Alright, what the bloody hell is this text, and Peter Galison, attempting to say?
> A black hole hoards images of the past. Light is composed of photons, and each one carries a bit of the image of whatever it hits. So when you see a tree the light hits the tree, bounces to your eye, and your brain eventually pulls it all together like a mosaic. Light stuck in a black hole’s gravitational pull can loop once, twice, or an infinite number of times, depending on its angle of approach, Lupsasca said. Those[sic] that finally escape in the direction of Earth carry a reflection of what the universe looked like when they entered the black hole’s pull. The longer light was held captive, the earlier in the past their image shows.
> “As we peer into these rings, first, second, third, etc., we are looking at light from all over the visible universe; we are seeing farther and farther into the past, a movie, so to speak, of the history of the visible universe,” said Peter Galison, the Joseph Pellegrino University Professor of the History of Science and of Physics...
Does anyone assert that any photon carries information about what it interacted with in the past? Does a photon have any information content other than its energy/wavelength? So "each one carries a bit of the image of whatever it hits" is total (ok, be nice now) poetic blarney.
So then, you observe some photons coming away from a black hole, how the AF are you supposed to be able to know which ones circled it once, twice, 42 times, whatever? And in what possible sense, Prof. Galison, is any of that like a "movie of the history of the universe"?
Of course photons carry information about whatever it interacted with. Photons don't have past (photons do not experience time). Photon's energy is the information as well as its polarization, direction from which it comes, etc.
Single photon is very small amount of information, you typically need more photons to get some kind of picture. But then when you think about it, the picture must have been carried by those photons so the information about the object IS in the photons you are receiving.
Now, photons coming away from black hole do not actually come from black hole. The photons that you can see are radiated by matter orbiting and falling into black hole. This carries no information from the black hole other than the geometry of the space (size/mass of the black hole), its angular velocity (angular momentum) and the charge which are the only measurable information about the black hole.
> the geometry of the space (size/mass of the black hole), its angular velocity (angular momentum) and the charge which are the only measurable information about the black hole.
This is true for a stationary black hole, i.e., one that never changes with time, but no real black hole is exactly stationary; every real black hole has matter and radiation falling into it. Non-stationary black holes can have more measurable information than just mass, spin, and charge.
I think his points are valid and shouldn’t be dismissed as “poetic blarney”. His words may inspire and other scientists to look into these details more and look for clues to decipher the photons that come from black holes. His statements are qualitative observations, just because something is not quantitatively measurable due to tech limitations doesn’t mean it should be dismissed (much of Einstein’s works was done in his head and confirmed decades later from experiments). Some advancements such as ability to watch for minuscule changes in stars and to determine planetary movements and sizes is probably a good example of some techniques that can be used for studying the photons that are coming off of black holes.. perhaps some day we may indeed be able to measure certain characteristics that give us clues to ancient photons.
> Does anyone assert that any photon carries information about what it interacted with in the past?
Yes. Each photon carries one bit of information. (I don't know if Galison intended "a bit" to mean literally a bit of information, but it's true.)
> Does a photon have any information content other than its energy/wavelength?
Yes, its polarization. In laboratory experiments in quantum information, photon polarization is what encodes the information.
> how the AF are you supposed to be able to know which ones circled it once, twice, 42 times, whatever?
By putting together the images conveyed by very large numbers of photons and looking for patterns in them. Yes, the information can be hard to decode, but it's there.
You can't measure all of these things simultaneously (the operators don't commute), so even if all these observables exist for a photon, they don't all carry independent information. You can only "read out" what you can get from one observable.
In most quantum information experiments, that observable is polarization (more precisely, polarization in some particular direction), which is an exact one-bit measurement result.
In radio telescope observations, it's typically amplitude and wavelength that are measured. I don't know that "the information content of a single photon" is really meaningful for such observations since they don't observe single photons, they observe the overall intensity at different wavelengths.
Even accounting for the uncertainty principle, the amount of information you can transfer in principle with a single photon is far more than one bit. For example, you can distinguish the single photon hitting a screen at more than two different times, or at more than two different places, or at more than two different energies.
> you can distinguish the single photon hitting a screen at more than two different times, or at more than two different places, or at more than two different energies.
You can't transfer information by any of these means because you can't control the time or the location or the energy of the photon hitting the screen.
The reason polarization is used to carry information in quantum computers is that you can control it, so you can use it to transfer information. But, as I said, a single photon can only be used to read out one bit in this way.
Measurements of wavelength don't measure single photons. They measure overall intensity, which amounts to making a single measurement that involves a huge number of photons.
