studiot

What is uncertainty ?

Recommended Posts

4 hours ago, koti said:

Certainty is when a system acts according to a rule or a set of rules when time passes. Uncertainty is when a system does not act according to rules with time passing. Isn’t it that simple?

 

2 hours ago, swansont said:

No. There is noise in all systems, so there will be variation in any repeated measurement. There is no violation of any rule involved.

 

 

2 hours ago, koti said:

What if we exclude the „noise” which is always random anyway?

 

2 hours ago, swansont said:

You can't 

Swansont's replies are incorrect. The reason, though subtle, is of vital importance to all sophisticated communications systems. Koti specifically noted "when a system acts..." Swanson misinterpreted "act" as "measure", but all sophisticated communication systems do much more than just make measurements - they make decisions based upon their measurements; and the entire point of making those decisions is to COMPLETELY eliminate the noise from the decision output sequences. This is done by exploiting a priori known limitations on ALLOWED measurable states. Think of the English alphabet; there are only 26 allowed letters (symbols). Consequently, if a system (such as your brain) receives a letter that is so noisy that it does not appear to be one of the allowed letters, it will detect that error and take steps to correct it. As a simple example, sequences of letters spell words (super symbols); a crude spell-checker can simply compare each received super-symbol (word) to a dictionary of all ALLOWED words, detect unallowed words and then substitute the closet matching allowed word, to correct most noise-induced errors. More sophisticated systems can create super-super-symbols, as allowed sequences (grammatically correct) of multiple words, to further reduce the number of errors in the decision sequences.

This is why your HD television picture is so much cleaner than an old analog TV picture. The analog TV merely measures the input voltage, as Swansont has assumed, and paints the screen with it, received noise and all. But an HDTV system transmits easily recognizable sequences of allowable symbols, that enable gargled symbols to be detected and completely replaced by clean ones. This does not merely reduce the noise, it entirely eliminates it, under most circumstances. When the system is unable to correct the measurement errors, the result is usually a catastrophic failure - the created picture, if there even is one (most HDTVs will blank the screen if the reconstruction is too bad), is highly distorted, like pieces of a jig-saw puzzle that has been assembled incorrectly.

Edited by Rob McEachern

Share this post


Link to post
Share on other sites
20 minutes ago, Rob McEachern said:

 

Swansort's replies are incorrect. The reason, though subtle, is of vital importance to all sophisticated communications systems. Koti specifically noted "when a system acts..." Swanson misinterpreted "act" as "measure",

No, I didn't. 

Quote

but all sophisticated communication systems do much more than just make measurements - they make decisions based upon their measurements; and the entire point of making those decisions is to COMPLETELY eliminate the noise from the decision output sequences.

You can't. You can reduce it to the point where it is basically irrelevant.

Quote

This is done by exploiting a priori known limitations on ALLOWED measurable states. Think of the English alphabet; there are only 26 allowed letters (symbols). Consequently, if a system (such as your brain) receives a letter that is so noisy that it does not appear to be one of the allowed letters, it will detect that error and take steps to correct it. As a simple example, sequences of letters spell words (super symbols); a crude spell-checker can simply compare each received super-symbol (word) to a dictionary of all ALLOWED words, detect unallowed words and then substitute the closet matching allowed word, to correct most noise-induced errors. More sophisticated systems can create super-super-symbols, as allowed sequences (grammatically correct) of multiple words, to further reduce the number of errors in the decision sequences.

Compensating for noise is not the same as claiming it is not present. If a letter is garbled but you can figure out if it was an a or an o by the context, that's error correction. It's not the elimination of noise. It's also not physics, as such.

Quote

This is why your HD television picture is so much cleaner than an old analog TV picture. The analog TV merely measures the input voltage, as Swansort has assumed, and paints the screen with it, received noise and all. But an HDTV system transmits easily recognizable sequences of allowable symbols, that enable gargled symbols to be detected and completely replaced by clean ones. This does not merely reduce the noise, it entirely eliminates it, under most circumstances. When the system is unable to correct the measurement errors, the result is usually a catastrophic failure - the created picture, if there even is one (most HDTVs will blank the screen if the reconstruction is too bad), is highly distorted, like pieces of a jig-saw puzzle that has been assembled incorrectly.

Unable to currect the errors? You just claimed they don't exist.

You can reduce white noise by averaging - it's reduced by the square root of the number of samples. But that in no way eliminates it. The bottom line is if there was no uncertainty, error correction would be unnecessary.

Also, it's swansont. So much for error correction.

Share this post


Link to post
Share on other sites
19 minutes ago, swansont said:

It reflects that we can never measure with arbitrary precision. Measurement devices limit this, as does nature itself.

