Artipixoids!

Xentica 0.2.0 Release

2019-06-28T08:11:11+00:00

Good day, the next major version is out at last, and we are glad to present you Xentica 0.2.0 release. A lot of new features were added to the core engine. And the good news, it’s backward compatible with the previous version, so you could upgrade safely.

The main goal accomplished, we built the EvoLife model purely with Xentica. The structures with the most interesting dynamics (so-called ‘Bliambas’) are achievable with it, although with a bit different meta-parameters. You could check how it runs on a demo video. Another interesting stable ecosystem was discovered in the process of development, video is available. Both experiments shot on those videos are included in the official EvoLife example, you could always reproduce them yourself.

Also, there is an experiment with EvoLife modification in 6D, but you should have a powerful video card with at least 9GB RAM in order to run it. And even with GTX 1080Ti, it runs very slow (~7FPS). If you are interested, check the glitchy time-lapse video of this experiment.

In order to make run all the things described above, we had to implement a number of new features to the core engine:

Per-cell RNG property. If specified, each cell will have a separate pseudo-random stream, with the ability to draw direct integer values and values from the uniform distribution.
Dynamic totalistic rule property. This is a regular integer property with the encoded lifelike/totalistic rule. It could also easily test a cell for born / dead status.
Genome crossover. It helps you to crossover several genomes in a stochastic way, bit by bit. Also, has an option to mutate a gene with a certain probability.
Meta-parameters. Constants that could be changed per each experiment. Makes experimenting more convenient, since you could easily inherit and re-use them.
Interactive parameters. A special meta-parameters case, that could be changed in runtime. It would be useful for the values you control interactively, like render options or mutation probability.
Math functions. A collection of functions, mapping to C counterparts. min/max, typecasting and so on.
Genome coloring. A collection of algorithms helping your evolutionary models to look just awesome out of the box.

A solid part of the work also done to make the testing process super easy and informative. Now, we have a powerful test suite, managed by a single tox command. It will:

run tests in all supported Python environments;
do sanity checks for the build;
test how docs are building;
run code style checks;
do code static analysis;
build code coverage;
run a benchmark.

As well, almost all major bugs were fixed, except the new one with DeferredExpression as an initial value for Variable.

You could check the full list of changes for this version in the CHANGELOG.

As in the previous release, no changes were made for Moire, you can safely use 0.1.0 with the latest Xentica. Although, we found several bugs for cases when the frame rate is too low. They gonna be fixed along with the next Xentica release.

Now, for future plans. As the main goal, we’ll focus on a Stochastic Corrector (or alike) model in the next version. Most probably, it will be a bugfix release (0.2.1), since we already have all the necessary features to implement it. But, there is also a couple of features in the backlog that would be necessary. If so, we’ll skip to the next major (0.3) version.

Traditionally, there are performance results, in gigacells per second:

GPU	Tester	Conway’s Life	Shifting Sands	Noise TV	EvoLife
GTX 1080Ti	@a5kin	6.38 Gcells/s	7.25 Gcells/s	7.33 Gcells/s	1.31 Gcells/s

If you wish to extend this table, please send your own results, along with your GPU model. You could run the benchmark with the tox -e benchmark command.

Stay connected,
Xentica/Moire Team

In Anticipation Of Xentica 0.2

2019-05-14T18:46:00+00:00

Greetings from every cell of our hearts! Xentica (formerly HECATE) has anniversary today: 2 years since the first actual code committed. And we are celebrating it with the 500th commit to the master branch. Especially proud that it deals with dimensionality, one of the main Xentica features.

So, what is the status for the moment? The bad news first. We missed the deadline on May 5. We excluded almost all secondary features from 0.2, in favor of the main goal accomplished. But the good news, we are very close to the actual release. And even without secondary goals, it looks much more powerful than the last version. It has meta-parameters, it has per-cell RNG, genetics, lots of helpers, and many bugs fixed. That’s all we need to build EvoLife model and beyond it.

Current EvoLife/Xentica model runs smoothly, but unfortunately, it shows a bit different behavior from the legacy version. The reasons are unknown, we’ll investigate it further. But if the investigation will take too long, we will skip it for the next bugfix version. StochasticCorrector / MSE model also left for the future version. We still need to finish tests, docs, and other maintenance stuff. Hopefully, you can expect 0.2 release somewhere in late May.

Stay connected,
Xentica/Moire Team

Xentica 0.2 Development Started

2019-03-02T20:11:23+00:00

High fives, if you did not notice, Xentica 0.2 active development just started this week. We have a primary goal to

run Evolife model on the renewed Xentica engine

And the secondary goals are:

integrate meta-params with Moire for automated UI build
give the ability to edit those params interactively
save/load experiments at runtime

Hopefully, most of those changes would be done before the Xentica 2nd anniversary (May 5th).

There are more things to be done for real-life science application, as we discussed with @Arcprotorp in Artipixoids utility in Origin of Life research issue.

So, probably we could sacrifice secondary goals (dealing most with visual/usability stuff) in favor of early core features, like: fields, distributions and other tools necessary to run Xentica powered version of MSE or similar model.

