Simple City

Richard Elwes's Blog and Website

  • Book: Maths 1001
    • Maths 1001: Errata
    • Maths 1001: Reviews
  • Book: Chaotic Fishponds…
  • Book: How to Build a Brain
  • Book: Maths in 100 Key Breakthroughs
  • Book: The Maths Handbook
  • Writing
  • Talking
  • Teaching
  • Research
  • Who is Richard Elwes?

Learning to Learn from Babies

Posted by richardelwes on July 16, 2015
Posted in: Brain Science, Education. 1 Comment

cakes

Yesterday, my twin sons turned one. I have spent an amazing number of hours over the last year watching them. I wondered if this experience might teach me something too, about how to learn. After all, babies are the grandmasters on this subject. In the same time that it has taken me to incrementally advance my knowledge of some tiny corner of mathematics, my children have moved from a total inability to do anything besides scream, crap themselves, and scream again, to being able to feed themselves (messily, so messily, but still), crawl, clap, grab, wave, recognise people, stand, and so on. And these are just the most visible manifestations of a deep mental transformation during which their brains have learnt huge amounts about processing sensory data and coordinating muscle movement.

If that rate of learning was to continue through their lives, they would grow into geniuses far surpassing anything humanity has seen so far. So how do they do it? Of course a large part of the answer is physiological: babies’ brains are a lot more plastic than adults’, highly efficient sponges for the absorption of new skills. There’s not a great deal we can do about that (although there may be something). All the same, I think we might be able to see some other, useful principles in action too:

  1. Play. We don’t usually talk about babies “working” – but they are, just as assuredly as a student revising or a scientist researching. The difference is that babies are also undeniably playing – and we wouldn’t usually describe either of the other two in that way. Babies are not motivated by exam grades or pressure to publish. The more fun you find your work, the better you do it.
  2. Be curious. The babies immediately home in on any new item which appears in their playing area, and start investigating. They are always exploring the room’s boundaries, and grabbing at anything unfamiliar or interesting (my laptop, mugs of hot tea, etc..). But they are not searching for anything in particular. Set aside some time for open-ended exploration and experimentation.
  3. First, learn one thing well. The boys learned to clap quite early on. With this under their belt, other manual skills such as waving and pointing were comparatively easy to pick up. Likewise, an experienced mathematician will find it easier than a novice to master an unfamiliar mathematical topic. Even if neither has any directly relevant knowledge, the fact that one is practised in the art of learning mathematics should carry them a long way. Building skills can be worthwhile, even when the skills themselves are not.
  4. A change is as good as a rest. In the opposite direction, the twins do not spend hours at a stretch practising one thing, such as walking. Instead they do it for a little bit, then get distracted by a toy, move onto another toy, have a crawling race, try holding a conversation with their mother, then they do some more walking, and so it goes on. Have more than one project on the go.
  5. Don’t be scared. Most of the time, my sons appear completely fearless. They happily crawl into perilous situations, pull over heavy objects, and invite disaster in any number of imaginative ways. This is despite the fact that they regularly do fall down and otherwise upset themselves. Take risks. Even if they don’t immediately pay off, continue to take risks.
  6. Don’t be embarrassed. Babies are not only unworried about taking a tumble, they’re also unafraid of looking like fools. The more I ponder this, the more important I think it is. In my efforts to learn Japanese, for example, I am hindered (perhaps more than I have realised) by the fear of making embarrassing mistakes in conversation with my in-laws. Likewise, mathematicians do not enjoy admitting errors, or gaps in their knowledge, in front of their colleagues (let alone their students). I think this is a bad habit. In order to learn from your mistakes, you must first allow yourself to make some.
  7. Accept help. The boys are utterly dependent my wife, me, and the other generous people who help us look after them (thank you!). Obviously, adults shouldn’t be that reliant on others, except in extremis. Nevertheless, there may be people in your life who would like to help you succeed. Let them.
  8. “Good enough” isn’t good enough. My children have reached the point where crawling is a highly efficient form of travel – they can zoom around the house to wherever they want to be. Walking, meanwhile, is a faltering, risky business. It would be a perfectly rational short term decision if they opted not to bother with it. Of course, babies don’t reason that way, which is just as well. Invest in your long-term skills, even at short term cost.
  9. Don’t focus on the scale of the challenge. My children’s vocabulary currently consists of little more than “dadadada”, “mamamama”, “aaaaghh”, and “pmmpphh”. It will be quite a journey from these noises to mastery of the language of Shakespeare (and indeed that of Chikamatsu). Of course they have no idea about that. The journey matters more then the destination.