Sure. But this doesn’t contradict my original, simplistic definition: „Certainty is when a system acts according to a rule or a set of rules when time passes. Uncertainty is when a system does not act according to rules with time passing” 

I’m not going to enforce this definition as its based mainly on gut feeling but it still seems right to me. 

If we stick to QM then „uncertainty” is simply everything, nature in its whole...at least as far as we know up untill now. That is still not a good definition imo.

Edited by koti

Share this post


Link to post
Share on other sites
4 minutes ago, koti said:

Sure. But this doesn’t contradict my original, simplistic definition: „Certainty is when a system acts according to a rule or a set of rules when time passes. Uncertainty is when a system does not act according to rules with time passing” 

I’m not going to enforce this definition as its based mainly on gut feeling but it still seems right to me. 

It has nothing to do with acting according to a rule. When you drop something, it falls according to rules: Newton laws. It is not violating those rules when you get scatter in your results.

There is inherent uncertainty if you try and measure momentum and position of quantum systems. You are following, not violating a rule when you get scatter in your results.

 

Share this post


Link to post
Share on other sites
20 minutes ago, swansont said:

It has nothing to do with acting according to a rule. When you drop something, it falls according to rules: Newton laws. It is not violating those rules when you get scatter in your results.

There is inherent uncertainty if you try and measure momentum and position of quantum systems. You are following, not violating a rule when you get scatter in your results.

 

 Doesn’t scatter act acording to rules as well? If we digg to the bottom of the scatter effect of a ball of blob splashing onto the floor we end up with Heisenberg’s principle which is a rule too. I’m trying to avoid involving randomness but Im not sure its possible.

Edited by koti

Share this post


Link to post
Share on other sites
7 minutes ago, swansont said:

Also, it's swansont. So much for error correction.

Note the timing of my edit, in which I corrected your name, before your post. So much indeed.

11 minutes ago, swansont said:

Compensating for noise is not the same as claiming it is not present.

Noise on the input has been totally eliminated from the output, not merely reduced. The claim pertains to the output, not the input.

14 minutes ago, swansont said:

It's also not physics, as such.

It is directly relevant to the interpretations (decisions) being made in physics. For example, if you think the allowed states of a Bell polarization state measurement must always be either 1 or -1, but 0 (zero) is also a rare, but allowed state (like viewing a coin edge-on), you will end up with some rather weird interpretations of the measurement correlation statistics, which is exactly what has happened.

Share this post


Link to post
Share on other sites
1 minute ago, Rob McEachern said:

 Noise on the input has been totally eliminated from the output, not merely reduced. The claim pertains to the output, not the input.

And it's not what is meant by uncertainty in physics measurements.

1 minute ago, Rob McEachern said:

It is directly relevant to the interpretations (decisions) being made in physics. For example, if you think the allowed states of a Bell polarization state measurement must always be either 1 or -1, but 0 (zero) is also a rare, but allowed state (like viewing a coin edge-on), you will end up with some rather weird interpretations of the measurement correlation statistics, which is exactly what has happened.

I don't recall any Bell examples where that's the case, but it's rather off-topic here.

11 minutes ago, koti said:

 Doesn’t scatter act acording to rules as well? If we digg to the bottom of the scatter effect of a ball of blob splashing onto the floor we end up with Heisenberg’s law which is a rule too. I’m trying to avoid involving randomness but Im not sure its possible.

Yes. You get a Gaussian distribution for white noise. Different kinds of noise will manifest itself in various ways. But it's always in your measurement.

Share this post


Link to post
Share on other sites
4 hours ago, swansont said:

And it's not what is meant by uncertainty in physics measurements.

Exactly the problem, and the subject of this thread.  Uncertainty of what? That is the question. If you always experience exactly what you expected, then you are unlikely to say there is anything uncertain about that experience. So uncertainty is a function of expectations (think allowed symbols, in the above posts). So if you expect the measurements to behave one way and they actually behave a different way, then there are several possible alternative causes for the difference. One is that there is something wrong with the measurements (corrupted by noise) as you have stated. But another is that there is something wrong with your expectations - your a priori model of the way things are supposed to turn out. If you expect to be able to measure two or more independent variables, where only one actually exists, then the correlations between your measurements will be unexpected (and thus proclaimed to be weird), but not due to being uncertain. See my post here: link removed by moderator

Edited by Phi for All
Don't advertise your other posts

Share this post


Link to post
Share on other sites
12 minutes ago, Rob McEachern said:

Exactly the problem, and the subject of this thread.  Uncertainty of what? That is the question. If you always experience exactly what you expected, then you are unlikely to say there is anything uncertain about that experience.

Patently untrue. It is standard practice to report uncertainty on any result.

Quote

So uncertainty is a function of expectations (think allowed symbols, in the above posts). So if you expect the measurements to behave one way and they actually behave a different way, then there are several possible alternative causes for the difference. One is that there is something wrong with the measurements (corrupted by noise) as you have stated.

