Tuesday, May 4, 2010

Invisible dust around spiral galaxies

In a recent post at the Galaxy Zoo Blog, Bill Keel wrote about the research on the dust distribution in galaxies, as studied by the fortunate existence of overlapping galaxies. One of those was UGC 4008, with the added specialty that it's not the bright part of the galaxies overlapping, but a dark part of the spiral (the left side of the upper, small galaxy is darkened and reddened by dust from the galaxy in front):

Now this is only visible because the bottom galaxy is in front of the top galaxy. We would have been unable to see this dark dust by any other means... (at least in these images, as an infrared survey might pick up thermal radiation from the dust)

If you look closely at the 'dark dust ring' part of the galaxy, you see that there is more noise in the image there. Seeing this might require adjusting your monitor to have good contrast. We can hypothesize that the noise is a result of real light coming from the dust disk, that has just about the intensity required to reach into significant signal-to-noise ratios. As such, we can integrate spacially by hand, effectively accumulating the radiation picked up by the separate pixels.

To do this using Mathematica or any graphics editing program, it is pretty trivial to look at the noise by substracting the image from a blurred (0,5 <= radius <= 1) version of itself, and then stretching the contrast:

As you can see when you compare the above image to the original, this does a pretty good job of showing the 'pixel dust area' in the picture. In fact, the 'pixel dust' of the bottom galaxy clearly reaches out to the top galaxy, what wouldn't have been obvious in the original image. It also shows that the top galaxy also has a dust disk or halo.

Basically, this method works as an edge-detection system for diffuse objects, as it is brightest where the signal is only just entering the noise. Another way to get this final bit (or subbit, rather?) of data out of the original image is by using a blur in which only values are added, which gives you the following image:

This method effectively just amplifies and smears the data out, but shows clearly the extend of the bottom galaxy. That the effect is real, is supported by the fact that there's no pixel dust to the bottom right of the bottom galaxy, and there's actually a pretty sharp edge where there is to where there isn't any pixel dust. Although in part that might come from the signal dropping below any measurable value, it might be expected to be a more gradual decay.

Since there will be good images of UGC 4008 soon, it will be nice to compare the real dust to the pixel dust above..

Thursday, November 5, 2009

Mirror, mirror, on the wall, I give you momentum, that is all!

There's more to do with mirrors. What happens when you look at yourself, in a mirror, while you're moving?

Special relativity has a lot to say on the subject. To begin with, you can't easily move faster than the speed of light, so your photons will hit the mirror earlier than you do (or fly by the mirror). That basically means that you can see yourself. Nothing strange there, yet. As you are supposedly moving at a respectable fraction of the speed of light, you're doing a pretty good job keeping up with you photons. Any photons hitting the mirror will have left you some while ago, more than they would've if you stood still. It's the opposite of running into the field of view of a camera after pressing the self-timer: in this case you run towards the mirror showing you an old image. The quicker you run, the older the image. By the way, you will still be equally far away from the mirror as you were when the photons left, so be sure to bring a telescope of some sort..

What else happens when you move towards a mirror? I covered the fact that you actually interact with yourself - or to be more precise - your past self. Now, since you're moving really quickly, your photons will be Doppler shifted. In this case of moving towards the mirror, they'll be blue shifted. The mirror will nicely reflect the blue-shifted photons, with you ending up seeing a bluer version of yourself. Now let's suppose you're an atom.

