“A Proper Scientist”

I’ve always been a little careful when talking about science not to put myself forward too much as “a scientist” and more as “a student”. Partially, this is because I like to have an excuse when I do things wrong. However, I also felt that to be “a scientist” you have to have something to back up your claim. Since I’ve never had a paper published, that means I’ve always felt uncomfortable describing myself as “a scientist”.

Until now.

This week I got the news that a paper which I worked on was published in the journal IEEE Transactions on Electron Devices. While the paper wasn’t my idea, I was a big part of getting it produced and my expertise helped to refine it. I’m proud of that contribution, so I want to share that with everyone. But also, after seven years, I feel that I can finally call myself a scientist without having to add a caveat.

So if it floats your boat, have a read of the paper. It’s called;

“Measurements and Simulations of Low Dark Count Rate Single Photon Avalanche Diode Device in a Low Voltage 180-nm CMOS Image Sensor Technology”

The authors are;

Tomer Leitner, Amos Feiningstein, Renato Turchetta, Rebecca Coath, Steven Chick, Gil Visokolov, Vitali Savuskan, Michael Javitt, Lior Gal, Igor Brouk, Sharon Bar-Lev, and Yael Nemirovsky

And yes, one of them is me!

You can find the paper here.

Revival! And More SPADs

Well I honestly thought that final year at University would have me seeing more spare time, but that was naive of me and of course I had to neglect one of my hobbies. Anyway, I’ve managed to find some time to get back into posting here and I’m looking forward to putting in more posts.

As a start, here is an update on my general guide to SPADs! It’s a whole chapter giving a basic introduction to what image sensors are and how we talk about them. I’m happy with how it came out, and if you have any opinions on it let me know. I’m keen to make the necessary corrections if I’ve made a mistake.

In other news, I now have an official offer to continue my studies at the University of Surrey – I’ll be studying novel algorithms for data cleaning and interpretation with an application to spectroscopic data in quantum computing. It promises to have an interesting side-feature of learning about how electrons transition from fully quantum behaviour to classically chaotic behaviour. I’m completely excited about it! Especially because I’ll be doing Physics that’s fun and interesting again.

SPADs Update: Introduction!

Today I published the first little bit of the “Rough Guide to SPADs” project which I’ve been going on about. I’m publishing these sections more or less as I’m writing them, so they might be a little occasional. However, you can find the Introduction, Publications, and Contents sections in the dropdown on the “Single Photon Avalanche Diodes” tab. There will be a long list eventually – I’m in the middle of writing the first “chapter”, which is a bit of an introduction to diodes and image sensors in general.

I’m excited about this! I hope to have the first actual chapter of content done soon.

SPAD Life

Tuesday marked the final nail in the coffin of my MPhys research year at the Rutherford Appleton Laboratory. As I’ve previously mentioned, I worked at the CMOS Sensor Design Group there and… didn’t have a very easy time. Nevertheless, I’ve come out of it confident in my ability to produce good research. For me, the next step is further study in the area of Photonics, leading to a PhD in the area.

Back on topic: on Tuesday I presented a short talk on the subject of the research I undertook at RAL. Since it was only 15 minutes and aimed at second year undergraduate Physicists, I wasn’t able to cover everything I’d have liked to. I think, however, that I gave a reasonable account of Single Photon Avalanche Diodes and dead time effects – that was the meat of the work suitable for actual publication. I was pleased at how it was received, but disappointed that I didn’t get harder questions.

Finishing the whole ordeal and being free is nice, but I can’t help thinking of returning to the subject for a kind of encore. There aren’t very many easy ways of learning about SPADs, so my intention is to try to produce a “Rough Guide To SPADs” type thing. I am going to proceed by sketching out the subject areas I’d like to cover.

Diodes, Image Sensors, Single Photon Counters

An introduction of sorts; the important principles of what a SPAD is.

SPADs in “Contrast” to Other Imagers

AKA: “But why can’t we use a Photomultiplier Tube?”

What a SPAD “Looks Like”

How to tell a SPAD from a spud from an APD.

SPAD Fundamentals – Analogies and Pictures

The bit where I explain how SPADs actually work. More or less.

Quenching, Reset, and SPAD Outputs

Why SPADs don’t break, and why you never get a “perfect” SPAD.

Practical SPADs – Important Parameters

How SPADs are set up and what to measure to test how good they are.

The Efficiency Problem

AKA: “Why you really do need conventions in engineering”.

The Photon Transfer Curve

A short description of one potential (general) solution to “The Efficiency Problem” and a related experiment.

External Reset and Dead Time (1) – The Problem

Yo dawg. I herd you like problems, so I put a problem in your problem.

External Reset and Dead Time (2) – The Model

Solving while we solve.

External Reset and Dead Time (3) – The Experiment

Demonstrating that it works.

