About Evert van Nieuwenburg

Theoretical physicist @ Caltech Working on condensed matter and machine learning.

Introducing a new game: Quantum TiqTaqToe

A passing conversation with my supervisor

Video games have been a part of my life for about as long as I can remember. From Paperboy and The Last Ninja on the Commodore 64 when I was barely old enough to operate a keyboard, to Mario Kart 8 and Zelda on the Nintendo Switch, as a postdoc at Caltech, working on quantum computing and condensed matter physics. Up until recently, I have kept my two lives separate: my love of video games and my career in quantum physics.

The realization that I could combine quantum physics with games came during an entertaining discussion with my current supervisor, Gil Refael. Gil and I were brainstorming approaches to develop a quantum version of Tetris. Instead of stopping and laughing it off, or even keeping the idea on the horizon, Gil suggested that we talk to Spyridon (Spiros) Michalakis for some guidance. 

This is not the story of Quantum Tetris (yet), but rather the story of how we made a quantum version of a much older, and possibly more universally known game. This is a new game that Spiros and myself have been testing at elementary schools.

And so I am super excited to be able to finally present to you: Quantum TiqTaqToe! As of right now, the app is available both for Android devices and iPhone/iPad:

Get it on Google Play

 

Previous quantum games

Gil and I knew that Spiros had been involved in prior quantum games (most notably qCraft and Quantum Chess), so he seemed like the perfect contact point. He was conveniently located on the same campus, and even in the same department. But more importantly, he was curious about the idea and eager to talk. 

After introducing the idea of Quantum Tetris, Spiros came up with an alternative approach. Seeing as this was going to be my first attempt at creating a video game, not to mention building a game from the ground up with quantum physics, he proposed to put me in touch with Chris Cantwell and help him improve the AI for Quantum Chess.

I thought long and hard about this proposition. Like five seconds. It was an amazing opportunity. I would get to look under the hood of a working and incredibly sophisticated video game, unlike any game ever made: the only game in the world I knew of that was truly based on quantum physics. And I would be solving a critical problem that I would have to deal with eventually, by adapting a conventional, classical rules-based game AI for quantum. 

Fun and Games

My first focus was to jump on Quantum Chess full-force, with the aim of helping Chris implement a new AI player for the game. After evaluating some possible chess-playing AI engines, including state-of-the-art players based off of Google’s AlphaZero, we landed on Stockfish as our best candidate for integration. The AI is currently hot-swappable though, so users can try to develop their own! 

While some of the work for implementing the AI could be done directly using Chris’s C++ implementation of Quantum Chess, other aspects of the work required me to learn the program he had used to develop the user interface. That program is called Unity. Unity is a free game development program that I would highly recommend trying out and playing around with. 

This experience was essential to the birth of Quantum TiqTaqToe. In my quest to understand Unity and Quantum Games, I set out to implement a “simple” game to get a handle on how all the different game components worked together. Having a game based on quantum mechanics is one thing; making sure it is fun to play requires an entirely different skill set.

Perspective

Classic Tic-Tac-Toe is a game in which two players, called X and O, take turns in placing their symbols on a 3×3 grid. The first player to get 3 of their symbols in a line (diagonally, vertically or horizontally) wins. The game goes as far back as ancient Egypt, and evidence of the game has been found on roof tiles dating to 1300 BC [1].

Many variations of the game have existed across many cultures. The first print reference to a game called “tick-tack-toe” was in 1884. In the US the game was renamed “tic-tac-toe” sometime in the 20th century. Here’s a random fun fact: in Dutch, the game is most often referred to as “Butter-Cheese-and-Eggs” [2]. In 1952, computer scientist Alexander S. Douglas at the University of Cambridge turned it into one of the first computer games, featuring an AI player that could play perfect games against a human opponent.

Combinatorics has determined that whoever plays first will win 91 out of 138 possible board combinations. The second player will win in 44 boards. However, if both players play optimally, looking ahead through all the possible future outcomes, neither player should ever win and the game always ends in a draw, in one of only 3 board combinations.