Atoms have their energy bands nicely quantized, so not all energy levels are allowed. An atom can switch between energy levels by emitting or receiving a photon of the corresponding energy difference. So, suppose a theoretical atom with three energy states 0, 1 and 2, with 0 having the lowest energy and 2 having the highest energy. Suppose the atom is in the 1 state, and is moving rather quickly towards a conveniently located mirror. The atom, in its rest frame, drops to the 0 state by emitting a photon, conveniently in the direction of the mirror. As the mirror is in the direction of travel, this means the photon is blue shifted with respect to the mirrors rest frame, and the atom loses a tiny bit of momentum. No sweat though, the atom was going fast enough for all relativistic purposes. At some point, the blue-shifted photon hits the mirror and is reflected back towards the atom. The mirror gained a bit of momentum doing this, but the photon still has equal energy content. As the photon meets its parent atom, they find that the blue shift allows the atom to absorb the photon, and jump directly from state 0 to state 2. It does cost a little bit of momentum, as this absorption slows the atom a tiny bit further down. It is, however, in a higher energy state than before the interaction through the mirror.
As a question to you, my dear reader: how is this agreeable with energy conservation (though I expect it is perfectly fine, through the exchange of momentum), and how does this work in terms of entropy?

Interestingly, this means that atoms might be able to ionize themselves (instead of eachother) through mirroring the radiation they emit. What about a stream of atoms coming at you, which you can turn into a plasma by aiming a mirror at them? Might it be possible to brake incoming plasma by using mirrors, as plasma responds to a wide range of frequencies? What else can be done using mirrors?
As is proposed in this great article, it might be said that photons are only a concept, that the interaction is direct, between emitter and absorber, the two being null-separated over a single world line. In this case, the atom interacts with itself (through the mirror). What is interesting, is that the atom in the example has a relative velocity with respect to its reflected self, resulting in a net energy gain. In the mean time, some momentum was transferred to the mirror. Basically, some of the 'virtual' kinetic energy the atom has due to its reflection 'moving towards it', is converted into potential energy in the atom, by means of transferring momentum (only!) to the mirror. That is, the mirror directly converts kinetic energy into potential energy!
I leave calculating the necessary speeds for various atoms and energy levels to the reader...

Memories from the past

You might have seen the 2006 movie called Deja Vu. In it, the government found a way to look exactly 9/2 day back in time, and it's used to prevent some generic terrorist attack (the main character also falls in love, but this is omitted in my summary due to it being part of standard conditions for movies). Of course, controlling a wormhole for these purposes seems far-fetched at the moment, but let's think about the looking-a-fixed-time-into-the-past part.

That ain't fiction. Please take a moment (no pun intended) to look at your nearest mirror. You might see yourself, looking at you, but that's not necessarily so. The image in the mirror is a few nanoseconds delayed (in air and vacuum, about 3 nanoseconds per meter), so it's showing you in the past. Unfortunately, the mirror is really close by, so you won't be able to look at that itching spot on your back or see yourself looking sideways. It seems to be an image of the present, for all general purposes.

So, what delay do you need in order to find some actual use for it? I don't think you need to go that far, and mirrors are both cheap and versatile. To look at the back of your head though, you need a delay of about a second at the very least.. With the current speed of light, that means placing a mirror on the moon. Or, to be actually able to use it without needing any instrumentation, turn the moon into a mirror (of the magnifying kind, preferably). This is less feasible than it might look, not in the last place because there are people who prefer looking at the moon instead of looking at themselves. It's a pretty egocentric to do, actually.

Looking into the past might not be very suitable for direct human consumption then. But what about computers? We broadcast huge amounts of information, and send huge amounts of information around the internet. For computers and other information handling equipment, a second is a long time. Two seconds is even longer. What if we place a data relay on the moon? It would be limited by the maximum amount of information sent by the relay, but it does mean the ultimate in information storage capacity: it can store everything you send! However, it'll only hold it for about two seconds though, and not more or less. The exact delay is time and location dependent, etc. etc. It is, however, a perfectly honest kind of memory. There's no way there will be tempered with it, and it's not actually stored anywhere. Plus, if you have the absolute need, then you can keep bouncing information back and forth, up to a maximum capacity of the upload rate times about two seconds. I call it Short Term Access Memory, or STAM.