Moire itself would most likely stay at 0.1 version, giving us just basic visualisation abilities. We’ll proceed to automated UI generation only after Xentica 0.2 is released.

Traditionally, we have an updated performance results. Author’s shining new GTX 1080Ti, and another GTX 1050Ti version by @Arcprotorp were added.

GPU	Tester	Conway’s Life	Shifting Sands
GTX 1080Ti	@a5kin	6.61 Gcells/s	7.39 Gcells/s
GTX TITAN	@a5kin	3.31 Gcells/s	3.43 Gcells/s
GTX 1050 Ti	@Arcprotorp	2.29 Gcells/s	2.54 Gcells/s
GTX 1050 Ti	@robotateme	1.82 Gcells/s	2.02 Gcells/s
GTX 750 Ti	@alexeyshch	1.19 Gcells/s	1.32 Gcells/s
GTX 860M	@a5kin	1.05 Gcells/s	1.13 Gcells/s

Stay connected,
Xentica/Moire Team

Xentica 0.1.1 Release

2018-07-24T20:11:23+00:00

Greetings, today we are glad to update you with Xentica 0.1.1 bugfix release. A number of major optimization and bugfix changes were made, so we highly recommend you to upgrade to this version.

Firstly, we’ve greatly sped up the process of seed generation. It’s powered by NumPy, and running orders of magnitude faster now. The only problem introduced with this optimization, is that it take 2x more memory during generation process. So, right now, we can only generate fields taking no more than half of your available RAM. In fact, it is much better option than waiting for hours in hope for seed fitting in your whole memory.

Secondly, we’ve fixed several nasty bugs with direct buffers/variables assignments. It is now working as expected, and you can safely assign directly to the buffer/variable, it will produce the correct code without errors.

We’ve also checked the correctness of non-uniform buffers interactions. For that, we wrote a specific model called Shifting Sands. Despite of a loud name, it’s just shifting the whole board in a constant direction via buffers. As a result, we can test if buffers are working correctly in every direction. Surprisingly, they are working as expected, so we’ve just left the new model as another one example. It’s also added to the official benchmark.

In background, we’ve done a good job in refactoring, fixing most of pylint issues and making object model more consistent. Only 2 pylint categories were disabled in the process: 1) no-member, since it can not detect CellularAutomaton members generated by the metaclass, 2) too-few-public-methods, because experiment classes are holding “data” only, which allows them to be easily inherited.

No bugfixes were made for Moire, you can safely use 0.1.0 with latest Xentica.

We have enough issues for another optimization/bugfix release, but not decided yet, should we handle them in 0.1.2, or dive right into 0.2 with all new features. It will be announced later after new release planning is done.

Traditionally, there are performance results, in gigacells per second:

GPU	Conway’s Life	Shifting Sands
GTX TITAN	3.31 Gcells/s	3.43 Gcells/s
GTX 1050 Ti	1.82 Gcells/s	2.02 Gcells/s
GTX 750 Ti	1.19 Gcells/s	1.32 Gcells/s
GTX 860M	1.05 Gcells/s	1.13 Gcells/s

Special thanks to @alexeyshch and @robotateme for assistance in testing.

Stay connected,
Xentica/Moire Team

Xentica/Moire 0.1.0 Release

2018-05-05T15:00:00+00:00

Today is the anniversary of both Xentica and Moire, a year since the development started. On this day, we are glad to present you the first public release of the system.

Version 0.1.0 is a minimum viable product. The main goal was to make Game Of Life model runnable with Xentica/Moire ecosystem. We succeed with it, and in the process, some cool extra features were implemented, like multi-dimensionality and multiple properties per cell. See the full list of features in the Xentica Changelog.

Moire is a mere satellite of Xentica, but also has its own standalone example, called Noise TV. Current functionality is very minimal, you can see the detailed description in Moire Changelog.

The system was tested on Debian/Ubuntu only, installation instructions are included. If you’ll manage to run it on other OS, please update us with details/hints.

There are also a number of bugs and optimization issues we are going to fix in 0.1.x releases. Then, we will focus on version 0.2.0, the main goal of which is to run Evolife with Xentica/Moire.

Current performance results for Game Of Life, in gigacells per second:

GPU	Performance
GTX 860M	1.05 Gcells/s
GTX TITAN	3.27 Gcells/s

Special thanks to @robotateme for testing and help with installation instructions.

Stay connected,
Xentica/Moire Team

Xentica / Moire: Last Preparations

2018-02-28T18:46:00+00:00

We are doing last preparations before first Xentica release. Actually, it is already fully prepared to be released. It has a solid documentation, covered 100% by tests, and setup script is ready too. The only thing left is to publish the package at PyPI.

But along with Xentica we also have to release Moire, since they are closely connected. And Moire is in a bit far from release state. So, in order to publish both packages simultaneously, first we have to document Moire and write at least some tests. Some refactoring and integration with CI systems are desirable too.

Hopefully, it will not take a lot of time, cause Moire is much smaller package in comparison to Xentica.

Stay connected,
Xentica/Moire Team