
fluidfcs said:
It's my understanding
Obtained from where? Surely from somewhere more reliable than AI models.

vanhees71
The problem is noit asking some AI. You can ask whomever or whatever you want whatever you want. The problem is when you then ask us to explain the answer to you. As you yourself said, they can't be trusted,

And why is that a problem? You said I can ask whomever I want but then that it's a problem if I ask you. Are questions not allowed on this forum? I'm trying to use available resources to come to the best understanding I can. That includes books, videos, websites, AI, forums, etc. On this particular point I've had a very hard time getting accurate information and so wanted to get clarification from other humans. Sorry if I've offended you. Also, my follow-up question above remains unanswered. If you are able and willing to confirm it that would be much appreciated. Thanks and have a good day.

fluidfcs said:
Are questions not allowed on this forum?
AI models are not valid references on this forum. They're not reliable. So questions about what an AI model says are pointless; the only answer is that they're not reliable and you need to go look at a reliable source.

fluidfcs said:
On this particular point I've had a very hard time getting accurate information
I have great difficulty believing that since polarization is treated in pretty much every textbook on EM radiation. You described the understanding you had before you asked any AI models any questions; where did you get it?

vanhees71
PeterDonis said:
AI models are not valid references on this forum. They're not reliable. So questions about what an AI model says are pointless; the only answer is that they're not reliable and you need to go look at a reliable source.I have great difficulty believing that since polarization is treated in pretty much every textbook on EM radiation. You described the understanding you had before you asked any AI models any questions; where did you get it?
I answered your question above. I read many different text books and checked a lot of different online resources but was not finding an answer to this specific situation. Please feel to answer my follow-up question if you can while you're responding here.

fluidfcs said:
No, you didn't. I wasn't asking a general question about what sources you consulted. I was asking specifically where you got this from:

fluidfcs said:
It's my understanding that circularly polarized light is composed of two perpendicular linearly polarized components with a 90 degree phase shift between them. When considered individually, horizontally or vertically polarized light encountering a 45 degree oriented linear polarizer will be absorbed 50% of the time and pass through 50% while of the time (while being converted to 45 degree linear polarization) - assuming a theoretically 100% efficiency of our optical components. It seems logical to me that circularly polarized light would therefore be able to pass through a 45 degree oriented linear filter similarly 50% of the time while being converted to 45 degree linear polarization.
Surely you can give an answer more specific than "I looked at a lot of stuff".

fluidfcs said:
You should be able to answer it yourself based on the understanding I quoted in my previous post. In fact you state the answer explicitly in what I quoted.

PeterDonis said:
You should be able to answer it yourself based on the understanding I quoted in my previous post. In fact you state the answer explicitly in what I quoted.
So you're saying my understanding is correct?

fluidfcs said:
So your response to my question is basically to go try to figure it out elsewhere.
Not at all. You've obviously already tried to do that. But you don't seem to have any confidence in what you figured out, and that confuses me. See further comments at the end of this post.

fluidfcs said:
You're offended at my use of AI, I get that.
It has nothing to do with being offended. AIs are not reliable sources and we don't accept them as sources here. That's just a PF rule.

fluidfcs said:
Still looking to have my understanding confirmed
If you're asking if what I quoted from you in post #9 is a reasonable understanding, yes, it looks reasonable to me.

What I'm struggling with is why you appear to have no confidence in that understanding. That's why I keep asking where you got it from. All you have said so far is "I read a lot of stuff". What stuff? Did you read any textbooks? Any peer-reviewed papers? Is there some particular reason why you are unsure whether to trust what you read?

PeterDonis said:
Not at all. You've obviously already tried to do that. But you don't seem to have any confidence in what you figured out, and that confuses me. See further comments at the end of this post.It has nothing to do with being offended. AIs are not reliable sources and we don't accept them as sources here. That's just a PF rule.If you're asking if what I quoted from you in post #9 is a reasonable understanding, yes, it looks reasonable to me.