It's not corruption, as such, since you will always have it.

 

Share this post


Link to post
Share on other sites
10 minutes ago, swansont said:

Patently untrue. It is standard practice to report uncertainty on any result.

Not when you get EXACTLY what you expect. Think about dialing a friend's phone number. If your friend actually answers, do you report that there is some remaining uncertainty about having actually dialed the correct number, or some uncertainty that the telephone system may have not correctly connected your call? Of course not, because the mere act of your friend answering (your expectation has been perfectly met), confirms that fact that there was no uncertainty, in that particular experience.

Share this post


Link to post
Share on other sites

 

13 minutes ago, Rob McEachern said:

Not when you get EXACTLY what you expect. Think about dialing a friend's phone number. If your friend actually answers, do you report that there is some remaining uncertainty about having actually dialed the correct number, or some uncertainty that the telephone system may have not correctly connected your call? Of course not, because the mere act of your friend answering (your expectation has been perfectly met), confirms that fact that there was no uncertainty, in that particular experience.

You're kind of right but not really plus this is in physics so I guess its irrelevant. Plus if an enginner is configuring a router while you dial it is possible you get through to Donald Trump instead of your friend. There is a non zero probability of this.
From physics point of view, uncertainty seems to be present in every aspect of nature so we could define it as an intrinsic part of nature which is always present at different rates. In this case uncertainty seems to be coming only from probability.
From philosophical point of view like your telephone call example (which after consideration is false) or my "tomorrow is Wednesday" example (which is more valid) my simple definition of uncertainty seems to work fine I guess.

Edit: Edited the whole post.

Edited by koti

Share this post


Link to post
Share on other sites
16 minutes ago, Rob McEachern said:

Not when you get EXACTLY what you expect. Think about dialing a friend's phone number. If your friend actually answers, do you report that there is some remaining uncertainty about having actually dialed the correct number, or some uncertainty that the telephone system may have not correctly connected your call? Of course not, because the mere act of your friend answering (your expectation has been perfectly met), confirms that fact that there was no uncertainty, in that particular experience.

Cite a paper where this has happened, because I do believe you are spouting BS.

Share this post


Link to post
Share on other sites
4 minutes ago, swansont said:

Cite a paper where this has happened, because I do believe you are spouting BS.

Is there a non zero probability that tomorrow will NOT be Wednesday for my frame of reference?
It's a genuine question, I'm trying to come up with an answer. If the probability of this is non zero then I guess we can assume that the definition of uncertainty is "intrinsic and unavoidable property of nature" If the probability is exactly zero then I'm confused. 
 

Share this post


Link to post
Share on other sites
37 minutes ago, koti said:

Is there a non zero probability that tomorrow will NOT be Wednesday for my frame of reference?
It's a genuine question, I'm trying to come up with an answer. If the probability of this is non zero then I guess we can assume that the definition of uncertainty is "intrinsic and unavoidable property of nature" If the probability is exactly zero then I'm confused. 
 

Is that a physics measurement? The concept really doesn't apply, in the way you have phrased it.

Now you could ask what time it is, and you would have some uncertainty in the answer because clocks have limited precision. That would be a measurement to which you can apply the concept.

 

Share this post


Link to post
Share on other sites
7 minutes ago, swansont said:

Is that a physics measurement? The concept really doesn't apply, in the way you have phrased it.

Now you could ask what time it is, and you would have some uncertainty in the answer because clocks have limited precision. That would be a measurement to which you can apply the concept.

 

Your clock example positions the attempt to find an answer to the OP in the time measurement "comfort zone" and in this case it works really well I agree. 
Are you saying that it doesn't make sense to apply the same comfort zone to my "zero/non zero tomorrow will be Wednesday probability" dilemma? The only variable which changes is the granulity. I mean purely from GR and QM point of view, can we model a scenario in which there will be something else than a Wednesday tomorrow? We can substitute Wednesday with whatever other point in time in the future (larger than Planck time and smaller than infinity) and apply whatever means of measurement we desire. Or am I being dense here?

Share this post


Link to post
Share on other sites
1 hour ago, koti said:

it is possible you get through to Donald Trump instead of your friend.

I specifically stated the condition "If your friend actually answers" - so getting through to only someone else is no longer even a possibility.

Share this post


Link to post
Share on other sites
8 minutes ago, Rob McEachern said:

I specifically stated the condition "If your friend actually answers" - so getting through to only someone else is no longer even a possibility.

If he answers its a done deal and the question about certainty/uncertainty doesn't exist. Don't pursue this, it's a dead end.