The top 10 mathematical achievements of the last 5ish years, maybe

Posted by richardelwes on June 18, 2015
Posted in: Maths. Tagged: abc conjecture, Joan Taylor, Shinichi Mochizuki, Thomas Hales. 12 Comments

I have recently been going through my book Maths 1001 making updates for a forthcoming foreign edition (of which more in future). So I have been looking over mathematical developments since approximately 2009. Thus, I present ten major developments in the subject since around then, arranged arbitrarily in ascending order of top-ness.

"Socolar-Taylor tile" by Parcly Taxel - Own work. Licensed under CC BY-SA 4.0 via Wikimedia Commons - https://commons.wikimedia.org/wiki/File:Socolar-Taylor_tile.svg#/media/File:Socolar-Taylor_tile.svg

“Socolar-Taylor tile” by Parcly Taxel – Own work. Licensed under CC BY-SA 4.0 via Wikimedia Commons – https://commons.wikimedia.org/wiki/File:Socolar-Taylor_tile.svg#/media/File:Socolar-Taylor_tile.svg

 

10. Mochizuki’s claimed proof of the abc conjecture.   The countdown kicks off on an awkward note. If Shinichi Mochizuki’s 2012 claimed proof of the abc conjecture had gained widespread acceptance, it would definitely top this list. As it is, it remains in limbo, to the enormous frustration of everyone involved.

 

9. The weak Goldbach conjecture. “From 7 onwards, every odd number is the sum of three primes.” We have known since 1937 that this holds for all large enough odd numbers, but in 2013 Harald Helfgott brought the threshold down to 1030, and separately with David Platt checked odd numbers up to that limit by computer.

 

8. Ngô Bảo Châu’s proof of the Fundamental Lemma. Bending the rules to scrape in (time-wise) is this 2009 proof of a terrifyingly technical but highly important plank of the Langlands Program.

 

7. Seventeen Sudoku Clues. In 2012, McGuire, Tugemann, and Civario proved that the smallest number of clues which uniquely determine a Sudoku puzzle is 17. (Although not every collection of 17 clues yields a unique solution, their theorem establishes that there can never be a valid Sudoku puzzle with only 16 clues.)

 

6. The Growth of Univalent Foundations/ Homotopy Type Theory. This new approach to the foundations of mathematics, led by Vladimir Voevodsky, is attracting huge attention. Apart from its inherent mathematical appeal, it promises to recast swathes of higher mathematics in a language more accessible to computerised proof-assistants.

 

5. Untriangulatable spaces. In sixth position is the stunning discovery, by Ciprian Manolescu, of untriangilatable manifolds in all dimensions from 5 upwards.

 

4. The Socolar–Taylor tile. Penrose tiles, famously, are sets of tiles which can tile the plane, but only aperiodically. It was an open question, for many years, whether it is possible to achieve the same effect with just one tile. Then Joan Taylor and Joshua Socolar found one (pictured above).

 

3. Completion of the Flyspeck project. In 1998, Thomas Hales announced a proof of the classic Kepler conjecture on the most efficient way to stack cannon-balls. Unfortunately, his proof was so long and computationally involved that the referees assigned to verify it couldn’t complete the task. So Hales and his team set about it themselves, using the Isabelle and HOL Light computational proof assistants. The result is not only a milestone in discrete geometry, but also in automated reasoning.

 

2. Partition numbers. In how many ways can a positive integer be written as a sum of smaller integers? In 2011, Ken Ono and Jan Bruinier provided the long-sought answer.

