# I How does observation affect reality

1. Apr 21, 2017

### Trollfaz

In quantum physics, one can change a system just by observing it, such as wave function collapse and quantum Zeno effect. I don't quite get how observation affects them, unless we are interacting with them when observing them.

2. Apr 21, 2017

### Staff: Mentor

That's the point. Observation requires interaction. As to how that change actually takes place is a question of interpretation.

3. Apr 21, 2017

### Trollfaz

I believe that only certain types of interaction can affect the system

4. Apr 21, 2017

### Staff: Mentor

Technically, any interaction will affect a system, but indeed not all interactions lead to a significant change in the wave function of the system in question.

By the way, nature doesn't care what you believe. Could you clarify what you mean, and how it is related to your original question?

5. Apr 21, 2017

### Trollfaz

I mean what types of interactions will affect the system but this is recently answered

6. Apr 21, 2017

### hilbert2

Even in classical mechanics you need to have light reflecting from an object before you are able to see it, and that always exerts an electromagnetic radiation pressure on it, affecting its state of motion. The difference to quantum mechanics is that in QM you can't make the effect of observation on the object as small as you want just by having detectors that can see smaller light intensities (so that more dim lighting is sufficient for monitoring the object).

7. Apr 21, 2017

### stevendaryl

Staff Emeritus
Well, an especially weird case is EPR, where observing one system can seemingly affect a second, very distant (but correlated) system. It's not too difficult to give a technical explanation of why observation can do this, although the explanation still leaves a seemingly unresolvable nugget of mystery.

The hallmark of quantum mechanics is interference effects. Suppose you set up a system in initial state $|I\rangle$, and then later test whether it is in final state $|F\rangle$. Suppose that there are two alternative ways for the system to go from $|I\rangle$ to $|F\rangle$, via intermediate state $|A\rangle$, or via intermediate state $|B\rangle$. Then the probability of finding the system in state $|F\rangle$ at the end is given by:

$P(I,F) = P(I, A) P(A, F) + P(I, B) P(B, F) + \Phi(I,A,B,F)$

where $\Phi(I,A,B,F)$ is an "interference term". Without the interference term, the probabilities are easy to interpret:
• There are two ways to get from $I$ to $F$: (1) along the path $I$ to $A$ to $F$, and (2) along the path $I$ to $B$ to $F$.
• The probability for the first path is just $P(I,A)$, the probability for the first "leg" of that path, times $P(A,F)$, the probability for the second "leg".
• Similarly for the second path.
• The total probability is just the sum of the probabilities for each path, taken separately.
But the interference term is the thing that is hard to understand in pre-quantum terms. And it's the source of the nonlocal observer effect in quantum mechanics. There can either be destructive interference, making the probability smaller for both paths than for either path separately, or there can be constructive interference, making the probability larger than for either path. This is illustrated by the famous "two-slit interference" experiment. Shine light on one screen with two small slits allowing the light to pass through, and then a second screen will show dark and bright lines, the dark lines where the light from one slit destructively interferes with light from the second slit, and the bright lines where there is constructive interference. Intuitively, it's hard to understand why opening a second slit for light to pass through would ever make it darker in some region of the second screen. The interference pattern of bright and dark lines persists even when the light intensity is reduced so low that you are seeing single photons pass through the slit. Each photon (or you can use electrons instead) seems to experience interference between the two alternative paths.

An example of the observer effect is that if you try to observe which slit each photon travels through, then the interference pattern is destroyed. The extra, nonclassical interference term goes away. It doesn't matter how unobtrusive your observation is; if it is possible to determine which path the photon takes, there is no interference.

Some people describe this effect as "observation collapses the wave function", but there is nothing special about observation. The key to understanding the observer effect is to realize that in order for there to be interference between two paths, the two paths have to lead to the same final state. If there is some difference, no matter how small, in the final states for the two paths, then there is no interference. If you observe which path a photon takes, then that makes the final states different, because your state is different. The total state consists of the state of the photon plus the state of the observer (or a measuring device, or another particle, or whatever interacts with the photon). So depending on which path was taken, the total final state is one of two possibilities:
1. The photon is in final state $F$, and you are in a state of remembering that it took path $A$
2. The photon is in final state $F$, and you are in a state of remembering that it took path $B$
Those are different final states, so there is no interference between those two possibilities. It doesn't matter how unobtrusive your measurement was, or how slightly it affected the photon--if your final state is different at the end, then there is no interference.

For a human being, there is no way to reverse the effects of observing something, so there is no way to restore the interference pattern. But if the second system that the photon interacted with is something very small---say another particle--then it is possible in some circumstances to reverse the effects of the interaction so that the final states are the same. In that case, the interference pattern is restored. This is the essence of "quantum eraser" experiments, described here: https://en.wikipedia.org/wiki/Quantum_eraser_experiment

8. Apr 21, 2017

### Mentz114

I'm sorry to be picky but it seems that experimental support for the electron case is not strong. In this paper**, the authors say
Note, they do not say 'interference pattern'. This helps to debunk a persistent quantum myth.

