Problems with PyTorch in Conda

So... I made a tutorial for using the Deep Lagrangian Fluids, but the training gets stuck in what I suppose it is a deadlock. However, the training works fine with Tensorflow. I know I said it was impossible to install the version I needed via conda (still is), but via pip, it is possible.

I'll post the tutorial update ASAP. There are some differences in the steps. I'm thinking about how to organize them.

Starting Unity

I finally decided to start learning Unity. Before building a Game Engine, I need to know how one works, right? For now I'll follow the official tutorial, and in the end I want to make a 2D spaceship game, probably similar to Chicken Invaders. I played it a lot in my childhood.

See ya.

Japanese Keyboard in Ubuntu

Hello, there. I'll just post the step by step to add the Japanese Keyboard in Ubuntu 20.04 because I always forget how.

It's here.

Cheers.

Starting Arena Engine

Yes, yes. It's been a long time. But now I found a goal for my engine. It will only implement a feature for another engine, like Quixel that was integrated in Unreal 5. Since I'm studying fluid (liquid) simulation with Deep Learning, I'm going to implement the Smoothed Particle Hydrodynamics (SPH). This is a relative old technique, so there are many works, even using CUDA and open source. This will help me a lot. 

Nevertheless, wish me luck.

New page released

I just made a tutorial to build and install all the dependencies to run the Deep Learning network I'm studying for fluid simulation, right here. For now, I can just execute it and visualize the results. The presentation video can be remade on your computer if you follow all the steps, no skipping or changing the order (please).

Soon I'm gonna find out the next step.

If you are interested on this topic and you want to work with Disney, check out the lib this project uses: Partio. And also, the SPlisH SPlasH that uses it.

I'm Back

Hello there. Almost a year, huh? 

Last time I stopped because I needed to generate data for a paper, which is now at the last step to be accepted. It is about Rigid Registration with Lie Groups. When it is released, I'll post here.

Since then, I started to research about Deep Learning for this topic. But I found out it is very complicated when you don't know much of the theory. 

For instance, I made a Toy Model with a Multilayer Perceptron to compute the triangle area. However, I couldn't make it have a good accuracy. 

So the task was assigned to another student who knows better about neural networks, and he concluded that triangles with very small area are being interpreted as outliers by the network, and making the training harder. Also, there is an error accumulation due to numerical instability, so the training needed 64 bit float and the error measurement, a 32 bit float.

Right now, I changed my research to fluid simulation with Deep Learning, far more interesting to me. I just made this code work, and very soon I'll post the step by step to make it work for you too.

See ya.

Where was I?

Hello, future me.

In case you are wondering where did I go these days since the last post, I've been busy studying many different things, from deep learning to gamification. Both are courses from Coursera, where you can watch the videos, but in order to get the certification, you must pay. Fair enough.

Also, I upgraded my Ubuntu version to 20.04, and it is a lot prettier. Now my simple laptop is faster and my desktop cleaner. Luckily, I manage to compile Yade DEM from source to activate the Potential Particles computation, which once was a real headache. But not before installing 1GB of dependencies. I know I probably wont use Yade anymore, but I found a workshop from a Ph.D in rock mechanics that opened my eyes for certain details when dealing with DEM:

1. If the particles are getting high velocities, decrease the time step for numerical stability;
2. If the real sample exhibits plastic behavior, select the simulation material that can recreate it;
   
3. Know the shape and size distribution of the grain of the material you want to simulate.

Besides, my sensei recommended me a YouTube channel that teaches Japanese in Japanse. The name is 日本語の森, and they teach grammar of any level.

So, I need to watch some of these videos, test Yade, and get the Deep Learning networks for rigid registration, that I found here, working.

I'm feeling like a workaholic. If it wasn't for my dog getting my attention, I would be lost.

PS: Today the blog completes 1 month!