Yes, you can prove a theory
You just have to be very careful about how you define "proof" and "theory". Which sounds silly, but it isn't really. If you accept that reality is an illusion, and/or that our senses cannot be trusted or give information which is always unreliable, that we cannot ever really know what reality is like, then it's true you can't prove a theory. But in that case you can't even have true facts either, only things which (at best) can be assigned probabilities.
But this is not typically how most of us think or how science normally works either. It operates under the assumption that we can actually objectively measure the world, even if our information about it is not complete. Our measurements do represent reality in some way. They may occasionally be wrong or people may lie about them and deceive us, but this is the exception rather than the rule. Thus, under this assumption, it is perfectly possible to measure things we may consider to be absolute, true facts. The only way to say the Earth is flat is to invoke a massive conspiracy theory and/or our memories are being constantly manipulated. This isn't wrong, but it isn't science.
Once you allow absolute facts, you also allow provable theories. Evolution has been proven to occur, the Earth has been proven to be roughly spherical. These are both theories, in that they tell us how the world works, and facts, in that they are known to be true. They have varying degrees of predictive power, and you have to specify precisely under what conditions they should occur - but if those conditions are indeed replicated precisely, then evolution will always happen and gravity will always operate.
It's true that most of the time things are not always so black and white - most theories are varying shades of grey. But the absolute extremes - the definitely true and the definitely false - do happen. "Alternative facts" are indeed simply lies.
https://astrorhysy.blogspot.com/2017/02/you-cant-not-prove-it-wasnt-me-who.html
Wednesday, 15 February 2017
Thursday, 9 February 2017
QA2 training
Today's ALMA QA2 training was relatively painless. Oh, it's a silly, badly-organised process in which files are given helpful labels like, "uid__Xf43e_2016.SB.00982.ms.split.cal" or worse, "textfile.txt", and finding the relevant information is like reading a dictionary when you know the meaning of the word but not the word itself.... BUT at least each individual step is not that bad. To tackle every possible eventuality you'd have to have years and years of experience, but that isn't going to happen most of the time. It's just a matter of writing a sensible set of ordered instructions (and using a Linux environment which FFS lets you minimise windows !)... there's light at the end of the tunnel and it's not an oncoming train.
So here is a short visual guide to observing with ALMA. Totally 100% accurate.
Wednesday, 8 February 2017
Not amused
Imagine, if you will, an enormous online technical manual from which you need to extract a few key points. Some of these are easily visible but most are hidden in obscure hyperlinks. You can control+f but not search the entire site in any way. Some points are labelled in an obvious way but others require reading and understanding to extract the relevant detail. You'll need at least a dozen different pages from which to extract all the information, but maybe not all at the same time and you can't minimize windows or have multiple desktops (because frak you, that's why). Some information can't be extracted directly from the manual but only by running a series of long, complicated tasks which then require you to check yet more extremely long web pages that are incredibly poorly-labelled and use a program which just plain doesn't work for no reason. The result of all this is access to data which doesn't interest you and you're not allowed to use in any case, and the absolute best you can hope for is that someone will eventually write you a thank-you email.
That's what ALMA QA2 is like.
Nope nope nope.
That's what ALMA QA2 is like.
Nope nope nope.
Tuesday, 7 February 2017
TMI
I will never, ever again fall back on the old adage, "in the age of information, ignorance is a choice". Anyone quoting that has never read the ALMA CASA cookbook. Or the example training PDF. And the less said about the vast, stupendous data reduction wiki, the better.
In the age of information overload, I need more tea. Much more tea.
In the age of information overload, I need more tea. Much more tea.
QA2 begins
This week I'm in Garching learning ALMA QA2 (quality assessment level 2) procedures. All we have to do is determine if the data have the correct sensitivity and resolution. Which is like saying that in order to go to the Moon, all you have to do is build a big enough rocket...