In Quantum TiqTaqToe, with the current ruleset, we don’t yet know if a winning strategy exists.

I explicitly refer to the current ruleset because we currently limit the amount of quantumness in the game. We want to make sure the game is fun to play and ‘graspable’ for now. In addition, it turns out there already is a game called Quantum TicTacToe, developed by Allan Goff [3]. That version of TicTacToe has similar concepts but has a different set of rules. 

The Game

A typical game of Quantum TiqTaqToe will look very much like regular Tic-Tac-Toe until one of the players decides to make a quantum move:

A quantum superposition is made by dragging from one empty square to another.

At this point, the game board enters into a superposition. The X is in each position with 50/50 chance; in one universe the X is on the left and in the other it is on the right. Neither player knows how things will play out. And the game only gets more interesting from here. The opponent can choose to place their O in a superposition between an empty square and a square occupied by a quantum X.

A quantum entanglement move.

Et voilà, player O has entangled his fate with his opponent’s. Once the two squares become entangled, the only outcomes are X-O or O-X, each with probability ½. Interestingly, since the game is fully quantum, the phase between the two entangled outcomes can in principle be leveraged to create interesting plays through destructive and constructive interference. The app features a simple tutorial (to be updated) that teaches you these moves and a few others. There are boards that classically result in a draw but are quantumly “winnable”.

A quick note on the quantumness

The squares in TiqTaqToe are all fully quantum. I represent them as qutrits (like qubits, but instead of having states 0 and 1 my qutrits have states 0, 1 and 2), and moves made by the players are unitary operations acting on them. So the game consists of these essential elements:

  1. The squares of the 3×3 grid are turned into qutrits (Empty, X, O). Each move is a unitary gate operation on those qutrits. I’ll leave the details of the math out, but for the case of qubits check out Chris’ detailed writeup on Quantum Chess [4].
  2. Quantum TiqTaqToe allows you to select two squares in the grid, providing you with the option of creating a superposition or an entangled state. For the sake of simplicity (i.e. keeping the game fun to play and ‘graspable’ for now), no more than 3 squares can be involved in a given entangled state.

I chose to explicitly track sets of qutrits that share a Hilbert space. The entire quantum state of the game combines these sets with classical strings of the form “XEEOXEOXE”, indicating that the first square is an X, the second is Empty, etc.

Victory in the multiverse

So, when does the game end if these quantum states are in play? In Quantum TiqTaqToe, the board collapses to a single classical state as soon as it is full (i.e. every square is non-empty). The resulting state is randomly chosen from all the possible outcomes, with a probability that is equal to the (square of the) wave-function amplitude (basic quantum mechanics). If there is a winner after the collapse, the game ends. Otherwise, the game continues until either there is a winner or until there are no more moves to be made (ending in a draw). On top of this, players get the option to forfeit their move for the opportunity to cause a partial collapse of the state, by using the collapse-mode. Future versions may include other ways of collapse, including one that does not involve rolling dice! [5]

Can you beat the quantum AI?

Due to quantum physics and the collapse of the state, the game is inherently statistical. So instead of asking: “Can I beat my opponent in a game of Quantum TiqTaqToe?” one should ask “If I play 100 games against my opponent, can I consistently win more than 50 of them?”

You can test your skill against the in-game quantum AI to see if you’ve indeed mastered Quantum TiqTaqToe yet. At the hardest setting, winning even 30% of the time after, say, 20 games may be extraordinary. The implementation of this AI, by the way, would have been a blog-post by itself. For the curious, I can say it is based on the ExpectiMiniMax algorithm. As of the moment of this writing, the hardest AI setting is not available in the app yet. Keep your eyes out for an update soon though!

The future

Perhaps kids who grow up playing quantum games will acquire a visceral understanding of quantum phenomena that our generation lacks.” – John Preskill, in his recent article [6].