Summary

Hopefully where it’s all brought together.

My aim is to make all of the discussions in the above accessible and fun, to a certain extent. Ideally, I would like the whole series of articles to be accessible to a first year Physics undergraduate, but still satisfying to a Professor in Photonic Devices.

Getting Through a Tough Placement

This is a post dedicated to the placement student of the future. Most placements go off without any major problems, result in a good mark at the end and a good learning experience. The vast majority of people won’t be unhappy, but it is inevitable that someone will have a less than good experience. If, like me, you get demoralized with your placement for whatever reason… hopefully you’ll find this motivating.

My motivation for writing is that my past year – my placement year – has been packed full of stuff. Highs, yes, but a majority of lows. It’s difficult to describe in detail (a year is a lot of busy), but this post is hopefully going to provide good information for future MPhys students who end up in my position; not really happy, and unable to do much about it. It’s even somewhat extendable to BSc sandwich students. The aim of this post is that if anyone should need it, there’s a resource to help see them through a bit better.

I wrote a while ago about my placement and how it’s not really been what I’d hoped. I won’t go into detail, but let’s suffice it to say that it has all been challenging for me in ways it probably shouldn’t have. I’ve regularly been overwhelmed with stress, and I have to say that I wish I had someone or somewhere I could have gone to who’d have told me it’s not a unique situation.

You, the reader, might not be on placement (or even a student) yourself. In that case, I urge you to add to the following list by commenting, and try to send this to other people who you know who might find it useful. Hopefully this will eventually be seen by someone who’ll benefit from it. So here are the things I wish I’d been regularly reminded over the past year. Some of them are simple, even obvious, but I don’t think that makes them any less worth saying.

1. Try to maintain your enthusiasm. Things get boring, dull, slow, difficult, stressful, or all of the above. Don’t let it get to you. What you’re doing is worth doing, and there are better things around the metaphorical corner.
2. Stay in contact with your friends. Seriously, if it wasn’t for a couple of really good supportive people in my friends network, I’d have given up. I really would.
3. Try your hardest to learn from every bad thing that happens. One of the things which I’m taking out of this is a new perspective on how to deal with other personality types at work. You might not need to learn that, but you’ll learn something useful that you didn’t expect to. That’s a very useful thing, even if it’s not really the reason you’re there.
4. Luckily, it’s only ten months. You don’t have to be there for that long, so even if it’s bad you’ll get out of there. It might seem like a distant prospect, but worry not; the end is in sight.
5. It’s never all bad. There have certainly been highlights to my stay, and you take those away with you too. I’ll never forget the feeling of knowing something nobody else has realized yet – the essence of discovery is one of the most motivating things I’ve ever experienced.
6. Visit your department as often as you can. Being in a relaxed academic zone rather than a workplace is incredibly refreshing, especially when you get to chat about Carl Sagan with your peers. If you’re too far away to visit, then stay in regular contact with your University’s Visiting Tutor or use a convenient Facebook page (Physoc, I’m looking at you!).
7. Try to be proud of your work. The morale boost of just understanding that what you’re doing is important and useful and above all good quality is huge. Do not underestimate it.
8. You won’t regret it. Providing you work hard and fulfil your own expectations of yourself, you’ll come out of the placement without regrets. You might not have been happy there, but at least you’ll have done it.

Finally, and very importantly, please remember this; you are not alone. Although you’ve been unlucky, you’re not the first person to have problems. You have people’s sympathy and you have the benefit of a support network at your University, as well as your friends. You’re on the course for a challenge, and you will certainly get one. By pushing through, even just by coping day by day, you are showing that you’re the good person and good researcher/scientist that you want to be. Well done, and keep going!

Up-Goer Five – Single Photon Avalanche Diodes!

I am going to explain what I did at the place where they help some strange people to do hard things they can’t do on their own because it needs lots of money. Often this means making very small things go very fast or sending lots and lots of light at something all at once.

We like looking at things and taking pictures. Sometimes we want to look at things that happen really fast, like when light comes out of stuff for a very very short time, or when there isn’t enough light to see stuff easily. You can also time when light gets to you to look inside someone to find what’s wrong with them. That way you can also use to find out about what happens when you put tiny things together or hit them together.

The thing I worked on was very very very tiny – about as wide as a hair. The idea is you can put lots of them together and they make a thing like an eye but black and white and which is really good at seeing very little bits of light. Light comes in single bits which you can’t cut up, and most eyes can only see seven or more at once when they’re really really good. It’s really hard to see just one at a time! But the things I worked on can see just one at a time because they are so tiny.

When a tiny bit of light hits the thing, it knocks out another tiny thing which is the same kind of thing that makes lights work. The thing that gets knocked out goes faster and faster and faster also for the same reason that lights work. It goes so fast that it knocks more of the things out! You can tell when a bit of light hits the thing because you get lots and lots and lots of makes-light-work things coming out, which is easy to tell when it happens.