Wednesday, 1 February 2017
Preparing for the VLA
So I've submitted my "scheduling blocks" to the VLA to get better resolution images of these optically dark hydrogen clouds in the Virgo cluster. With Arecibo they are single pixels : we get their maximum size and direct measurements of their position, mass and velocity width, but that's all. That's enough to say, "these little buggers are very interesting", but not enough to determine what the hell they actually are. We can run simulations for years and years, but we'll never really know without more data.
These observations will get us resolution that's about 4x better than Arecibo - not a huge gain, but potentially enough to distinguish between possible explanations. Given their mass we can work out a lower size limit at which they should have a density high enough to form stars - if they're close to that we'll know if we've caught them at a strange phase in their evolution just before they ignite star formation. In that case we might consider searching for molecular hydrogen. Or they might be more extended, rotating objects which could be stable on long timescales, in which case that's probably a Nature paper. Tonnes of stuff we can potentially get from this.
Of course, we'd like even better resolution, but that's not so easy. Interferometers like the VLA have sensitivity effects that single-dishes like Arecibo just don't have - making them (in some situations) far worse than due to the difference in collecting area. So their sensitivity to diffuse gas is ~100-1000x lower, and while these clouds were detected at Arecibo in 5 minutes of observations, we've requested 2 hours for each one at the VLA. Even then, since we don't know the true structure of the clouds, we're not guaranteed to detect anything. Fingers crossed...
Reading the VLA observing guide and links therein has been a less than fun affair, much like reading an old computer manual. Great if you need reference info, not so good if you need to learn stuff for the first time. The sheer amount of information is mmmmwwwaaaaaarrrgh. It takes a lot of getting used to the different terminologies : "scheduling block" really mean "observing script", "resource file" really means "telescope configuration file". I wouldn't say these are particularly intuitive labels and it makes mentally processing the information very much harder than it needs to be.
On the positive sides, the observing preparation tool is very nice. It's a simple, reasonably intuitive GUI webpage rather than a scripting language like Arecibo uses (though Arecibo does have an interactive GUI for doing the observations too). And even better, the NRAO helpdesk is genuinely very, very helpful, responding to my inept inquires in a few hours and in great detail.
And so now we play the waiting game, until the data arrives and I have to learn how the heck to process it.
These observations will get us resolution that's about 4x better than Arecibo - not a huge gain, but potentially enough to distinguish between possible explanations. Given their mass we can work out a lower size limit at which they should have a density high enough to form stars - if they're close to that we'll know if we've caught them at a strange phase in their evolution just before they ignite star formation. In that case we might consider searching for molecular hydrogen. Or they might be more extended, rotating objects which could be stable on long timescales, in which case that's probably a Nature paper. Tonnes of stuff we can potentially get from this.
Of course, we'd like even better resolution, but that's not so easy. Interferometers like the VLA have sensitivity effects that single-dishes like Arecibo just don't have - making them (in some situations) far worse than due to the difference in collecting area. So their sensitivity to diffuse gas is ~100-1000x lower, and while these clouds were detected at Arecibo in 5 minutes of observations, we've requested 2 hours for each one at the VLA. Even then, since we don't know the true structure of the clouds, we're not guaranteed to detect anything. Fingers crossed...
Reading the VLA observing guide and links therein has been a less than fun affair, much like reading an old computer manual. Great if you need reference info, not so good if you need to learn stuff for the first time. The sheer amount of information is mmmmwwwaaaaaarrrgh. It takes a lot of getting used to the different terminologies : "scheduling block" really mean "observing script", "resource file" really means "telescope configuration file". I wouldn't say these are particularly intuitive labels and it makes mentally processing the information very much harder than it needs to be.
On the positive sides, the observing preparation tool is very nice. It's a simple, reasonably intuitive GUI webpage rather than a scripting language like Arecibo uses (though Arecibo does have an interactive GUI for doing the observations too). And even better, the NRAO helpdesk is genuinely very, very helpful, responding to my inept inquires in a few hours and in great detail.