What I'm struggling with is why you appear to have no confidence in that understanding. That's why I keep asking where you got it from. All you have said so far is "I read a lot of stuff". What stuff? Did you read any textbooks? Any peer-reviewed papers? Is there some particular reason why you are unsure whether to trust what you read?
I can't quote you a single source that I got my understanding from. I've been studying this stuff for awhile and have used many difference source to build that understanding. Thank you for confirming, I appreciate it.

fluidfcs said:
So, we would expect the circularly polarized light to get past the diagonal linear polarizer roughly 50% of the time and become diagonally polarized, right?
Sure, sounds good! But then again, how do you know I and the others are not all sophisticated AIs posting answers to questions here and therefore can't be trusted?

My question to you is, why did the AI responses make you doubt your own conclusions? Based on what you wrote, I thought you should've been easily able to point out the flaw in the first response you posted, but I'm inferring that you have a certain level of understanding. If you're working from "I 'know' that if vertically polarized light hits a polarizer at 45 degree, half of the light gets through" because you read that in a book somewhere without understanding why that's indeed the case, your doubt perhaps make sense. You can address it by learning why half the light gets through. On the other hand, if you do understand why, then you should've been able to see exactly why the first AI response was wrong and dismissed it.

Picking apart an AI's response can actually be a good learning experience for you. It will help you to see things with a critical eye and evaluate what people tell you, instead of having to rely on trust and authority.

fluidfcs said:
I can't quote you a single source that I got my understanding from.
Then my advice to you is to change that practice: pay more attention to where you get information from. To evaluate the reliability of your sources, you have to know which sources you used. And evaluating the reliability of your sources seems to be a key issue that you have had: you had so little confidence in what was actually a correct understanding, that you went and asked AI models, the least reliable sources you could possibly have picked, and they made you doubt what was actually a correct belief.

Dale
fluidfcs said:
And why is that a problem?
"Hey! I read this in a fortune cookie! Now explain it to me, you guys." Do you not see how this is a problem?

An AI strings words together without any understanding of what they mean. How can this possibly be considered a reasonable resource?

"Hey! I read this in a fortune cookie! Now explain it to me, you guys." Do you not see how this is a problem?

An AI strings words together without any understanding of what they mean. How can this possibly be considered a reasonable resource?
I've been able to learn quite a lot from AI. They don't get everything right, but if you have a solid foundation in the topic you can often identify the small number of cases where things don't look right. Definitely comparing to a fortune cookie is an exaggerated and bad faith comparison. Some of the AIs I use (Elicit, Perplexity, You) even link me to their sources, which I can use to validate and/or dig further. They are imperfect but still valuable and useful learning and research tools.

Clearly many of you are strongly biased against using AI tools, that's fine. I don't feel a need to sell you on their merit.

weirdoguy and PeroK
fluidfcs said:
I've been able to learn quite a lot from AI.
How can you possibly know that? You don't know if anything the AI tells you is correct.

fluidfcs said:
if you have a solid foundation in the topic
Then why do you need an AI to learn from?

fluidfcs said:
Some of the AIs I use (Elicit, Perplexity, You) even link me to their sources
Then you're not really using the AI as a source, you're using it as a search engine. Yes, if you get a valid source you can use it here no matter where you got it from. That's very different from citing the AI itself as a source.

fluidfcs said:
Clearly you guys are all strongly biased against using AI tools
It's not a matter of bias. It's a matter of reliability. The fact that AIs are not reliable is commonplace.

fluidfcs said:
So, we would expect the circularly polarized light to get past the diagonal linear polarizer roughly 50% of the time and become diagonally polarized, right?
Yes. You can consider the circularly polarized light to be two linear polarized sources, with a 90 deg phase offset. Then the one parallel to the diagonal passes unattenuated and the one perpendicular is completely attenuated. So half of the energy is attenuated.