From the get-go, Quantum TiqTaqToe (and Quantum Chess) have had outreach as a core motivation. Perhaps future quantum engineers and quantum programmers will look back on their youth and remember playing Quantum TiqTaqToe as I remember my Commodore 64 games. I am convinced that these small steps into the realm of Quantum Games are only just the beginning of an entirely new genre of fun and useful games.

In the meantime, we are hard at work implementing an Online mode so you can play with your fellow human friends remotely too. This online mode, plus the option of fighting a strong quantum AI, will be unlockable in-game through a small fee (unless you are an educator who wishes to introduce quantum physics in class through this game; those use cases are fee-free courtesy of IQIM and NSF). Each purchase will go towards supporting the future development of exciting new Quantum TiqTaqToe features, as well as other exciting Quantum Games (Tetris, anyone?)

Just in case you missed it: the app is available both for Android devices and iPhone/iPad right now:

Get it on Google Play

 

I really hope you enjoy the game, and perhaps use it to get your friends and family excited about quantum physics. Oh, and start practicing! You never know if the online mode will bring along with it a real Quantum TiqTaqToe Tournament down the road 😉

References

[1] https://en.wikipedia.org/wiki/Tic-tac-toe

[2] The origin of this name in Dutch isn’t really certain as far as I know. Alledgedly, it is a left-over from the period in which butter, cheese and eggs were sold at the door (so was milk, but that was done separately since it was sold daily). The salesman had a list with columns for each of these three products, and would jot down a cross or a zero whenever a customer at an address bought or declined a product. Three crosses in a row would earn them praise from the boss. 

[3] https://en.wikipedia.org/wiki/Quantum_tic-tac-toe

[4] https://arxiv.org/abs/1906.05836

[5] https://en.wiktionary.org/wiki/God_does_not_play_dice_with_the_universe

[6] https://arxiv.org/abs/1801.00862

Machine learning the arXiv

Over the last year or so, the machine learning wave has really been sweeping through the field of condensed matter physics. Machine learning techniques have been applied to condensed matter physics before, but very sparsely and with little recognition. These days, I guess (partially) due to the general machine learning and AI hype, the amount of such studies skyrocketed (I admit to contributing to that..). I’ve been keeping track of this using the arXiv and Twitter (@Evert_v_N), but you should know about this website for getting an overview of the physics & machine learning papers: https://physicsml.github.io/pages/papers.html.

This effort of applying machine learning to physics is a serious attempt at trying to understand how such tools could be useful in a variety of ways. It isn’t very hard to get a neural network to learn ‘something’ from physics data, but it is really hard to find out what – and especially how – the network does that. That’s why toy cases such as the Ising model or the Kosterlitz-Thouless transition have been so important!

When you’re keeping track of machine learning and AI developments, you soon realize that there are examples out there of amazing feats. Being able to generate photo-realistic pictures given just a sentence. e.g. “a brown bird with golden speckles and red wings is sitting on a yellow flower with pointy petals”, is (I think..) pretty cool. I can’t help but wonder if we’ll get to a point where we can ask it to generate “the groundstate of the Heisenberg model on a Kagome lattice of 100×100”…

Another feat I want to mention, and the main motivation for this post, is that of being able to encode words as vectors. That doesn’t immediately seem like a big achievement, but it is once you want to have ‘similar’ words have ‘similar’ vectors. That is, you intuitively understand that Queen and King are very similar, but differ basically only in gender. Can we teach that to a computer (read: neural network) by just having it read some text? Turns out we can. The general encoding of words to vectors is aptly named ‘Word2Vec’, and some of the top algorithms that do that were introduced here (https://arxiv.org/abs/1301.3781) and here (https://arxiv.org/abs/1310.4546). The neat thing is that we can actually do arithmetics with these words encoded as vectors, so that the network learns (with no other input than text!):

  • King – Man + Woman = Queen
  • Paris – France + Italy = Rome

In that spirit, I wondered if we can achieve the same thing with physics jargon. Everyone knows, namely, that “electrons + two dimensions + magnetic field = Landau levels”. But is that clear from condensed matter titles?