 

1. Bounded gaps between primes. It’s no real surprise to find that the top spot is taken by Yitang Zhang’s wonderful 2013 result that there is some number n, below 70 million, such that there are infinitely many pairs of consecutive primes exactly n apart. The subsequent flurry of activity saw James Maynard, and a Polymath Project organised by Terence Tao, bring the bound down to 246.

 

But, but,…?!

Where’s Hairer’s work on the KPZ equation? What about Friedman’s new examples of concrete incompleteness?! What can I say? It’s just for fun, folks. If you think I’ve got it horribly wrong, then feel free to compile your own lists. (The real answer for such things being left out is that I couldn’t easily update my book to include them.) And now…

 

Bonus feature! Progress in computational verifications and searches

In no particular order:

  • The simple continued fraction of π has now been computed to the first 15 billion terms by Eric Weisstein, up from 100 million.

 

  • The decimal expansion of π has been computed to 13.3 trillion digits, up from 2.69999999 trillion.

 

  • The search for the perfect cuboid: if one exists, one of its sides must be at least 3 trillion units long, up from 9 billion.

 

  •  Goldbach’s conjecture has been verified for even numbers up to 4 × 1018 by Oliveira e Silva, up from 1018.

 

  • The largest known twin primes are the pair either side of 3756801695685 × 2666669, up from 2003663613 × 2195000.

 

  • The largest known prime, and the 48th known Mersenne prime (up from 47), is 257885161-1, up from 243112609-1.

 

  • The Encyclopedia of Triangle Centres contains 7719 entries (up from 3587).

 

  • The longest known Optimal Golomb Ruler is now 27 notches long (up from 26):
    (0, 3, 15, 41, 66, 95, 97, 106, 142, 152, 220, 221, 225, 242, 295, 330, 338, 354, 382, 388, 402, 415, 486, 504, 523, 546, 553)

 

  • The most impressive feat of integer-factorisation using classical computers, is that of the 232-digit number RSA-768:

    1230186684530117755130494958384962720772853569595334792197322452 1517264005072636575187452021997864693899564749427740638459251925 5732630345373154826850791702612214291346167042921431160222124047 9274737794080665351419597459856902143413

    into two 116 digit primes

    3347807169895689878604416984821269081770479498371376856891243138 8982883793878002287614711652531743087737814467999489
    and
    3674604366679959042824463379962795263227915816434308764267603228 3815739666511279233373417143396810270092798736308917.

    (The previous record was the 200 digit semiprime RSA-200.)

 

  • The most impressive feat of integer-factorisation using a quantum computer is that of 56,153=233 × 241. The previous record was 15.

 

  • The Collatz Conjecture has been verified for numbers beyond 2 × 1021. The previous record was 5.76 × 1018. (However, this has happened via a patchwork of distributed computing projects, and I have not been able to establish with any certainty that every number up to the new higher limit has been checked. I encourage someone in this community to organise all the results in a single location.)

Book Review: “What the Best College Teachers Do” by Ken Bain, 2004

Posted by richardelwes on March 31, 2015
Posted in: Bookery, Elwes Elsewhere. Leave a comment

I have a new post at The De Morgan Forum.

Euler’s Partition Theorem

Posted by richardelwes on March 7, 2015
Posted in: Number theory. Leave a comment

euler I’ve recently been revisiting Euler’s Theorem. Not that one. No, not the one on this commemorative stamp either. No, no, not the totient function one. (Good guess though.)

I mean the one about partitions.

What is a partition? It’s simply a way of splitting up a number (a positive integer) into smaller pieces. For instance 2+2+1 is a partition of 5. The question is: how many ways of doing this are there?

Well there’s just one partition of one, namely 1.

There are 2 partitions of two: 2 and 1+1.

There are the 3 partitions of three: 3, 2+1, 1+1+1. (Notice that for these purposes, we count 2+1 and 1+2 as the same partition.)