fluidfcs said:
I'm trying to use available resources to come to the best understanding I can
Then wait for the AI developers to develop an AI physicist. Again, current AI models are language models. They know language, not physics. Using them to learn physics is using them outside of their intended purpose. They are neither designed nor trained to write correct physics, just good English.

fluidfcs said:
if you have a solid foundation in the topic you can often identify the small number of cases where things don't look right.
I disagree with this quite strongly. If you already have a strong foundation then you don’t need to be asking an AI. Second, it is not a small number of cases that they get wrong. Again, these are language models. They know 0 physics. They only know what words often go together and how to string them together with good grammar and style. Asking physics questions of a language model makes no sense.

Furthermore these language models sound confident. So not only do they not know any physics, when you ask them about physics they respond with all of the relevant vocabulary in a very confident sounding style with no knowledge about the underlying physics.

You had a solid foundation but were fooled by a confident but incorrect response. The AI confidence is dangerous.

fluidfcs said:
Some of the AIs I use (Elicit, Perplexity, You) even link me to their sources, which I can use to validate and/or dig further.
A lot of them just make up the sources. Including putting real recognized expert authors who never collaborated together on papers that don’t exist in reputable journals that do exist. And since they don’t know any physics if they do happen to write an actual reference they don’t know if it actually supports what they said. All they know is that that reference sometimes comes up in conversations with those words.

Instead of asking an AI about physics, you should ask an English teacher. The language models are trying to replicate an English teacher, so you will be better off asking the real thing rather than the AI imitation. If trying to learn physics from an English teacher seems ineffective, then learning it from an imitation English teacher would be even less effective.

Last edited:
weirdoguy, vanhees71, Lord Jestocost and 3 others
Perhaps one should answer the physical question too. It's simply out of question that ChatGPT is not a reliable source. I don't think we need 19 postings to clarify this ;-).

The polarization state of a circular polarized photon is
$$|\psi \rangle=\frac{1}{\sqrt{2}} (|H \rangle + \mathrm{i} |V \rangle),$$
taking the ##h=+1## state as an example.

Then the probability that the photon passes a "diagonally oriented polarization filter" is given by the scalar product with the corresponding eigenvector,
$$|D \rangle=\frac{1}{\sqrt{2}} (|H \rangle + |V \rangle),$$
i.e.,
$$P=|\langle D|\psi \rangle|^2.$$
Now
$$\langle D | \psi \rangle=\frac{1}{2} (\langle H + \langle V|)(\langle H + \mathrm{i} \langle V)=\frac{1}{2} (1+\mathrm{i}) \; \Rightarrow \; P=\frac{1}{2}.$$
After the polarization filter any photon that comes through is in the polarization state described by ##|D \rangle##.

DrChinese, Dale, Lord Jestocost and 1 other person
fluidfcs said:
I can't quote you a single source that I got my understanding from. I've been studying this stuff for awhile and have used many difference source to build that understanding. Thank you for confirming, I appreciate it.
The question is simply answered by expressing the circular polarised state in the diagonal basis. As @vanhees71 has done.

There's nothing in your posts that suggests you understand polarisation. I suggest that your understanding is akin to that of an AI. You have some knowledge of how various words and phrases go together but no real understanding of the basics.

I suggest, therefore, that you review your approach to learning physics. Especially in regard to the mathematical basis for QM generally.

DrChinese and vanhees71

• Quantum Physics
Replies
43
Views
2K
• Quantum Physics
Replies
8
Views
1K
• Quantum Physics
Replies
41
Views
3K
• Quantum Physics
Replies
8
Views
1K
• Quantum Physics
Replies
2
Views
647
• Quantum Physics
Replies
28
Views
677
• Quantum Physics
Replies
17
Views
2K
• Quantum Physics
Replies
4
Views
1K
• Quantum Physics
Replies
6
Views
1K
• Quantum Physics
Replies
49
Views
2K