Can you give a more technical description of "each photon carries one bit?" I don't doubt it, but it seems like a remarkable result!
The number of photons is itself a quantum observable (number operator) which is non-commuting with other observables. Does "each photon carries one bit" mean that if there are known N photons (eigenstate of number operator) there are N bits? Or is it saying something about distributions?
> Can you give a more technical description of "each photon carries one bit?"
The strictly technically correct version applies (as I've said in response to others) to polarization in a setting like quantum information or quantum computing experiments, where the polarization is controlled in various ways, and where measuring the polarization of a single photon at the end of the experiment corresponds to "reading out" one bit of the result.
For cases like the radio telescope observations of light from the black hole in the article this discussion was originally about, the "one bit" more or less corresponds to a "pixel" of the image that gets constructed from the data collected by multiple telescopes and combined by computers.
> The number of photons is itself a quantum observable (number operator) which is non-commuting with other observables.
There isn't a single "number operator" for photons. There is a number operator for each "mode" (or Fock state in more technical language), which means for each possible combination (k, u) of momentum k and polarization u.
Eigenstates of the number operator for each mode are also eigenstates of the Hamiltonian and the polarization operator, so it is not true that the number operators do not commute with any other observables.
Actual states in laboratory experiments are not always Fock states, but AFAIK they are close to it in quantum information and quantum computing experiments. However, the states being detected by the radio telescopes are not Fock states or anything close to it; they are coherent states, which are basically classical EM field states and are not eigenstates of the number operator, or indeed of any observable.
> Does anyone assert that any photon carries information about what it interacted with in the past? Does a photon have any information content other than its energy/wavelength? So "each one carries a bit of the image of whatever it hits" is total (ok, be nice now) poetic blarney.
An analogy to how we perceive things seems appropriate. The light from some source (i.e the sun) hits an object, the light rays are then reflected/transmitted/refracted through the surface based on various material properties. Your eye, acting like a pinhole camera, captures all the light rays from that surface, and then projects it onto the sensor reception surface (I think the retina?) creating the 'image' that we see.
So the specific way the light rays interacts with the surface is what creates the 'information content'. For example, we perceive a blue-colored surface because the surface properties (i.e roughness) will specifically reflect wavelengths associated with blue. A specular surface will reflect light back to the eye, creating a 'shiny' surface.
They shine a light source onto an object, take a picture, and then by processing the light rays in the photo, are able to generate the same scene but from the perspective of the light source. It's really astounding when you see it. They can do this reconstruction because the flow of light rays from the light source to the object, that is then reflected to the camera can be reversed (it follows Helmhotz Reciprocity - light ray that is reflected off an object is the same as the light ray that hits the surface that creates the reflected ray). So by processing the light rays in the camera image, they can reverse it, and calculate how the light source would have "perceived" the original scene.
In the case of the blackhole, I wonder if it's just that the light that enters the blackhole might loop around multiple times, but if it doesn't hit a surface, then the light that they can process/interpret would literally show what it reflected off before it entered the blackhole, which would be an image from the past.
If you know the exact position and direction of a photon coming out of the accretion disk, you could theoretically reverse out a trajectory and calculate what direction it approached the black hole from at what time.
Of course, that relies on incredibly high resolution images and knowledge of the black hole, but it's not impossible.
Of course light carries information about its past. Else you would not be able to tell information about the room you are in from the light that has been there.
As for how we tell how it traveled, light that is bent around the black hole and does not go into orbit makes a ring. Light that is bent around, loops once, then comes out makes a second ring. It is very hard for any light to come out between these rings. Ditto for 3x. And 4x. In theory, if we could see enough rings, we can look back as far as we want. In practice we can only make out so many. But comparing one ring with the next we can see back in the past. Finding out the delay tells us things about the black hole.
(It gets complicated, because it depends on the mass, the spin, and angle of the spin.)
We can look at light from old parts of the universe and understand the chemical composition at that place and time the light came from, and how it was moving relative to us (i'm guessing most people will know about red shift, not everyone will, it's essentially if something is moving away from us, the wavelength gets stretched or in other words the light gets redder.) We've seen examples, where light (ok space) is bent by a heavy mass and we've seen light from the same source arrive to earth via different paths. This approach with the black whole might allow us to do the same thing, but with light from the same object coming to us via an enormous number of paths, and having left the object at enormous number of different times, allowing us to simultaneously observe light from same object over millions of years
If you could collect enough photons and crunch them, you can imagine being able to reconstruct a map of matter moving and changing relative to the black hole over time. No idea if we can really do this, but its an exciting idea
That's one of the strangest, and angriest, questions I've seen on an HN science post.
