gngdb

import torch

if  __name__ == '__main__':
    X = torch.randn(100)
    out_shape = (100,100)
    idxs = torch.randint(high=100, size=out_shape).long()
    assert torch.abs(X[idxs] - X.index_select(0, idxs.view(-1)).view(*out_shape)).max() < 1e-3
    from timeit import timeit
    setup = 'import torch; X = torch.randn(100); out_shape=(100,100); idxs = torch.randint(high=100, size=out_shape).long()'
    print("X[idxs]: ", timeit("_ = X[idxs]", setup=setup, number=100))

gngdb

# implementation of https://arxiv.org/abs/1504.04788
import torch
import torch.nn as nn
import torch.nn.functional as F
import xxhash


class HashFunction(object):
    """Hash function as described in the paper, maps a key (i,j) to a natural number
    in {1,...,K_L}"""

gngdb

import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F

import numpy as np
from scipy import optimize

from obj import PyTorchObjective

gngdb

gngdb

import torch
from functools import reduce
import math

def is_power(n):
    # https://stackoverflow.com/a/29480710/6937913
    n = n/2
    if n == 2:
        return True
    elif n > 2:

gngdb

gngdb

gngdb

import random
from io import StringIO

def write_bitstream(fname, bits):
    # bits are a string of ones and zeros, based on this
    # stackoverflow answer: https://stackoverflow.com/a/16888829/6938913
    # was broken due to utf-8 encoding using up to 4 bytes: https://stackoverflow.com/a/33349765/6937913
    sio = StringIO(bits)
    with open(fname, 'wb') as f:
        while 1:

gngdb

from tqdm import tqdm

class TrainingProgress(tqdm):
    """Make the progress bar store progress as it displays it when the log method is called."""
    def log(self, *args, **kwargs):
        if 'trace' not in self.__dict__.keys():
            self.trace = {}
        # store anything with an integer or float datatype in the trace dictionary
        # with global index as the key
        for k in kwargs:

gngdb

import math

import numpy as np
from scipy.fftpack import dct

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim