rejuvyesh

using SimpleChains
function f(x)
    N = Base.isqrt(length(x))
    A = reshape(view(x, 1:N*N), (N,N))
    expA = exp(A)
    vec(expA)
end
  
T = Float32;
D = 2 # 2x2 matrices

rejuvyesh

# install tinycudann via 
# pip install git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch

import torch
import tinycudann as tcnn
import time

class TCNNMatrixExponentEstimator1(torch.nn.Module):
    def __init__(self, hidden=16) -> None:
        super().__init__()

rejuvyesh

using PyCall
using CUDA
using DLPack
using Test
#using Zygote
#using ChainRulesCore

@show DLPack.PYCALL_NOOP_DELETER

jax = pyimport("jax")

rejuvyesh

using PyCall
using DLPack
using Test
using Zygote
using ChainRulesCore

torch = pyimport("torch")
functorch = pyimport("functorch")
dlpack = pyimport("torch.utils.dlpack")
py"""

rejuvyesh

using Flux

qdim = 2
nn = Chain(Dense(qdim, 32, tanh), Dense(32, 2));
q = rand(2, 5);

function jac(x)
    o = nn(x)
    return reduce(hcat, [o[:, i] for i in 1:size(x)[end]])
end

rejuvyesh

#!/usr/bin/env python3
#
# File: test_dircol.py
#
import numpy as np

import torch

from optimalcontrol.dircolproblem import DIRCOLProblem
from mechamodlearn import utils

rejuvyesh

using MuJoCo

modelfile = "test/humanoid.xml"
pm = mj_loadXML(modelfile)  # Raw C pointer to mjModel
pd = mj_makeData(pm)            # Raw C pointer to mjData

m, d = mj.mapmujoco(pm, pd) # wrap with our jlModel, jlData types

# we can manipulate data in the raw C structs now
nq = mj.get(m, :nq)

rejuvyesh

rejuvyesh

\newcommand{\ba}{\mathbf{a}}
\newcommand{\bb}{\mathbf{b}}
\newcommand{\bc}{\mathbf{c}}
\newcommand{\bd}{\mathbf{d}}
\newcommand{\be}{\mathbf{e}}
\newcommand{\bg}{\mathbf{g}}
\newcommand{\bh}{\mathbf{h}}
\newcommand{\bi}{\mathbf{i}}
\newcommand{\bj}{\mathbf{j}}
\newcommand{\bk}{\mathbf{k}}

rejuvyesh

agents:
  - idx: 0
    status: 1
    rate: 1.0
    learner: 
      cls: SharedFFA3CLearner
      lidx: 0
  - idx: 1
    status: 1
    rate: 1.0