> Does anyone assert that any photon carries information about what it interacted with in the past?
Yes. I do. The Cosmic Microwave Background encodes structure from many foregrounds in its wavelength, polarization, and momentum vector. Moreover, if one is able to build a well-motivated model for the photon's source and path of propagation, one can extract still more.
Grab a linear polarizer and play with it while you look at the daytime sky. The light you see is polarized differently depending upon the path it took to reach you from the Sun.
Look at your hand. Does the color of the light that reaches your eye not encode a convolution of the emission spectrum of the nearby light source and the reflection spectrum of your hand?
Regarding the black hole question, I'm no GR expert, though I am a professional gravitational physicist, but my guess is that the photons that make more orbits emerge from closer to the event horizon. The article discusses a notion of "shells", so I surmise that there may be favorable resonances in the photon-scattering cross-section at integer orbits.
Scientists are occasionally wrong about something, sometimes in spectacular ways, but when multiple scientists at top institutions all agree about a few basic physics concepts in a popular-physics article, there's a pretty good chance that they know what they're talking about.
> if one is able to build a well-motivated model for the photon's source and path of propagation, one can extract still more.
I think this is what the parent comment is getting at.
How are you supposed to build a model of a black hole given the source is universe-scaled and the time-line for receipt of any particular photon is probably(?) vast and our observations are distant.
Not to be flippant, but cosmology is essentially the science of dealing with these kinds of scales. It’s an entire field whose goal is to make questions like the above tractable.
Completely agreed. In the era of concordance cosmology, one can, if desired, use the relative abundance of the elements, the distribution of galaxies in the sky, and the distance to far-away supernovae and, with two free fit parameters (dark-energy and dark-matter density) predict the properties of the Cosmic Microwave Background anisotropy.
It's an incredible fact that one can do this -- predicting the properties of the photons emitted at the very first instant that the universe became willing to allow a photon to propagate freely.
A young Randall Munroe could have picked from a number of really well-motivated reasons to state, "Science, It works...", but he picked this one: https://xkcd.com/54/
Modern precision cosmology works really, really well.
That’s a very good point, and obvious when you point it out.
Still, I have a sneaking suspicion we are making flawed assumptions because our observations are from this point, and only this point, in space and time.
Just my opinion, and it’ll probably turn out to be flawed itself like most of my opinions ;)
If anyone is not a physicist but wants to dive a bit deeper into cosmology and quantum theory without insurmountable math - I can recommend PBS Spacetime[1] YouTube channel and their Discord community. I was pleasantly surprised to see the most updated information about the current knowledge, including the results from 2019.
A healthy second to this suggestion! The channel and community they've built is a high quality, informal classroom with some really thought provoking content. They actively engage with audience questions, provide "homework" in the form of challenge questions and are very welcoming. And amazingly, even during the pandemic, with their shooting from the hosts home, they didn't skip a beat. I've been a patreon supporter for awhile and it is the best damn value for my contribution.
Can VLBI be done with a single moving telescope, with a very precise clock?
I'm thinking of a single frequency point source in space. The received signal will be periodic. Thus we can move the receiver to a new point in space and still know what the received waveform would be at old position based on its periodicity (if we have an accurate clock). From there we can do interferometry?
A Fourier transform allows the same reasoning to be extended to a multi-frequency source. We can also extend the source spatially, in the same way an arbitrary hologram can be considered to be a collection of "zone plates" due to a set of point sources?
If the above is true, can we use a single Earth based telescope (with clock) to form a telescope with a diameter of the Earth's orbit over the course of a year?
Maybe the flaw is that it assumes that the scene being observed is static over the period of observation? How much change can be tolerated before a scene is no longer static?
So, maybe someone can explain in more concrete terms. The distance from us to the supermassive black hole in the center of the galaxy is 25k years. A photon traveling from us to the black hole, that happens to go around the black hole and come back would reach us in 50k years, or in 50k years plus a few million years due to the local time-space distortion around the black hole? If the photon goes around the black hole once and escapes and comes back, will it be back in 50k years, or in 1 billion years?
Do black holes appear the same no matter what direction you see them from? Every one of the theoretical views (and indeed the "real" one) are the same. I get the layman's description of why and how spacetime curvature causes it, but do they look the same if you view them from the plane of the accretion disk vs through the "poles" because of this?