And so now we play the waiting game, until the data arrives and I have to learn how the heck to process it.
Tuesday, 31 January 2017
The velocity function of missing satellites
A very intriguing paper on astro-ph yesterday, submitted to ApJ but not yet accepted.
This is yet another one attempting to solve the missing galaxy problem. The problem is that at low masses there are far fewer galaxies than simulations predict. Simulations generally only use dark matter since it's computationally cheap and normal matter only accounts for about 10% of the total mass anyway. This means it can be tricky to predict what the observable mass (i.e. of normal matter) of the simulated galaxies would really be. So people have started predicting their rotation velocities instead. This should depend much more strongly on the dark matter mass and less on the normal matter, so the prediction should be more robust. Past simulations have found there's still a great big problem, with galaxies of low velocities (i.e. low total mass) being much too numerous in the simulations compared to reality.
This paper compares simulations that use only dark matter with ones that use normal matter too. They find that the physics of the gas can explain the discrepancy very well. In their model, it turns out that the gas discs don't trace out the full size of the dark matter halos, so the rotation that would actually be measured would be lower than the true maximum of the halo. Also, the gas mass in many of the smaller halos (which rotate less quickly) would be so low it would be undetectable. That brings the number of small, slowly-rotating galaxies into good agreement with the observations.
It's a very interesting result and definitely one to watch as it evolves with the the referee reports. One thing they barely mention is the "too big to fail" problem, where there seems to be missing galaxies which are so large there's no way they should have been able to avoid forming stars. A potentially more serious flaw is (if I understand them correctly) that they select galaxies only if they have a gas mass above a certain threshold. If so, this will introduce a weird selection effect because real observations aren't like that. The observed "brightness" of the gas depends on its apparent velocity width as well as its mass. So I'm not convinced their results are really directly comparable to observations, but this should be an easy problem to address.
https://arxiv.org/abs/1701.07835
This is yet another one attempting to solve the missing galaxy problem. The problem is that at low masses there are far fewer galaxies than simulations predict. Simulations generally only use dark matter since it's computationally cheap and normal matter only accounts for about 10% of the total mass anyway. This means it can be tricky to predict what the observable mass (i.e. of normal matter) of the simulated galaxies would really be. So people have started predicting their rotation velocities instead. This should depend much more strongly on the dark matter mass and less on the normal matter, so the prediction should be more robust. Past simulations have found there's still a great big problem, with galaxies of low velocities (i.e. low total mass) being much too numerous in the simulations compared to reality.
This paper compares simulations that use only dark matter with ones that use normal matter too. They find that the physics of the gas can explain the discrepancy very well. In their model, it turns out that the gas discs don't trace out the full size of the dark matter halos, so the rotation that would actually be measured would be lower than the true maximum of the halo. Also, the gas mass in many of the smaller halos (which rotate less quickly) would be so low it would be undetectable. That brings the number of small, slowly-rotating galaxies into good agreement with the observations.
It's a very interesting result and definitely one to watch as it evolves with the the referee reports. One thing they barely mention is the "too big to fail" problem, where there seems to be missing galaxies which are so large there's no way they should have been able to avoid forming stars. A potentially more serious flaw is (if I understand them correctly) that they select galaxies only if they have a gas mass above a certain threshold. If so, this will introduce a weird selection effect because real observations aren't like that. The observed "brightness" of the gas depends on its apparent velocity width as well as its mass. So I'm not convinced their results are really directly comparable to observations, but this should be an easy problem to address.
https://arxiv.org/abs/1701.07835
Friday, 20 January 2017
Check out my kinky curves
So here it is, my eighth paper as first author. It's very similar to the sixth, except that it's much better because it's half as long. I'll have a detailed blog post up in a few days, but for now here's the super-short version.