Try it yourself

If you decide at this point that the rest of the blog is too long, at least have a look here: everthemore.pythonanywhere.com or skip to the last section. That website demonstrates the main point of this post. If that sparks your curiosity, read on!

This post is mainly for entertainment, and so a small disclaimer is in order: in all of the results below, I am sure things can be improved upon. Consider this a ‘proof of principle’. However, I would be thrilled to see what kind of trained models you can come up with yourself! So for that purpose, all of the code (plus some bonus content!) can be found on this github repository: https://github.com/everthemore/physics2vec.

Harvesting the arXiv

The perfect dataset for our endeavor can be found in the form of the arXiv. I’ve written a small script (see github repository) that harvests the titles of a given section from the arXiv. It also has options for getting the abstracts, but I’ll leave that for a separate investigation. Note that in principle we could also get the source-files of all of these papers, but doing that in bulk requires a payment; and getting them one by one will 1) take forever and 2) probably get us banned.

Collecting all this data of the physics:cond-mat subsection took right about 1.5 hours and resulted in 240737 titles and abstracts (I last ran this script on November 20th, 2017). I’ve filtered them by year and month, and you can see the result in Fig.1 below. Seems like we have some catching up to do in 2017 still (although as the inset shows, we have nothing to fear. November is almost over, but we still have the ‘getting things done before x-mas’ rush coming up!).

numpapers

Figure 1: The number of papers in the cond-mat arXiv section over the years. We’re behind, but the year isn’t over yet! (Data up to Nov 20th 2017)

Analyzing n-grams

After tidying up the titles (removing LaTeX, converting everything to lowercase, etc.), the next thing to do is to train a language model on finding n-grams. N-grams are basically fixed n-word expressions such as ‘cooper pair’ (bigram) or ‘metal insulator transition’ (trigram). This makes it easier to train a Word2Vec encoding, since these phrases are fixed and can be considered a single word. The python module we’ll use for Word2Vec is gensim (https://radimrehurek.com/gensim/), and it conveniently has phrase-detection built-in. The language model it builds reports back to us the n-grams it finds, and assigns them a score indicating how certain it is about them. Notice that this is not the same as how frequently it appears in the dataset. Hence an n-gram can appear fewer times than another, but have a higher certainty because it always appears in the same combination. For example, ‘de-haas-van-alphen’ appears less than, but is more certain than, ‘cooper-pair’, because ‘pair’ does not always come paired (pun intended) with ‘cooper’. I’ve analyzed up to 4-grams in the analysis below.

I can tell you’re curious by now to find out what some of the most certain n-grams in cond-mat are (again, these are not necessarily the most frequent), so here are some interesting findings:

  • The most certain n-grams are all surname combo’s, Affleck-Kennedy-Lieb-Tasaki being the number 1. Kugel-Khomskii is the most certain 2-name combo and Einstein-Podolksi-Rosen the most certain 3-name combo.
  • The first certain non-name n-gram is a ‘quartz tuning fork’, followed by a ‘superconducting coplanar waveguide resonator’. Who knew.
  • The bigram ‘phys. rev.’ and trigram ‘phys. rev. lett.’ are relatively high up in the confidence lists. These seem to come from the “Comment on […]”-titles on the arXiv.
  • I learned that there is such a thing as a Lefschetz thimble. I also learned that those things are called thimbles in English (we (in Holland) call them ‘finger-hats’!).

In terms of frequency however, which is probably more of interest to us, the most dominant n-grams are Two-dimensional, Quantum dot, Phase transition, Magnetic field, One dimensional and Bose-Einstein (in descending order). It seems 2D is still more popular than 1D, and all in all the top n-grams do a good job at ‘defining’ condensed matter physics. I’ll refer you to the github repository code if you want to see a full list! You’ll find there a piece of code that produces wordclouds from the dominant words and n-grams too, such as this one:

caltechwordcloud.png

For fun though, before we finally get to the Word2Vec encoding, I’ve also kept track of all of these as a function of year, so that we can now turn to finding out which bigrams have been gaining the most popularity. The table below shows the top 5 n-grams for the period 2010 – 2016 (not including 2017) and for the period 2015 – Nov 20th 2017.