35 comments
[ 3.3 ms ] story [ 87.8 ms ] thread> A black hole hoards images of the past. Light is composed of photons, and each one carries a bit of the image of whatever it hits. So when you see a tree the light hits the tree, bounces to your eye, and your brain eventually pulls it all together like a mosaic. Light stuck in a black hole’s gravitational pull can loop once, twice, or an infinite number of times, depending on its angle of approach, Lupsasca said. Those[sic] that finally escape in the direction of Earth carry a reflection of what the universe looked like when they entered the black hole’s pull. The longer light was held captive, the earlier in the past their image shows.
> “As we peer into these rings, first, second, third, etc., we are looking at light from all over the visible universe; we are seeing farther and farther into the past, a movie, so to speak, of the history of the visible universe,” said Peter Galison, the Joseph Pellegrino University Professor of the History of Science and of Physics...
Does anyone assert that any photon carries information about what it interacted with in the past? Does a photon have any information content other than its energy/wavelength? So "each one carries a bit of the image of whatever it hits" is total (ok, be nice now) poetic blarney.
So then, you observe some photons coming away from a black hole, how the AF are you supposed to be able to know which ones circled it once, twice, 42 times, whatever? And in what possible sense, Prof. Galison, is any of that like a "movie of the history of the universe"?
Single photon is very small amount of information, you typically need more photons to get some kind of picture. But then when you think about it, the picture must have been carried by those photons so the information about the object IS in the photons you are receiving.
Now, photons coming away from black hole do not actually come from black hole. The photons that you can see are radiated by matter orbiting and falling into black hole. This carries no information from the black hole other than the geometry of the space (size/mass of the black hole), its angular velocity (angular momentum) and the charge which are the only measurable information about the black hole.
Or by matter elsewhere in the universe, whose light got bent by the hole.
This is true for a stationary black hole, i.e., one that never changes with time, but no real black hole is exactly stationary; every real black hole has matter and radiation falling into it. Non-stationary black holes can have more measurable information than just mass, spin, and charge.
Yes. Each photon carries one bit of information. (I don't know if Galison intended "a bit" to mean literally a bit of information, but it's true.)
> Does a photon have any information content other than its energy/wavelength?
Yes, its polarization. In laboratory experiments in quantum information, photon polarization is what encodes the information.
> how the AF are you supposed to be able to know which ones circled it once, twice, 42 times, whatever?
By putting together the images conveyed by very large numbers of photons and looking for patterns in them. Yes, the information can be hard to decode, but it's there.
More than a bit. Between its wavelength, its presence (or absence), polarization, and timing, a photon can carry a great deal more than one bit.
In most quantum information experiments, that observable is polarization (more precisely, polarization in some particular direction), which is an exact one-bit measurement result.
In radio telescope observations, it's typically amplitude and wavelength that are measured. I don't know that "the information content of a single photon" is really meaningful for such observations since they don't observe single photons, they observe the overall intensity at different wavelengths.
You can't transfer information by any of these means because you can't control the time or the location or the energy of the photon hitting the screen.
The reason polarization is used to carry information in quantum computers is that you can control it, so you can use it to transfer information. But, as I said, a single photon can only be used to read out one bit in this way.
The number of photons is itself a quantum observable (number operator) which is non-commuting with other observables. Does "each photon carries one bit" mean that if there are known N photons (eigenstate of number operator) there are N bits? Or is it saying something about distributions?
The strictly technically correct version applies (as I've said in response to others) to polarization in a setting like quantum information or quantum computing experiments, where the polarization is controlled in various ways, and where measuring the polarization of a single photon at the end of the experiment corresponds to "reading out" one bit of the result.
For cases like the radio telescope observations of light from the black hole in the article this discussion was originally about, the "one bit" more or less corresponds to a "pixel" of the image that gets constructed from the data collected by multiple telescopes and combined by computers.
> The number of photons is itself a quantum observable (number operator) which is non-commuting with other observables.
There isn't a single "number operator" for photons. There is a number operator for each "mode" (or Fock state in more technical language), which means for each possible combination (k, u) of momentum k and polarization u.
Eigenstates of the number operator for each mode are also eigenstates of the Hamiltonian and the polarization operator, so it is not true that the number operators do not commute with any other observables.
Actual states in laboratory experiments are not always Fock states, but AFAIK they are close to it in quantum information and quantum computing experiments. However, the states being detected by the radio telescopes are not Fock states or anything close to it; they are coherent states, which are basically classical EM field states and are not eigenstates of the number operator, or indeed of any observable.
An analogy to how we perceive things seems appropriate. The light from some source (i.e the sun) hits an object, the light rays are then reflected/transmitted/refracted through the surface based on various material properties. Your eye, acting like a pinhole camera, captures all the light rays from that surface, and then projects it onto the sensor reception surface (I think the retina?) creating the 'image' that we see.