Most neutral hydrogen gas (HI, pronounced H-one) is associated with optically bright galaxies, but there are a few weird gas clouds that aren't. In particular, there are these six HI blobs in the Virgo cluster that look like they're rotating as fast as massive galaxies - but optically they're dark. They're miles and miles away from any other galaxies and there's no sign of any extended HI streams anywhere nearby. So the most obvious explanation - that they were just ripped out of ordinary galaxies as they fly past each other - has problems.
What's particularly weird about these clouds is their apparent rotation. We don't really measure rotation directly, just how fast the gas is moving along the line of sight. The difference between the fastest and slowest-moving gas gives us a line width, which for normal galaxies it's safe to assume represents rotation. If that's the case for these clouds, they need a fairly substantial dark matter halo to hold them together, because they're rotating so fast that their gas mass is nowhere near large enough to stop them flying apart. They would effectively be galaxies without any stars.
But they might not be rotating at all. It's possible the high line widths are really just due to streaming motions with the gas flowing at different velocities in the same direction. Other people have claimed from numerical simulations that this explanation works well for some other, similar features. And it does. The problem is this has been widely taken to mean that's definitely the best explanation for any and all HI clouds in any situation.
To test this, we numerically simulated a spiral galaxy falling into a cluster. The target spiral is based on a fairly typical known object and we varied its parameters quite a lot to see how much difference that made (and also its trajectory through the cluster). The spiral has gas, stars, and dark matter. The galaxy cluster is much simpler : 400 point masses all buzzing around just like in a real cluster. Those point masses just have gravity - they don't have their own gas and stars because that's computationally very demanding indeed. This is a big improvement on the earlier works though, because they just did one point mass flying past a target galaxy.
What we show here is that the tidal debris/streaming motion hypothesis categorically, decisively, does not work for objects like these clouds. I would even go so far as to say it fundamentally cannot work. It's easy enough to make large clouds with high velocity widths, but's damn near impossible for clouds as small as those we see in Virgo. We saw gas being torn off the galaxies into long streams, and those streams get "harassed" into small clouds, which is nice... but there's a problem. The greater the streaming motions within a stream, the harder they are to detect and the quicker they disperse. It's incredibly difficult to disperse the rest of the stream while preserving the features of the highest velocity widths.
We also examined the formation mechanism of the famous VIRGOHI21. This well-known dark galaxy candidate is a very sharp "kink" in the velocity of a long stream from a spiral galaxy, which again looks like it might be rotating. This is what earlier works were trying to reproduce. We show that actually they didn't really do this, even though they claimed to, but our simulations did. That might sound like petty bickering, and it is. But it's important because we can now very clearly say if a cloud is likely to be tidal debris or not. If it's in a stream, then that's probably a good explanation. If it's isolated, then that's only a sensible explanation if the cloud is rather large or has a low velocity width. If it's isolated but small and with a high velocity width, tidal debris is a lousy explanation.
As for VIRGOHI21 itself, I would say that after more than a decade, jury's still out. Much to my annoyance, and despite lengthy explanations in the paper, I wasn't able to convince the referee that we don't really understand it. It's true that we can explain the velocity kink... but that's all we can explain. We can't easily account for the rest of the features of the system, and we haven't really tested the dark galaxy hypothesis - so we don't know how well that explanation actually works. Despite the referee's abject protestations that "any scientist" should know that the success of one model not preclude the success of another, they fell for this very same fallacy when I added in a footnote about the alternative VIRGOHI21 hypothesis. And because I'm not insanely belligerent, I caved in and took the footnote out. Fortunately I don't have to do that on social media... :)
A much more detailed write-up can be found here.
https://arxiv.org/abs/1701.05361
Most neutral hydrogen gas (HI, pronounced H-one) is associated with optically bright galaxies, but there are a few weird gas clouds that aren't. In particular, there are these six HI blobs in the Virgo cluster that look like they're rotating as fast as massive galaxies - but optically they're dark. They're miles and miles away from any other galaxies and there's no sign of any extended HI streams anywhere nearby. So the most obvious explanation - that they were just ripped out of ordinary galaxies as they fly past each other - has problems.
