Aart's Blog

It has been a while since I added major features to Reversi for Android , but I recently added something I had been planning for a long time: a notation window and full game navigation, similar to Chess for Android. The result is shown below. No more artificial restrictions on the undo, one can simply go back and forth in the full game, and try different strategies to learn from one's mistakes! An interesting factoid is that Reversi uses a slightly different board orientation for the algebraic notation: the a1 square is in the upper left corner with the h8 square in the bottom right corner. This goes a bit against my chess intuition, but obviously I had to follow the Reversi convention.

Around the holidays I am finally getting some spare time to implement new features for Chess for Android . I am actively "in the zone" working on: The ability to offer a draw or resign during an ongoing game. Since the UCI protocol does not provide this feature, the GUI will accept the draw simply based on past evaluations of the position. The ability to change the users name and ELO rating, which will appear in the header for all user games that are saved in PGN format to the SD card. The ability to connect to FICS , the free internet service. This is of course the most work and currently still in prototype stage. The first release will probably start simple by just allowing to watch demo games.

Before moving to the next book, first a posting on an example given in the Tensorflow book by Ramsundar and Zadeh. The chapter on convolutional neural networks discusses training a tensorflow architecture to recognize handwritten digits taken from the MNIST dataset. The given Python code automatically downloads the dataset from the Web and partitions the labeled data into a train, validation, and test set (as explained in the book, used to train the network, validate the performance of the model, and test the final model, respectively). The ultimate objective of the algorithm is, given the tensor with handwritten digits shown below to the left, finding the tensor with labels shown below to the right.  [7 2 1 0 4 1 4 9  5 9 0 6 9 0 1 5  9 7 3 4 9 6 6 5  4 0 7 4 0 1 3 1  3 4 7 2 7 1 2 1  1 7 4 2 3 5 1 2  4 4 6 3 5 5 6 0  4 1 9 5 7 8 9 3] Clearly a fun example, since recognizing digits is an intuitive, but non-straightforward problem. I highly recommend running a

A brief impression after finishing the book "TensorFlow for Deep Learning - From Linear Regression to Reinforcement Learning" (by Ramsundar and Zadeh).  The book introduces the concept of tensors, primitives and architectures for deep learning, and the basics of regression, various neural networks, hyperparameter optimization, and reinforcement learning. The art work in the figures is beautiful (something that convinced me to buy the book). The TensorFlow code examples can be downloaded from the book's website, making it easy to follow along with the discussion the book. The book falls a bit short on detailed explanation, however. I found that many times when the discussion in the book was about to get interesting, it referred to other work for details instead. Several architectures were merely "explained" with a figure, no accompanying details in the text. In addition, although I realize how hard it is to avoid errors in a book, the given linear r

As a CS student, a long time ago in a country far away, I was very interested in AI (Artificial Intelligence), and not just for chess playing programs. In fact, if it weren't for my professor convincing me to continue with compilers and high-performance computing, I may have ended up specializing in the field of AI. Perhaps lucky for me, since AI has gone through many rounds of boom-and-bust. Nowadays, however, machine learning in general, and deep learning in particular really seem to have taken AI in a very promising new direction. Since I feel machine learning will become an important, if not mandatory skill for computer scientists, I decided to buy a few books on TensorFlow and familiarize myself with the new paradigm. For starters, I bought the three O'Reilly books below (other recommendations are welcome) and plan to do a few brief follow-up posts on this topic.

After all the fun I had connecting Chess for Android with the Millennium  over Bluetooth, I was curious if I could provide similar support for the DGT electronic chess boards. Some of these have Bluetooth capabilities, most use USB connections, and some older models, like the one I have, still use the RS-232 connector. To my pleasant surprise, by combining the original serial cable of DGT with a USB-to-serial cable and a female-USB-to-USB-C cable, I was able to get an actually working connection between my DGT board and my tablet or phone. Next was implementing support in Chess for Android. Luckily, DGT kindly shared the protocol documentation with me, and after a fun weekend of hacking, Chess for Android now proudly supports DGT electronic chess boards as well.

Recently I have been very active adding new features to Chess for Android again. I have added support to connect to the Millennium ChessGenuis Exclusive electronic chessboard, added a new piece set (thanks Bryan Whitby), added various engine related features requested by users, and switched to the much better model where users can enable (and thus disable) permissions just for the features they like. Now, I also added optional piece animation and algebraic notation on the board. Hopefully this makes watching ongoing tournaments more smooth, as illustrated below for a match between Komodo and DiscoCheck. Keep an eye on Google Play for updates!

Did you ever wonder what an Android phone screen looks like under a microscope? So did I. So at the start of this weekend, I got the microscope out and took some photos, collected in one picture below. The results are amazing. What looks white to the naked eye, is really a large field of RGB (red-green-blue) elements under magnification. All colors are, of course, obtained by adjusting the brightness of each RGB element appropriately, as illustrated in this picture too.

I am rolling out Chess for Android version 5.4 on Google Play . Besides minor improvements, the major new feature consists of draw and resign adjudication during chess engines tournaments. As shown below, a new tournament dialog has been implemented which shows, besides familiar older options, a section for draw and resign adjudication. If during a game, after the given move number and during the given move count, the score drops below the requested draw score (in cp) or exceeds the requested resign score (in cp, either consistently for white or for black), the game is adjudicated rather than played in full. This feature has been requested many times by tournament managers to avoid wasting time playing e.g. boring drawn games until the 50-move rule applies. See this talkchess posting for an example game. As usual, let me know if you encounter problems with the new release. Also, I could use some help translating the new strings into several languages (most will display Eng

Let’s revisit the Micro-KIM memory map, introduced in the third tutorial. +-----------+ | 2K EPROM  |$1fff | monitor   | | program   |$1800 +-----------+ | 6532 RIOT |$17ff | I/O, timer| | and RAM   |$1740 +-----------+ | optional  |$173f | I/O, timer| | and RAM   |$1400 +-----------+ |           |$13ff | 5K RAM    | |           |$0000 +-----------+ Since the default kit (without any expansion) only uses the lower address bits to access 8K, memory repeats itself every 8K. You can verify this by storing and inspecting values in, for instance, addresses $0000 and $2000. Any value stored in one address will show up in the other. Although an interesting factoid, there is no reason to let Micro-KIM programs address anything outside the range $0000-$1fff. Addresses $0000-$13ff contain 5K free RAM (another interesting factoid: the Micro-KIM actually wastes 3K of its 8K RAM chip to keep compatibility with the original KIM-1). This memory region can be used to store da