What might be the use of such delay memory, you might be wondering? The key here is completeness. There is absolutely no way to filter any information that has been sent, assuming a decent relay station. Even if it is controllable, that is only possible ahead of time. The bounced data is free from any censorship. It might even be possible to create the memory using real mirrors, as is currently used for measuring the exact distance to the moon. I doubt that it's energy-efficient, but it has a virtually endless storage capacity..

Two seconds might be enough for computers to figure out whether there was important stuff to
get a hold of. For example, all e-mails sent through a server might be loaded to the STAM, while a computer has time to check the names of the senders. If it finds , it still has time to fetch the content of the e-mail. That might end up being cost-efficient. By the way, if computers can cope with a ~250 millisecond delay STAM, then it's not even necessary to place a relay on the moon: it can be placed in geostationary orbit, at a conveniently located fixed spot in the sky. The technology is there, both for active relaying and passive mirroring of data. So why hasn't this been done before? It's a matter of time, I suppose..

edit: Some theories suggest your brain uses this already, especially for the repetition of audible signals. Signals along nerve cells travel so slowly, that a delay of some part of a second is pretty feasible, especially by relaying along the spine...

Wednesday, March 18, 2009

Optimal product lifetimes

Disposable products probably are bad for the environment, because it is likely that they could be used again without much alteration. Take for example a bottle of juice: when it's empty it still is perfectly capable of containing fresh juice, although it might need some cleaning before re-use. It usually is dumped in a bin however, and at best it will go through the whole recycling machine before it arrives at the next consumer. There probably is a better solution here.. (like bringing your own bottle and cleaning it at home every day). This suggests that the longer the lifetime of the product, the better.

Very long lifetimes however, might not be the most efficient either. Suppose that your computer is able to operate perfectly for the upcoming fifty years or so. It might be outdated today, but in fifty years it will be so horribly outdated that the only useful usage will be as an inefficient room heater.. And that is a problem in itself, for two reasons: the overall usage of natural resources is higher, because old products still work, and are less quickly recycled than they would've been otherwise. (Note that this is true as soon as you replace it later than would be useful in terms of recycling, which might be much earlier than it's actual lifetime. Note also that I assumed perfect recycling, something we certainly don't achieve at this moment in time). The second reason why it's bad to have products lying around which aren't very useful anymore, is their energy use. How many homes today still have an old 300+ watt desktop computer (screen included) around, while the newest double flat screen quad-core monster uses less than 200 W at full force? In this case, innovation is slowed down because of the 'old' products still working. This effect is easy to predict by the way: imagine what will happen when electric vehicles are launched at a large scale in the next few years. They will enter the market far more easily if a lot of the current cars suddenly fail in an inexplicable way... or the regulations are altered to render a lot of cars useless. We still end up with the recycling issue, but in terms of energy efficiency that would be beneficial.

There's another way in which innovation might be slowed by long-lasting products, on the production side of things. If there are few products to be sold, because the old ones break down, there's not much capital available to invest in research and development. In such a market, research and development must come (all but) solely from indepently funded sources, like universities. There most likely will only be new players on the market, since businesses 'dry out' after supplying the majority with the improved product.
Extrapolating a bit, most consumer goods producers will dry out in this way, while markets for consumables like food and entertainment might remain largely unaffected...

It appears that there's an optimum in product lifetime, determined in a large part by the characteristics of innovation in the market. A nice example might be cell phones, which tend to break down right before they are outdated. It's one of the most innovation driven markets I can imagine right now, and a nice test case for the optimal lifetime conjecture posed here. Now all we need is the recycling capabilities necessary to make it sustainable...

Thursday, March 5, 2009

Continuous bank accounts

Ever since humans started trading goods, transactions were (more or less) instantaneous events. Even today, seemingly continuous transactions, like wages and insurances, are transformed into repeating instantaneous transactions, mostly once every month or year.
Even interest, which is continuous per definition, is averaged over a year and returned in one event.
Doesn't this seem strange to you?

