Christoph Heindl

physics.units - dimensional analysis/nondimensionalization addition

8

Hey! I have developed a prototype Python package that deals with physical dimensional analysis: https://github.com/cheind/sympy-nondim In particular, sympy-nondim calculates from an unknown relation of (dimensional) variables, a new relation of...

Hey, I'm currently benchmarking ZMQ/pyzmq for an application scenario that demands high throughput. For this reason, I created a minimal working example that computes throughput metrics and that runs either...

**Describe the bug** Maybe I'm not seeing something, but I believe the example https://github.com/ReactiveX/RxPY/blob/master/examples/parallel/timer.py is broken for RxPY 4.0.4. Instead of printing the output it prints nothing, which is due...

Extended benchmarks

3

Hi, Just wanted to let you know that I've been benchmarking various Python LAP solving implementations. https://github.com/cheind/py-lapsolver/tree/develop#benchmarks Until recently, the solver from [lapsolver](https://github.com/cheind/py-lapsolver/tree/develop) (my project), was actually the fastest among...

Hey, I've added a barebone GCSL implementation to https://github.com/cheind/gcsl, featuring a new goal-reaching cartpole environment. I believe the code is quite easy to digest in its current form and might...

Hi, the shape database loader does not fully load the entire database. There are at least 2/3 of the files missing. I've discovered this while working on https://github.com/cheind/dest. A correct...

Not sure about this right now, but when iterating in the gen. procrustes we might suffer from a scale / orientation drift. References to this are given in https://graphics.stanford.edu/courses/cs164-09-spring/Handouts/paper_shape_spaces_imm403.pdf

Reconsider the softmax distribution

2

PixelCNN++ argues that neighboring intensity usually correlate is not captured by the softmax distribution. Instead, they propose a mixture model consisting of the logitic distribution (like normal but with heavier...

Decouple input and output representations

2

currently we assume that what we get as input is what we will predict as output (just shifted). However, thinking towards other research areas it might make sense that we...

drawback: all conditions would have to be embeddable (i.e integers) - except the embedding layer is not part of the model, but lives outside.