**Controlled double-slit electron diffraction

https://arxiv.org/abs/1210.6243

9. Apr 22, 2017

### Staff: Mentor

This is quibbling over words. The pattern observed in electron double slit experiments, whether you use the words "interference pattern" or "diffraction pattern" to describe it, is explained by quantum interference between components of the electron wave function coming from each slit. That is what "interference between the two alternative paths" means.

What myth is that?

10. Apr 22, 2017

### hsdrop

Is there any way that we know of to interact with a wave function and not collapse it??

11. Apr 22, 2017

### DennisN

"Collapsing wavefunctions" actually assumes certain interpretations of Quantum Mechanics. There are interpretations without wavefunction collapse, see e.g. Interpretations of quantum mechanics - Comparison of interpretations.

Also,
is also a matter of interpretation, i.e. is the wave function physically real/existing? See e.g. Wave function - Ontology.

So, maybe your question could be reworded to "Is there any way that we know of to interact with a particle or system without disturbing it/changing it?" or something like that. If so, as far as I know, there is no way. But there is more that can be said about this, one thing you could have a look at is what is called "weak measurements".

12. Apr 22, 2017

### Mentz114

Acknowledged. Diffraction is a wave property like interference. Don't know why I thought otherwise.

13. Apr 22, 2017

### Staff: Mentor

Ah, ok. Yes, I agree this is a myth.

14. Apr 23, 2017

### vanhees71

It's not only a myth it's clearly disproven by experiment. You never find "half of an electron" anywhere but always "one electron" or "no electron".

15. Apr 23, 2017

### Staff: Mentor

There is no collapse in QM.

To see this you need to study an axiomatic treatment. See post 137:

Notice in the axiom nothing about collapse at all. It comes about from beginning texts and even some intermediate ones that don't explain things carefully enough. Study an advanced text like Ballentine for the details.

These days an observation is generally considered to occur once an interaction has happened that leads to decoherence and a superposition is converted to a mixed state.

Thanks
Bill

16. Apr 23, 2017

### stevendaryl

Staff Emeritus
I would say that "wave function collapse" is a rule of thumb for working with quantum mechanics, which reflects the following three interpretation-independent facts about observations:

1. If you measure a quantity, you always get an eigenvalue of the corresponding operator.
2. Measurement destroys interference between alternatives.
3. After a measurement, you can, for all practical purposes, treat the system as if it is now in an eigenstate of the operator corresponding to the observable being measured.
The meaning and/or explanation of these facts is a matter of interpretation, but the facts themselves are independent of interpretation. They are empirically confirmed. I think that they are almost completely explained by considering the measurement apparatus itself to be a quantum-mechanical system. Almost.

17. Apr 23, 2017

### vanhees71

1. is part of the theory. Ok.

2. This statement already is not so easy to understand for me. What do you mean by it? Given a state, I know the probability to measure a value of the measured observable, no more no less. What do you mean by "interfering"? There are just some more or less probable possibilities for the outcome of the measurement with the only exception that maybe the system is determined in a state, where the measured observable is determined (i.e., takes a one of its possible values with 100% probability).

3. Is true for a very rare class of measurements, the socalled von Neumann filter measurements. It's almost always technically impossible to perform such an experiment. In general you destroy the system measured. E.g., registering a photon usually absorbs it, and it's destroyed, providing a bit of heat energy resulting in a little temperature increase of the detector material.

18. Apr 23, 2017

### stevendaryl

Staff Emeritus
It actually shouldn't be, in my opinion. It's a rule of thumb at best.

2. This statement already is not so easy to understand for me. What do you mean by it? [/QUOTE]

It seems obvious to me. As I said in another thread: Interference is the phenomenon where there are two possible "paths" between a fixed initial state and a fixed final state. The probability of going from the initial state to the final state is found by adding the amplitudes for the two (or however many there are) alternative paths, and squaring. The result includes cross terms involving both paths, which are "interference" terms. If you perform an observation that attempts to see which path is actually taken, that destroys the interference pattern. The mathematical explanation is entanglement: the observation entangles the system being measured with the system doing the observation. The reduced density matrix for just the system being measured then has no interference terms.

They might be rare, but they are the most interesting (such as EPR).

19. Apr 23, 2017

### vanhees71

Ok, then 2. is also standard. EPR is all about entanglement. You don't need to perform von Neumann filter measurements to prove it, as zillions of such experiments with photons show, which are destroyed in the measurement rather than coming out in a definite new state.

20. Apr 24, 2017

### zonde

I am trying to follow discussions about measurements but one thing that makes it hard to follow these discussions is interpretation of term "measurement".
If I try to keep track of experimental equipment that corresponds to mathematical operations, there is some ambiguity. Let's take simple example of PBS and detector. PBS splits the beam of photons in two beams with certain polarizations. These polarizations are determined by orientation of PBS. Outputs are tested with detectors.
And it seems that there are two different mathematical operations that correspond to different pieces of equipment. PBS - projection operator, detector - Born rule. Respectively detector can only tell coefficient of eigenvector but orientation of PBS can tell what eigenvectors are it's outputs while it can't tell coefficients of eigenvectors.

So option 1. seems to correspond to combination of PBS and detector.
Option 2. seems to talk about detectors or even something more complex.
Option 3. corresponds to PBS.