Well, it does to me. There's a purely historical argument for keeping it like this, and that's 'because we've always done it this way'. That doesn't cut it, does it? Then there obviously might be an economic reason, ie. it is the cheapest method around. But is it? If you assume every transaction costs a certain amount, then obviously less transactions is better. Any payment method that involves a transaction cost with a nonzero starting price for any transaction will result in as few transactions as possible, as will any method which becomes relatively cheaper when the transactions are higher.
This logically lead to transactions occuring annually whenever possible, like taxes and insurance policies, monthly when annually really is inconvenient, like wages and temporary things (phone bills, etc.), and instantaneous when it's a one-time transaction, such as shopping transactions.
I think we can do with less transactions than that, and make some more improvements on the way..

To do this, let's make our bank accounts time-dependent. Let's have a look at some examples. First off, income. Let's suppose you earn 31536 $/year, ignoring bonuses and the like. As things are now, you'd receive 2628$ every month (I'm aware of the usual notation, it's just that monetary values are the only ones in which the unit is written in front of the numerical value..). With a time-dependent bank account, you'd receive one milli-dollar every second. If you think that sounds small, remember there are 3600 seconds in an hour, and you'll sleep about one-third of the time.

At this point, we've resolved all the 'it's close to the end of the month so I won't have money to buy food'-issues. That might not be enough to convince you though. So let's do another example: fixed expenses for housing.
Nowadays, every month you receive your wage one day, and you'll need to pay your fixed expenses the next.. Although you could automate this, in the time-dependent bank account this hassle wouldn't exist. Let's suppose you'd now pay 1000 $/month for the house you live in. In the new system, you'd pay about 381 micro-dollar per second, or 381 µ$/s. That means that you wouldn't receive 1000µ$/s from your wage, but that you would receive only 619 µ$/s. In other words, the money you would pay for your house passes by your bank account. It never actually resides on your account, so you'll not need to worry about it either. In the meanwhile, this speeds up the economy, because there's no delay from you having to arrange the transaction (whether automated or not).

You might start to become afraid of the microdollars and prices per second. Note though, that you can still use the normal ways to communicate about it: 1000 µ$/s is still equal to 2628 $/month, so you can still use all 'normal' values in your communication. In fact, you might not even notice the change if a bank would migrate to the new system. You would still see the amount of money you have available on your account, the only difference being that the amount has risen a bit after refreshing your browser...

So is it necessary to change the whole world in order to get this to work? Actually, the system would be perfectly backwards compatible. A now instantaneous transaction becomes an up-to-one-second transaction. It might still be instantaneous. It's lower than the frequency with which you can refresh your account info anyway, so it might not even matter as long as banks are quicker than you are. This also opens up possibilities: if you have only 500$ left on your bank account (yes that would be 500 $ + .001*t $ in the current example) but you really need a new washing machine of 700$, then instead of putting you at -200$ the bank might decide to put you down to 0$ for the next 200000 seconds (that sounds like an eternity, but it really is only 2.3 days). This might mean that you're officially broke at this point, but what if you need to buy some food? You'd like to spend 10$ right now, while there's no money on your account. There seems to be no way out: you'll have to spend the next few days hungry, or do you? Do you need to hunger now because the store where you bought the washing machine needed your money at that exact moment? They probably didn't need it at that moment, just in the month after buying. So the store demands its rightful money over a one month period, in this case meaning that there's a limit of about 266 µ$/s (based on the full 700$). That's no problem at all! You still had 619 µ$/s left from your wage, remember? You still have your 500$ left on your account, so the immediate eating won't be an issue either. Let's suppose you have paid the store the 500$ immediately, and there's 200$ left to be paid over the next month. This means you'll be paying the store 76 µ$/s for the next month (where the monetary month is redefined to be 2.628 million seconds, to prevent everyone from buying stuff in February), leaving you with a 0$ + 543 µ$/s bank account. If the grocery store where you're trying to buy your 10$ meal has a paying period of over ~5 hours, that won't be a problem at all.
This might seem overly complex, but in reality you won't notice much of it. The machine will still tell you that you've a) paid or b) have too little money. For stores however, this gives freedom to choose how they like to be paid. As in the example above, it is most convenient to smear out the whole payment over the maximum period, so it is subtracted from your wage during that period. This actually means that you receive less money from your job, as the money is directly transferred to the store, and you do your job for free while you've received a washing machine from the store.
This might sound a lot like the direct trading systems of early civilization, doesn't it? Maybe the new system is more natural...