So the specific way the light rays interacts with the surface is what creates the 'information content'. For example, we perceive a blue-colored surface because the surface properties (i.e roughness) will specifically reflect wavelengths associated with blue. A specular surface will reflect light back to the eye, creating a 'shiny' surface.
A very cool demonstration of how light conveys information is in dual photography: https://www.youtube.com/watch?v=p5_tpq5ejFQ
And the original paper: https://graphics.stanford.edu/papers/dual_photography/https:....
They shine a light source onto an object, take a picture, and then by processing the light rays in the photo, are able to generate the same scene but from the perspective of the light source. It's really astounding when you see it. They can do this reconstruction because the flow of light rays from the light source to the object, that is then reflected to the camera can be reversed (it follows Helmhotz Reciprocity - light ray that is reflected off an object is the same as the light ray that hits the surface that creates the reflected ray). So by processing the light rays in the camera image, they can reverse it, and calculate how the light source would have "perceived" the original scene.
In the case of the blackhole, I wonder if it's just that the light that enters the blackhole might loop around multiple times, but if it doesn't hit a surface, then the light that they can process/interpret would literally show what it reflected off before it entered the blackhole, which would be an image from the past.
Of course, that relies on incredibly high resolution images and knowledge of the black hole, but it's not impossible.
As for how we tell how it traveled, light that is bent around the black hole and does not go into orbit makes a ring. Light that is bent around, loops once, then comes out makes a second ring. It is very hard for any light to come out between these rings. Ditto for 3x. And 4x. In theory, if we could see enough rings, we can look back as far as we want. In practice we can only make out so many. But comparing one ring with the next we can see back in the past. Finding out the delay tells us things about the black hole.
(It gets complicated, because it depends on the mass, the spin, and angle of the spin.)
If you could collect enough photons and crunch them, you can imagine being able to reconstruct a map of matter moving and changing relative to the black hole over time. No idea if we can really do this, but its an exciting idea
> Does anyone assert that any photon carries information about what it interacted with in the past?
Yes. I do. The Cosmic Microwave Background encodes structure from many foregrounds in its wavelength, polarization, and momentum vector. Moreover, if one is able to build a well-motivated model for the photon's source and path of propagation, one can extract still more.
Grab a linear polarizer and play with it while you look at the daytime sky. The light you see is polarized differently depending upon the path it took to reach you from the Sun.
Look at your hand. Does the color of the light that reaches your eye not encode a convolution of the emission spectrum of the nearby light source and the reflection spectrum of your hand?
Regarding the black hole question, I'm no GR expert, though I am a professional gravitational physicist, but my guess is that the photons that make more orbits emerge from closer to the event horizon. The article discusses a notion of "shells", so I surmise that there may be favorable resonances in the photon-scattering cross-section at integer orbits.
Scientists are occasionally wrong about something, sometimes in spectacular ways, but when multiple scientists at top institutions all agree about a few basic physics concepts in a popular-physics article, there's a pretty good chance that they know what they're talking about.
I think this is what the parent comment is getting at.
How are you supposed to build a model of a black hole given the source is universe-scaled and the time-line for receipt of any particular photon is probably(?) vast and our observations are distant.
It's an incredible fact that one can do this -- predicting the properties of the photons emitted at the very first instant that the universe became willing to allow a photon to propagate freely.
A young Randall Munroe could have picked from a number of really well-motivated reasons to state, "Science, It works...", but he picked this one: https://xkcd.com/54/
Modern precision cosmology works really, really well.
Still, I have a sneaking suspicion we are making flawed assumptions because our observations are from this point, and only this point, in space and time.
Just my opinion, and it’ll probably turn out to be flawed itself like most of my opinions ;)
[1] https://m.youtube.com/channel/UC7_gcs09iThXybpVgjHZ_7g
[1] https://www.patreon.com/pbsspacetime
I'm thinking of a single frequency point source in space. The received signal will be periodic. Thus we can move the receiver to a new point in space and still know what the received waveform would be at old position based on its periodicity (if we have an accurate clock). From there we can do interferometry?
A Fourier transform allows the same reasoning to be extended to a multi-frequency source. We can also extend the source spatially, in the same way an arbitrary hologram can be considered to be a collection of "zone plates" due to a set of point sources?
If the above is true, can we use a single Earth based telescope (with clock) to form a telescope with a diameter of the Earth's orbit over the course of a year?
Maybe the flaw is that it assumes that the scene being observed is static over the period of observation? How much change can be tolerated before a scene is no longer static?