2010-2016

2015 – November 20th 2017

Spin liquids  Topological phases & transitions
 Weyl semimetals  Spin chains
 Topological phases & transitions  Machine learning
 Surface states  Transition metal dichalcogenides
 Transition metal dichalcogenides  Thermal transport
 Many-body localization  Open quantum systems

Actually, the real number 5 in the left column was ‘Topological insulators’, but given number 3 I skipped it. Also, this top 5 includes a number 6 (!), which I just could not leave off given that everyone seems to have been working on MBL. If we really want to be early adopters though, taking only the last 1.8 years (2015 – now, Nov 20th 2017)  in the right column of the table shows some interesting newcomers. Surprisingly, many-body localization is not even in the top 20 anymore. Suffice it to say, if you have been working on anything topology-related, you have nothing to worry about. Machine learning is indeed gaining lots of attention, but we’ve yet to see if it doesn’t go the MBL-route (I certainly don’t hope so!). Quantum computing does not seem to be on the cond-mat radar, but I’m certain we would find that high up in the quant-ph arXiv section.

CondMat2Vec

Alright, finally time to use some actual neural networks for machine learning. As I started this post, what we’re about to do is try to train a network to encode/decode words into vectors, while simultaneously making sure that similar words (by meaning!) have similar vectors. Now that we have the n-grams, we want the Word2Vec algorithm to treat these as words by themselves (they are, after all, fixed combinations).

In the Word2Vec algorithm, we get to decide the length of the vectors that encode words ourselves. Larger vectors means more freedom in encoding words, but also makes it harder to learn similarity. In addition, we get to choose a window-size, indicating how many words the algorithm will look ahead to analyze relations between words. Both of these parameters are free for you to play with if you have a look at the source code repository. For the website everthemore.pythonanywhere.com, I’ve uploaded a size 100 with window-size 10 model, which I found to produce sensible results. Sensible here means “based on my expectations”, such as the previous example of “2D + electrons + magnetic field = Landau levels”. Let’s ask our network some questions.

First, as a simple check, let’s see what our encoding thinks some jargon is similar to:

  • Superconductor ~ Superconducting, Cuprate superconductor, Superconductivity, Layered superconductor, Unconventional superconductor, Superconducting gap, Cuprate, Weyl semimetal, …
  • Majorana ~ Majorana fermion, Majorana mode, Non-abelian, Zero-energy, braiding, topologically protected, …

It seems we could start to cluster words based on this. But the real test comes now, in the form of arithmetics. According to our network (I am listing the top two choices in some cases; the encoder outputs a list of similar vectors, ordered by similarity):

  • Majorana + Braiding = Non-Abelian
  • Electron + Hole = Exciton, Carrier
  • Spin + Magnetic field = Magnetization, Antiferromagnetic
  • Particle + Charge = Electron, Charged particle

And, sure enough:

  • 2D + electrons + magnetic field = Landau level, Magnetoresistance oscillation

The above is just a small sample of the things I’ve tried. See the link in the try it yourself section above if you want to have a go. Not all of the examples work nicely. For example, neither lattice + wave nor lattice + excitation nor lattice + force seem to result in anything related to the word ‘phonon’. I would guess that increasing the window size will help remedy this problem. Even better probably would be to include abstracts!

Outlook

I could play with this for hours, and I’m sure that by including the abstracts and tweaking the vector size (plus some more parameters I haven’t even mentioned) one could optimize this more. Once we have an optimized model, we could start to cluster the vectors to define research fields, visualizing the relations between n-grams (both suggestions thanks to Thomas Vidick and John Preskill!), and many other things. This post has become rather long already however, and I will leave further investigation to a possible future post. I’d be very happy to incorporate anything cool you find yourselves though, so please let me know!