So, the system is still fully backwards compatible, only now stores have the convenience of allowing people to buy stuff on a longer timescale directly through their own bank accounts, instead of the credit way sometimes used for larger sales nowadays. Yet another way to battle eachother at! Also, spreading the income in times of recession might be a very smart way to work through a difficult period.

Finally, interest. It becomes possible to return interest continuously. This would mean that there's a bonus added every moment for the amount available on the moment before. On the other hand, it's common usage to have no interest at all on your pay account, and I think that would be the most logical choice in this case as well.

It would leave everyone with lots of choices though, because every payment could become a function of time... it might be possible to create some interesting (no pun intended) constructions.

More interesting still, is the short-circuiting effect of time-dependent banking. Consider a butcher and a cook: the cook buys his meat from the butcher, while the butcher dines at the cooks restaurant. They pay eachother significant amounts of money each month, but a lot of it cancels with the new system. If the monetary values are equal, only a net tax flow remains...
This effect manifests itself on a much larger scale as well: companies pay taxes to the government, the government pays it's officers and other employees, who buy stuff from the companies. Effectively, those employees just pay less.

So, using time-dependent banking can reduce a lot of transactions into single continuous transactions, allowing for much more flexible payment options. It might also create new risks and reduce the transparency of the market, but due to the backward compatibility it might allow for effective countermeasures like better and more open information systems and lending/spending rules. On the other hand, because of the cancelling effect of overlapping income and payments, people might be more aware of their actual wealth, thus reducing the chances on not reaching the end of the month...

Sunday, March 16, 2008

What's next?

Basically, all new inventions occupy only three categories: convenience, safety and healthcare, and energy saving/efficiency gain. Actually, they all fall in the first two categories, but some new inventions don't do anything that couldn't be done before; no, those inventions just do it more efficient. And although convenience is important and safety and healthcare are definately great for getting the Earth more crowded, as far as I see it, we need to redo quite a lot of what we've already done, by taking a good look at energy efficiency.
And maybe most of the upcoming 'inventions' might not be new at all...

Tuesday, February 20, 2007

Identity database [WIP]

Hi all (and iAll, my imaginary iAudience which makes this writing useful as long as there's no real audience)!

This is my first work-in-progress [WIP] post. It covers some idea which can and will be expanded in the future, for example through your comments. This time, a database in which every person is listed, including a lot of extra information.

Like most of you (I assume), I sometimes wonder how incredibly crappy the (in my case Dutch) government controls personal information. You need to fill in the same thing lots of times, for different departments, lots of times.
As such I propose one database for the listing of everyone in a nation (eventually maybe worldwide), in which a lot of personal data is stored. Apart from the obvious security issues, which someone needs to address very intelligently, there are two interesting questions I can come up with:

1) What information should be stored in such a database?

2) How large would this database become if you store N people?

Let's start with the first question. Some things I can think of one would like to have in such a database:

- Name
- Gender
- Date of birth
- Place of birth
- Nationality
- Postal Code
- House number
- Citizen Service Number (Used in Holland for fiscal and insurance transactions)
- Mother
- Father
- Partner
- Password of some sort?
- Children?
- Email address?

There might be a lot more information stored here.. First let's take on the question of who might use the database:
- Government
- Police
- Doctors
- Banks
- Companies
- Genealogy hobbyists?