So far the pattern looks rather easy. But there are the 5 partitions of four:
4, 3+1, 2+2, 2+1+1, 1+1+1+1

And then 7 partitions of five:
5, 4+1, 3+2, 3+1+1, 2+2+1, 2+1+1+, 1+1+1+1+1

Continuing the sequence, we find 1, 2, 3, 5, 7, 11, 15, 22, 30, 42, 56, 77, 101, 135, 176, 231, 297, 385, 490, 627, 792, 1002, 1255, 1575, 1958, 2436, 3010,…

What’s the pattern now? Finding an explicit formula for the number of partitions of n is seriously hard. Even Euler couldn’t manage it. But he was able to make some progress: he found a generating function.

Next question: what is a generating function? To answer with an example, the generating function for the sequence 1,2,3,4,5,6,\ldots is the power series x+2x^2 +3x^3 +4x^4 +\ldots .

More generally, the generating function for the sequence a_0, a_1, a_2, a_3, \ldots is a series: a_0 + a_1 x + a_2 x^2 + a_3 x^3 + \ldots .

Next question: what’s the point? Often the resulting series can be radically compressed. For instance, a bit of mathematical magic tells us that the series x+2x^2 +3x^3 +4x^4 +\ldots can be rewritten[1] as \frac{1}{(1-x)^2} . Now the whole infinite sequence has been encapsulated by a short, simple algebraic expression. So, if you want to know the nth element of the sequence, you can extract it (if you know how) fairly easily from the generating function.

Well, there is not a lot of mystery in the sequence 1,2,3,4,\ldots – certainly there are no prizes for guessing its nth term. But what of the partition sequence: 1, 2, 3, 5, 7, 11,\ldots? Euler’s great insight was that there is a generating function for this too:
\left( \frac{1}{1 - x} \right) \times \left( \frac{1}{1 - x^2} \right) \times \left(\frac{1}{1 - x^3} \right) \times \left(\frac{1}{1 - x^4} \right) \times \ldots

You can see that this is not as nice as the function above – and it’s certainly not nice enough to close the book on the whole problem of computing partition numbers. All the same, this function is hugely easier to work with than scrabbling about with partitions with bare hands. Euler’s generating function can also be written as
\left(1+x+x^2+x^3+\ldots \right) \times \left( 1+x^2+x^4+ \ldots \right) \times \left(1+x^3+x^6+ \ldots \right) \times \ldots

On first sight, this looks like an algebraic nightmare: you have to open up an infinite number of brackets, each of which contains an infinite series! However, if you want to compute the nth partition number, you only care about the first n brackets, and only the first few terms in each: from the first bracket you care about the first n terms, from the second bracket approximately the first \frac{n}{2}, from the third approximately the first \frac{n}{3}, and so on. In fact the total number of terms you need to worry about is approximately n \ln n, which is a manageable number when n is not too big.

So although Euler didn’t find a formula for partition numbers, his generating function does provide a manageable procedure for computing these numbers. But what of an actual formula?

Further progress came courtesy of a pair of mathematical superstars to rival Euler: Hardy and Ramanujan. The nth one is given approximately by the formula
\frac{1}{4 n \sqrt{3}} e^{\pi \sqrt{\frac{2n}{3}}}.

As n grows bigger and bigger, this number gets proportionally closer to the right answer. But what about an exact formula? That’s something there has been huge progress on in recent years… but perhaps that’s a story for another day.

 

[1] There are issues of convergence here, which I will ignore for now.

Math Frolicking

Posted by richardelwes on March 1, 2015
Posted in: Elwes Elsewhere. Leave a comment

As an ambitious young researcher, you can miss the wood for the trees. When you’re presented with a theorem, your inclination is to dive straight into the proof, and start grappling with the toughest ideas in there. But when you’re addressing a general audience, you have to step back, pause, and ask “What is the point of this theorem? Where did it come from? Why should we care?” You’re forced to adopt a different perspective on the subject, and that’s refreshing.

Read more of my interview over at Math Frolic.

Dr Who and the Quaternions

Posted by richardelwes on January 27, 2015
Posted in: Education, Elwes Elsewhere, Maths, Meedja, Technology. Leave a comment
“The parametric engines are jammed! Orthogonal vector’s gone! I’m almost out of ideas!”