Edited by koti

Share this post


Link to post
Share on other sites
48 minutes ago, koti said:

Your clock example positions the attempt to find an answer to the OP in the time measurement "comfort zone" and in this case it works really well I agree. 
Are you saying that it doesn't make sense to apply the same comfort zone to my "zero/non zero tomorrow will be Wednesday probability" dilemma? The only variable which changes is the granulity. I mean purely from GR and QM point of view, can we model a scenario in which there will be something else than a Wednesday tomorrow? We can substitute Wednesday with whatever other point in time in the future (larger than Planck time and smaller than infinity) and apply whatever means of measurement we desire. Or am I being dense here?

Probability is not measurement, and the results are discrete rather than a continuum. You can't get a result of "Wednesday and a tenth" whereas what time is it is both.

Share this post


Link to post
Share on other sites
19 minutes ago, swansont said:

Probability is not measurement, and the results are discrete rather than a continuum. You can't get a result of "Wednesday and a tenth" whereas what time is it is both.

Understood. But how else other than with probability can we aproach uncertainty? It should be safe to assume that uncertainty is a concept which deals only with future events so if following your „measurement” aproach to describing uncertainty we are hitting a wall here. I’m sure I’m missing something trivial...Wednesday inevitably came almost 2 hours ago so I wil continue tomorrow.

Share this post


Link to post
Share on other sites
1 hour ago, swansont said:

Cite a paper where this has happened, because I do believe you are spouting BS.

What "this" are you referring to? The issue I am attempting to point out, is not the uncertainty in the measurements, but the uncertainty in any decision/behavior based upon those measurements, when it is possibly based upon a false assumption. There are many famous examples in physics, like Lord Kelvin's wildly incorrect estimate of the age of the earth - caused not by uncertainty in the measurements, but uncertainty in the meaning of the measurements. Such uncertainties in meaning, rather than uncertainties in the measurements, are at the heart of all the uncertainties regarding interpretations of quantum theory.

But there is something much more subtle and interesting as well: are physical systems compelled, by the laws of nature, to respond to noise, *as if* it is not noise? Or can an entity ignore the noise, for example, by ignoring the least significant bits in any measurements, that are, a priori, known to be highly likely to be corrupted by noise? Macroscopic communications systems do this all the time. What makes you assume that microscopic systems cannot do the same thing? Because that is indeed an assumption at the heart of the classical versus quantum interpretation problem. If quantum systems *act* as if the least significant bits of fields, forces, potentials etc, do not matter, then all observable behaviors will be quantized - not because the fields etc. are quantized, but because the set of behaviors induced by the fields in other entities, like purported particles, are discontinuous and restricted to a small number of possible behaviors. Since information is quantized, by definition, all interactions driven by the recovery of that information, may also be discrete/quantized, even if the fields/forces that the information is being extracted from, are not quantized, but continuous. In such a case, it is unlikely that a quantized field theory will ever be entirely successful, because the cause of the observed, quantized behavior does not lie within the fields per se, but within the restricted behavioral repertoire of the entities responding to the fields, just as in a classical communications systems, responding discontinuously to continuous electromagnetic field measurements; it does not matter it the actual voltage measurement corresponding to a bit value, is equal to 1.176, the system is going to behave *as if* it was exactly equal to 1.000. Quantum entities may do the same. Attributing quantized behaviors entirely to quantized fields, is thus another source of uncertainty.

Share this post


Link to post
Share on other sites

I think I got it. We have to distinguish between defining uncertainty of past events and uncertainty of future events. Past events uncertainty can be measured, future events uncertainty can only be aproached with probability. @swansont, I think you might have had a conception of past events uncertainty in your mind hence your measurement aproach whereas I had my focus on future events uncertainty. My „Wednesday” example/question essentially becomes a question of „will the time arrow still flow forward” which is a different subject. 

Share this post


Link to post
Share on other sites
37 minutes ago, Rob McEachern said:

What "this" are you referring to? The issue I am attempting to point out, is not the uncertainty in the measurements, but the uncertainty in any decision/behavior based upon those measurements, when it is possibly based upon a false assumption.

Decisions are not what is being discussed. Kindly cease the off-topic distraction. start a new thread if you wish.

 

Share this post


Link to post
Share on other sites
4 hours ago, koti said:

If he answers its a done deal and the question about certainty/uncertainty doesn't exist.

Exactly my point. Understanding what caused the previous uncertainty to suddenly cease to exist, provides the answer to the original question "What is uncertainty?"

Share this post


Link to post
Share on other sites
On 26/12/2017 at 8:07 PM, swansont said:

No, it means I misread your post. Please disregard.

It reflects that we can never measure with arbitrary precision. Measurement devices limit this, as does nature itself.

Midway throught the festive break and I needed to reply to a PM, whilst briefly here, I saw this.

I think it shows that, but more than that as well.

 

This is where my two examples were going.

 

Good to see some responses.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now