The problem is that the eye-things I worked on doesn’t see all of the bits of light because sometimes it misses them. What you need to know to make the things I worked on better is how often they miss bits of light. That can be hard for lots of reasons, but some of the reasons can be got around because some people found out how. One problem is that the eye-things can’t always see the light-bits because they are too slow at looking, which is a different reason than the problem was to begin with. This is not a normal problem with other kinds of eye-thing so no-one knows how to get round it easily.

I found a way of making that second problem go away. It was easy to come up with but hard to find out if it was right, because you can’t tell exactly how much light there is without using the eye-things. We made a computer think very hard about it for a long time and it found that I was right. Then we tried a different way of telling whether I was right using the real eye-things and we found that I was right again. That was nice, and now we want to tell lots of other people about it so that they can all make better eye-things.

The Programming Game

When I was learning my first programming language (Fortran 95) two or three years ago, the students in my class had a fun little game; to see who could solve the problem in the fewest number of lines of code! It was an extremely good way of learning how to make programs more efficient, but it was a somewhat bad way of learning to comment code and make it clear.

Fast forward to 2013 and the Facebook group I started to support people learning and coding in Fortran at my University turned up an interesting example of the efficiency game;

Just solved the last ‘extra’ exercise on Paul Stevenson‘s tutorial in 14 lines of code without any new functions or KINDs. Do I win a prize?

The “extra exercise” refers to a set of problems which aren’t necessary to complete the course, but can be done for the student’s entertainment. The particular problem is quite simple, and is meant to simply give the student experience in the KIND statement and intrinsic function in Fortran 95. However, this serves as a perfect example of how success in computing does not solely rely on your efficiency at coding! Another important lesson is the simplification of your problems into a form which a computer will find easy to solve.

The problem is simple: Calculate the probability P  that in a class of 30, two or more people share the same birthday. It’s a classic problem in probability, and the equation typically used as a solution is;

P = 1 – ( n! / ( (n-c)! n^c ) )

Where n is the number of days in a year and c is the number of classmates. The stumbling block for a very simple calculation is that the program cannot normally handle such huge numbers as 365! ( = 1x2x3x…x364x365). It’s a massive number which can’t be handled by the normal protocol which is used in a standard Fortran compiler. One way to solve the problem is to force the compiler to allocate more space to the number so that it can be calculated. However, that’s not the only way of solving the problem.

What this clever chap Adam decided to do was to simplify the term with the n! in it. If you can remove the need to make that calculation, the program becomes trivial. So that’s what he did! You can simplify the formula by realising that n! and (n-c)! simplify to the product of the numbers from (n-c+1) to n. With that, it’s fairly easy to produce a DO loop which makes that calculation;

DO i=(n-c+1),n

product=product*i

END DO

P = 1 – product * n^-c

However, this is still an inefficient solution because 365^30 is still a massive number. Here’s another simplification which he used; since you multiply c times in the loop, and you have n^-c on the outside, you can collapse one of the n‘s in that term into each iteration of the loop! The line product=product*i goes to product=product*i/n. However, you can simplify it further by removing the need to have that ugly set of calculations in the DO statement.

Once you realise that the loop makes a substitution for i=(n-c+i), you can rewrite the inside of the loop with that substitution and remove it from the loop statement. With the two changes combined;

DO i=1,c

product=product*(1-(c-i)/n) !simplified

END DO

P = 1 – product

Doesn’t that look a lot nicer? I think it does. But there’s one more thing you can do to make it even easier. When you remember that the action taken here is multiplication, you know it can be done in any order! Which means that you don’t have to start at (c-i), rather you can start at c-(c-i) and work up to c. Since c-(c-i) simplifies to i, you can write;

DO i=1,(c-1)

product=product*(1-i/n) !simplified further

END DO

P = 1 – product

So all it takes is some creative thinking to make your complicated KIND statements disappear! The program was reduced to 14 lines in it’s entirety, so well done Adam.

Finally, if you want to half the number of lines again, try using Matlab. This is the code I made to check it, in it’s entirety;

n=365;
c=30;
cumulative=1;
for i=1:(c-1)
cumulative=cumulative*(1-(i/n));
end
P1=1-cumulative

Without even trying to minimize code length! 7 lines.

So that’s one of the things that I think people forget when they see someone solve a computing problem really easily; often, it takes intuitive leaps to see where the bit is which makes it difficult for you. Once you remedy that, your programming is often a lot faster. (and you can win code length competitions with your friends!)

Disclaimer: after adopting Matlab on my home PC, I haven’t written a full program in Fortran at all. I’m a bit rusty, so please don’t beat me up about the syntax!