matthewwasley.com

I was recently forwarded part of an email exchange. I think it’s worth sharing.

The background is whether poster boards in corridors constitute a fire hazard. Clarification was issued on this matter and the guidance was that all boards must be enclosed (i.e. in perspex or similar), unless the display material was laminated, in which case it would smoke but not combust. An email was sent around and the following reply received:

Purely out of a sense of curiosity, could I ask for clarification of the clarification?

We have it on good authority (Ray Bradbury) that the autoignition temperature of paper is 451’F (roughly 231’C).  As the temperature field in a corridor would be seriously anisotropic (very hot at the top, cold on the floor and cool on the walls due to conduction and wall mass) we could expect that the free air temperature would be much higher than this and that life would not be possible there (in any case the oxygen would be very low and smoke would make the area toxic).

So, surely, the odd bit of paper on the wall would be neither here nor there.  Paint, carpets, lino and , of course, wall paper will be much more of a problem.  

Counterintuitively perhaps, a piece of paper directly on a wall would be cooler than one on a noticeboard due to the thermal capacity of the wall and insulation provided by a notice board.

Perspex and acrylics in general melt around 120’C and poly carbonates around 150’C.  So by the time  we get to the autoignition point of paper the noticeboard covers will be dripping down the walls. It’s difficult to see what protection we get.

We may be confusing the issue with that of ‘fire load’ which is the amount of combustible material in a given volume.

A further point in passing:  I would have thought that the heat of combustion of the few to several mm of Perspex enclosing the boards would be orders more than a similar area of say 100 micron paper.  Also there would be proportionally more carbon and hence smoke.

The emailer goes on to ask for a detailed risk assessment comparing and quantifying the alternatives. The response is unknown, but a line from Basil Fawlty springs to mind: “… why do we bother? We should let you all burn“.

It works (kind of). My adventure in opinion mining has moved on a bit since the last post on that topic. My bodged together Python script now reads in half a dozen RSS feeds, looks for ones with a key word in the title (currently, and appropriately given my last sciency post, gold), downloads them and then passes the text to OpinionFinder.

In the next step, OpinionFinder has a look through the downloaded files and looks for positive and negative words. My script takes OpinionFinder’s conclusions on the individual files and sums them up into an overall positive or negative result.

The next part in my master plan is to compare the results on a given day to the market and see whether positive news correlates to upward movement in price.

To be continued…

Sometimes one thing leads to another and I find myself embarking on a new project. A conversation with a friend led to me thinking about whether opinion mining/sentiment analysis could be used as part of a stock market investment strategy. Opinion mining, roughly speaking, is the analysis of text to determine the attitude of the writer. There’s money to be made in this area and there are many companies doing just that. It’s also rich ground for academic research, which is good as it means there’s lots of information to be had.

Perhaps not unsurprisingly therefore, my idea of using sentiment analysis on the stock market is not a new one. Several groups have published papers on it (like this one) and an investment company has even launched a financial product based on a similar concept.

I still wanted to have a go myself though, so I set about looking for useful software. The academics in the paper I linked to above used a tool called OpinionFinder as part of their research. OpinionFinder is free to download here. Be warned though, if you’re a Windows user and plan on clicking a link and having the thing working straight away you’re going to be disappointed. OpinionFinder relies on other software which uses older versions of compilers and other things. On a newish version of Ubuntu (don’t even think about using Windows) this means a lot of jiggery pokery is needed to get the thing running, and that’s the point I’m at now.

My idea involves inputting various news feeds into OpinionFinder, so I also need to get/write software which will download RSS and Atom feeds and sort them. UniversalFeedParser should be able to help with this, and I’m pleased to report I downloaded and ran it without any problem. “All” I now need to do is add a few other modules and write a bit of code to join them up.

All very well, I hear you say, but I thought you claimed to be a writer so what has all of this geekery got to do with wordsmithing? Not a lot in fairness, save that I can write about it here, but it gives me something to do in the evenings instead of all the things that I’m meant to be doing like writing, jobs around the house, watching TV, speaking to my family etc.

Due to yet another clerical error, I didn’t get shortlisted for the Wellcome prize, so instead I’ll post my entry here for anyone that wishes to read it…

Quantum computing: a new application of an old idea

Quantum. Definitely a high tech word. A word fit for the 21^st century and our brave new world of hovering chrome robots and travel to the stars. Except that the quantum world isn’t all that new, in fact the first inroads into quantum technology were made over a hundred years ago. Around 1900, Max Plank proposed that the energy given off by hot things is quantised, i.e. broken up into discrete chunks. Albert Einstein and others built on Plank’s chunks and began to use this new quantum theory to gain a much better understanding of how things work at the atomic level and beyond.

This new science was key to the development of the transistor, the laser and many of the other inventions of the 20^th century – technology that we still rely on today for our iPADs, Blu-ray players, washing machines and just about anything else that runs on electricity. Quantum research has continued into the 21^st century too, and we now have an unprecedented level of control over the sub-atomic world. This has opened up fantastic opportunities but also created a few thorny problems, like quantum computers.

If you own a computer with internet access, the chances are you’ve used it to buy something online or check your bank account. Your internet browser, and your bank for that matter, will keep your information secure by using RSA encryption. The RSA scheme, named after Rivest, Shamir and Adleman who first published it, involves multiplying two random prime numbers (numbers divisible only by themselves and one) in order to encrypt a message. Anyone wanting to intercept and decode the message needs to know what the original numbers were. Finding which two numbers were multiplied together to make a third number is called factoring, and computers are very good at it. However, the bigger you make the number, the longer it takes and it’s possible to pick a number so big it can’t be cracked in any sensible timeframe, even with a modern supercomputer. The problem is, quantum computers can cheat.

Quantum computers take advantage of the peculiar nature of the sub atomic world. In this mysterious place, we can never know everything about a particle. We might be able to nail its position down, but we’re then stumped on how fast it’s going. Or we can know how fast it’s going but not know where it is. Things get odder still though: the particle is, in fact, in all places and at all speeds at the same time. This is known as superposition. Erwin Schrödinger famously showed just how strange this was by a thought experiment involving a cat that was both alive and dead at the same time. Aside from annoying the RSPCA, superposition comes in very handy for quantum computing. Whereas a conventional computer has to process one number at a time, superposition allows a quantum computer to do a whole lot of them at once. If you apply this ability to factoring a large number you can dramatically speed up the process, and RSA ­– the system behind the world’s financial transactions and hence the value of money – is suddenly breakable.

Don’t start stockpiling baked beans and heading for the Welsh mountains just yet though, because quantum theory also provides a solution: quantum cryptography. That same inability to know what a particle is doing can be put to use in making an unbreakable communications system. Things can be arranged such that any eavesdropping will unavoidably be detected, making the system completely secure – barring human error of course. There’s more good news, whereas the first commercial quantum cryptography systems are starting to appear, quantum computing appears much further behind. For example, to break RSA you’d need the ability to work with at least 10,000 bits of quantum information and, as of April 2011, the most credibly demonstrated number was 14.

Anyway, perhaps we shouldn’t focus on the super-villain application of quantum computing, its power can be used for good as well as evil. Quantum computers should be very good at modelling large complex systems, like the brain for example, or interactions between atoms that could lead to new drugs. In fact their awesome power will probably lead to applications we haven’t even thought of yet. The research into quantum technology going on right now could well change the 21^st century in the same way that the original research changed the 20^th. Not bad for an idea that first emerged when Queen Victoria was on the throne.