I have a guest post at The Aperiodical reporting on the London Mathematical Society’s birthday party last week, where Doctor Who was a recurring theme.

The Grothendieck Song

Posted by richardelwes on January 2, 2015
Posted in: Elwes Elsewhere, Geometry, Music, Nonsense. 2 Comments

There are many songs in the world about love and loss, heartbreak and heart-ache. There are altogether fewer about algebraic geometry in the style of Alexander Grothendieck. Here is my attempt to fill that gap:

 

Disclaimer: I hope none of this needs saying, but just in case of misinterpretation:

1. No views expressed therein are attributable to any organisation to which I am affiliated.

2. Most of the views expressed therein are not attributable to me either, but are a deliberate exaggeration, for comic effect, of a common initial reaction to one’s first meeting with Grothendieck-style algebraic geometry.

3. It is not intended as a serious critique of any mathematician or school of mathematics!

 

Lyrics

When I was a young boy doing maths in class
I thought I knew it all.
Every test that I took, I was sure to pass.
I felt pride, and there never came a fall.

Up at university, I found what life is for:
A world of mathematics, and all mine to explore.
Learning geometry and logic, I was having a ball.
Until I hit a wall…

For I adore Euler and Erdős,
Élie Cartan and Ramanujan
Newton and Noether. But not to sound churlish
There’s one man I cannot understand.

No, I can’t get to grips with Grothendieck,
My palms feel sweaty and my knees go weak.
I’m terrified that never will I master the technique
Of Les Éments de Géométrie Algébrique.

He’s a thoughtful and a thorough theory-builder, sans pareil.
But can anybody help me find the secret, s’il vous plaît
Of this awe-inspiring generality and abstraction?
I have to say it’s driving me to total distraction.

For instance… A Euclidean point is a location in space, and that we can all comprehend.
René Descartes added coordinates for the power and the rigour they lend.
Later came Zariski topology, where a point’s a type of algebraic set
Of dimension nought. Well, that’s not what I thought. But it’s ok. There’s hope for me yet!

But now and contra all prior belief
We hear a point’s a prime ideal
In a locally ringed space, overlaid with a sheaf.
Professor G, is truly this for real?

No, I can’t make head nor tail of Grothendieck
Or Deligne, or Serre, or any of that clique.
I’ll have to learn not to care whenever people speak
Of Les Fondements de la Géométrie Algébrique.

But don’t take me for a geometrical fool.
I can do much more than merely prove the cosine rule.
I’ll calculate exotic spheres in dimension 29
And a variety of varieties, projective and affine.

I’m comfortable with categories (though not if they’re derived)
I’ll tile hyperbolic space in dimension 25
I can compute curvature with the Gauss-Bonnet law
And just love the Leech Lattice in dimension 24.

But algebro-geometric scheming
Leaves me spluttering and screaming.
And in logic too, you may call me absurd
But I wouldn’t know a topos, if trampled by a herd.

I’ve tried Pursuing Stacks but they vanished out of sight,
I’ve fought with étale cohomology with all my might.
And Les Dérivateurs. It’s 2000 pages long.
I reach halfway through line 3, before it all goes badly wrong.

No, I’ll never get to grips with Grothendieck
And I’m frightened that I’m failing as a mathematics geek.
All the same, I can’t deny the lure and the mystique
Of Le Séminaire de Géométrie Algébrique.

– Richard Elwes, 2015

 

Comments are closed on the Youtube page (because obviously) and here (because broken) but open on Google plus, Facebook, and Twitter.

and here

Posts navigation

← Older Entries
Newer Entries →
  • Some of my interesting things

    • The Grothendieck Song
    • A rhopalic sentence
    • Wondering about this wallpaper?
    • Topological limericks
  • Affiliations

    • I work at University of Leeds, UK
    • I am the Publicicty Officer for the European Mathematical Society
  • Richard is on a social medium

    • Richard Elwes on Google+
  • And on another…

    My Tweets
  • And one more…

    And one more…
  • Subscribe to Simple City

    • RSS - Posts
    • RSS - Comments
Blog at WordPress.com.
Simple City
Blog at WordPress.com.
Cancel