Profile cover photo
Profile photo
Christian Lawson-Perfect
Mathematician, koala fan, Aperiodical triumvir
Mathematician, koala fan, Aperiodical triumvir

Christian's posts

Post has shared content
Back in 2013, our own Christian Lawson-Perfect came up with a way of making a solid from the smallest non-Hamiltonian graph, the Herschel Graph. Called the Herschel Enneahedron, it’s got nine faces (three squares and six kites) and the same symmetries as…

Post has attachment
I'm considering running a 'hack day' style event, but I'm very unsure about some things, and I'd like to hear from anyone with experience running/attending such an event.

* we recently ran a conference on e-assessment in mathematical sciences (, which showed that there are lots of people working in the area, and there are sufficiently many who don't mind travelling all the way to Newcastle to hear about it.
* I'm the lead developer of Numbas (, an open source e-assessment system aimed at maths. At the moment, I'm the only real contributor to the code, though the system is used by hundreds of people around the world.

We have some money left over in our budget, which I thought we could use to organise an event like a hack day. At the moment, I'm the only contributor of code to the Numbas project, so my main aim is to find a way of getting more people involved.

Now, Numbas doesn't have a huge number of users - on the order of a few hundred - and I think that the users who could helpfully contribute code are a very small minority. Getting that handful of people to take part would be good but difficult, so I wonder if I could attract good coders who don't already use Numbas to the project.
My guess is that a good route in to contributing code to Numbas would be to add a self-contained feature, maybe in the form of an extension: it wouldn't need much input from me in terms of refactoring existing code, and the stakes would be low - if they fail in some way, nothing else is broken.

So I have some questions, below. I don't expect answers to any of them - whatever input you can offer, even just gut feelings, would be helpful.

- should I already have some people contributing to the project before running a hack day, or is a hack day a good way of jump-starting collaboration?
- what do we need to offer to people who attend? Will people make their own way to Newcastle or do we really need to cover travel expenses?
- would offering a list of ideas for small, self-contained projects be more attractive than a more broad pitch "in this hack day we'll work on <accessibility/marking/graphics/etc.>"?
- what can I do to attract people who aren't currently involved with Numbas? Where can I advertise the event to reach the right people? I think CS departments would be a good place to start, but how do I pitch it? These people would have to get something out of the work, so I guess it would have to be related to whatever else they're working on.
- is it worth my time trying to recruit postgrad students, or should I focus on faculty? My assumption is that contributing to an established open source project would look good on a CV and potentially provide a good project - there are quite a few features in Numbas that I reckon would make a good thesis topic.
- I feel that a single day would be too short, especially considering most people's travel times to Newcastle. What do you think is the optimal length for an event? Should I expect everyone to attend the whole time, or allow people to just attend for a portion?
- when would be the best time to run the event? Is term-time a no-no?
- outside a hack day, what can I do to be more clear about how people can contribute to Numbas? If someone asked me what they could work on, I'd look at the GitHub issues list, but that doesn't do a good job of selling the idea of contributing.

Alternately, it would also be nice just to get people together to create some high-quality material using Numbas. That would mean I potentially attract more people who wouldn't be comfortable attending a code-focused workshop, but this way I think I'm less likely to get any serious contributions to the code. Does that sound right to you?

And now I'm going to tag a few people that I think might have something helpful to say: +William Stein, +Michael Croucher, +Vincent Knight, +James Denholm-Price, +Peter Rowlett

If you have something to say that you don't want to put in public, please email me at

Jeremy Corbyn just mentioned this stat in PMQs: "In 1998, more than half of working households of people aged 16 to 34 were buying their own homes. Today, the figure is 25%."

What's the correct thing to measure here? It seems sensible to me that 16-year-olds are much less likely to buy a home than 34-year-olds, no matter the economic conditions, and "16-34" is a very wide range.
A baby boom would change this stat as the cohort come of age and get older: first it would go down, as the proportion of 16-year-olds goes up, and then it would increase as they get older and more of them buy houses.

Is there a way of normalising the data in one-year intervals, to control for a difference in the distribution of ages?

Post has shared content
Two months ago, I bought and not only set up the internet’s fanciest primality-checking service, but also invented a rather addictive game. It quite quickly went viral, or as relatively viral as a maths game can get, with people tweeting…

Post has shared content
Around about exactly this time a year ago, I bought the frivolous domain name, to celebrate π Day and to indulge my curiosity about a marvellous algorithm to compute π’s digits. This year, I’ve been thinking about prime…

Post has attachment

Post has shared content
I had the idea of doing short videos about mathematical objects I’ve got lying around. First up is a very unconventional group theory textbook.

Post has attachment
Starting with some fun maths about the area of an inscribed circle, I made this:

The fun fact is this: a circle inscribed in a square of side length 1 covers pi/4 of the square (pi/4 = pi*(1/2)^2). A quadrant of a circle inscribed in a square of side length 1 also covers pi/4 of the area (pi/4 = (pi*1^2)/4).
So, where you've got a quadrant of a circle inscribed in a square, you can replace it with a circle and still cover the same area. You can then split that circle into quadrants, and do the same again, until you get bored.

The thing I made just does that a few times, and layers the resulting diagrams on top of each other, in different colours. I think it makes some pretty nice compositions. I recently decided to try my hand at mathematical art, and I think this is a good first step.

Post has attachment
Someone on facebook posted this infographic - - which claims that each airstrike Britain might conduct in Syria would cost £508,000. After looking at the equivalent numbers of helpful people that could have paid for, some wag then commented asking how much carpet it would buy.

So, if we were literally carpet bombing Syria, how much could we cover?

Carpetright has a wide range of carpets on sale, starting at around £5 per square metre, so £508,000 would buy 101,600 square metres of carpet.
By drawing a circle of that size on Ar Raqqah, it looks like one air strike could comfortably cover this stadium in Raqqa:

The first source I found giving a numerical blast radius for the kind of bomb used in these airstrikes is in a forum for the game "World of Warplanes" - A commenter there says that the blast radius of a mk-84 bomb is about 365 metres, similar to the radius of the circle of carpet.

In conclusion: for the same cost, you can carpet just about the same area as you can blow up. (minus delivery and fitting)

Post has attachment
Can anyone make sense of the little snippets of maths presented in this?

How is zeta * conj(zeta) bilinear? And the rest make just as little sense.
Wait while more posts are being loaded