What's particularly weird about these clouds is their apparent rotation. We don't really measure rotation directly, just how fast the gas is moving along the line of sight. The difference between the fastest and slowest-moving gas gives us a line width, which for normal galaxies it's safe to assume represents rotation. If that's the case for these clouds, they need a fairly substantial dark matter halo to hold them together, because they're rotating so fast that their gas mass is nowhere near large enough to stop them flying apart. They would effectively be galaxies without any stars.
But they might not be rotating at all. It's possible the high line widths are really just due to streaming motions with the gas flowing at different velocities in the same direction. Other people have claimed from numerical simulations that this explanation works well for some other, similar features. And it does. The problem is this has been widely taken to mean that's definitely the best explanation for any and all HI clouds in any situation.
To test this, we numerically simulated a spiral galaxy falling into a cluster. The target spiral is based on a fairly typical known object and we varied its parameters quite a lot to see how much difference that made (and also its trajectory through the cluster). The spiral has gas, stars, and dark matter. The galaxy cluster is much simpler : 400 point masses all buzzing around just like in a real cluster. Those point masses just have gravity - they don't have their own gas and stars because that's computationally very demanding indeed. This is a big improvement on the earlier works though, because they just did one point mass flying past a target galaxy.
What we show here is that the tidal debris/streaming motion hypothesis categorically, decisively, does not work for objects like these clouds. I would even go so far as to say it fundamentally cannot work. It's easy enough to make large clouds with high velocity widths, but's damn near impossible for clouds as small as those we see in Virgo. We saw gas being torn off the galaxies into long streams, and those streams get "harassed" into small clouds, which is nice... but there's a problem. The greater the streaming motions within a stream, the harder they are to detect and the quicker they disperse. It's incredibly difficult to disperse the rest of the stream while preserving the features of the highest velocity widths.
We also examined the formation mechanism of the famous VIRGOHI21. This well-known dark galaxy candidate is a very sharp "kink" in the velocity of a long stream from a spiral galaxy, which again looks like it might be rotating. This is what earlier works were trying to reproduce. We show that actually they didn't really do this, even though they claimed to, but our simulations did. That might sound like petty bickering, and it is. But it's important because we can now very clearly say if a cloud is likely to be tidal debris or not. If it's in a stream, then that's probably a good explanation. If it's isolated, then that's only a sensible explanation if the cloud is rather large or has a low velocity width. If it's isolated but small and with a high velocity width, tidal debris is a lousy explanation.
As for VIRGOHI21 itself, I would say that after more than a decade, jury's still out. Much to my annoyance, and despite lengthy explanations in the paper, I wasn't able to convince the referee that we don't really understand it. It's true that we can explain the velocity kink... but that's all we can explain. We can't easily account for the rest of the features of the system, and we haven't really tested the dark galaxy hypothesis - so we don't know how well that explanation actually works. Despite the referee's abject protestations that "any scientist" should know that the success of one model not preclude the success of another, they fell for this very same fallacy when I added in a footnote about the alternative VIRGOHI21 hypothesis. And because I'm not insanely belligerent, I caved in and took the footnote out. Fortunately I don't have to do that on social media... :)
A much more detailed write-up can be found here.
https://arxiv.org/abs/1701.05361
Wednesday, 14 December 2016
Maths is Christmassy too !
Some festive algebra spotted Cardiff University's astronomy PhD student's office, because just equations are super Christmassy !
Tuesday, 13 December 2016
Veeeerry accurate galaxy model in the Cardiff office.
Veeeerry accurate galaxy model in the Cardiff office. It's got a red central bulge and blue star-forming spiral arms. It's a bit warped, but that's because it's probably interacting with a massive companion (most likely angelic baby Tom Selleck Jesus on the tree).
Subscribe to:
Posts (Atom)