That issues another topic: who should be able to access the database, and who should be able to edit the database? I suppose it's unreasonable to give everyone access, because it then becomes possible to find any person for any reason, including malicious ones (finding some ex you're trying to kill or some celebrity you're trying to stalk). However, if you don't allow individuals but do allow all companies, then it's as easy as starting your own business to find the information you want. So, my best guess is that it's best to allow everyone to search the database, but demand that they login using their own data (maybe with some interesting random question as identification like "what is the name of your mom's brother?" or "How many children did you have on ?"). That way it is possible to track who searches for whom, and thus identify any misuse afterwards. When the criminal record is linked, the database could log whether any criminals search the database. This poses another question: What information should and could be linked to, and which of those are available to others?

Some stuff that could be linked:
- Medical file
- Financial data
- Bank account
- Internal Revenue Services / Credit Registration Services
- Criminal record
- Biometric database (DNA, fingerprint, irises, facial patterns)
- Company information (Chamber of Commerce)
- Curriculum vitae -> automated using school, college and company databases

I suppose no-one but yourself and special persons/institutions should be able to view any of the above, maybe except the curriculum vitae part.

Maybe the most interesting concept is the 'Unknown Person' (UP): someone who is not listed in the database (but is living in the country). Some UPs might be houseless persons and members of intelligence agencies. However, if looking someone up gives you 'no entry', then it must be someone of an intelligence agency. Most likely this would be prevented by creating false entries, which have some strange connections. This would for example result in complete false families (as it is rather odd to not know neither mother or father in the Netherlands at least, which would make it any such orphans a rather simple target and automatically make any orphan a possible intelligence agent..). I guess it all depends on how much data is openly available. If the name/address/mother/father/birth date information is available of everyone in a single downloadable archive, it will be possible to check for family connections all the way through. I suspect that there is no non-connected family to be found, so fake families are no way to hide 'secret people'. However, false connections might be a way to work with. Just connect the secret person to random familia and there is no problem until it is checked.. Maybe still too obvious for an intelligence agency..
So what can be done? If every person is listed by a number in the database then every person will have his or her own number, preferably in chronological order. This will come down to the birth date, so there is arbitrariness in the numbers of people born on the same day. This should be officially reported within several days after birth and is also available from any involved doctor or medical institution, so most people should be taken care of. What to do then with those that are withheld, like teenage mothers keeping their child away from family and friends? These will have a problem when they try to find a school, as the identity of the child will be checked by the school.
So what will happen to this mother, will she be taken in custody for kidnapping some child of someone? Maybe it is necessary to file the DNA-profile of everybody in the same database, as it will be otherwise impossible to proof an identity. But let's get back to the kid, which is by now four years old and uncharted up till now. Will he be added at his birth date, or at this new date? I would opt for adding him at the end of the row, thus ending up in between persons (in this case) four years younger. This would make sure every number is used and in the mean time makes it harder to falsely add persons. This would also allow the easy addition of immigrants, though subtraction of emigrants seems not useful. (Eventually a world database should overcome this issue)
Although this all seems very useful (at least to me), it is convenient for governments to cover up their intelligence agents, so the best option in my opinion is a restraint on the visible information (no downloadable archives, not more than a certain number of searches per day, etcetera). This way it's possible to hide any false links by lack of information, though this can become visible in the long run.
Another option is oversimplifying any job description ('civil servant' instead of 'secret spy'), but the issue of the visible family always is apparent. The entry logs could be checked however, showing everyone that searches for the secret agent, thus creating the famous 'reveal and be revealed' paradox. Being able to search the logs will be the most important tool for a government, as is searching in all directions (searching on birth date, on mother, on company, etc.).

Enough ranting about it for now; if I think of an addition, I'll post it in this very article/post thing, so be sure to check up. Please, now go out and convince YOUR government to install an identity database, today. "Who do you want to look UP, today?"

(Next time I'll add a discussion of question 2), the amount of information stored.)