shoyer

import numpy as np

def convert_label_to_indexer(index, label):
    """Given a pandas.Index and labels (e.g., from __getitem__) for one dimension,
    return an indexer suitable for indexing an ndarray along that dimension
    """
    if isinstance(label, slice):
        indexer = index.slice_indexer(label.start, label.stop, label.step)
    elif np.isscalar(label):
        indexer = indexer.get_loc(label)

shoyer

shoyer

shoyer

shoyer

"City"	"City_Alternate"	"Country"	"Latitude"	"Longitude"	"Country_ISO3"	"pop1950"	"pop1955"	"pop1960"	"pop1965"	"pop1970"	"pop1975"	"pop1980"	"pop1985"	"pop1990"	"pop1995"	"pop2000"	"pop2005"	"pop2010"	"pop2015"	"pop2020"	"pop2025"	"pop2050"
"Sofia"	""	"Bulgaria"	42.70	23.33	"BGR"	520.00	620.00	710.00	810.00	890.00	980.00	1070.00	1180.00	1190.00	1170.00	1130.00	1170.00	1180.00	1210.00	1230.00	1240.00	1236
"Mandalay"	""	"Myanmar"	21.97	96.08	"MMR"	170.00	200.00	250.00	310.00	370.00	440.00	500.00	560.00	640.00	720.00	810.00	920.00	960.00	1030.00	1170.00	1310.00	1446
"Nay Pyi Taw"		"Myanmar"	19.75	96.10	"MMR"	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	60.00	930.00	1020.00	1180.00	1320.00	1461
"Yangon"	"Rangoon"	"Myanmar"	16.87	96.12	"MMR"	1300.00	1440.00	1590.00	1760.00	1950.00	2150.00	2380.00	2630.00	2910.00	3210.00	3550.00	3930.00	4090.00	4350.00	4840.00	5360.00	5869
"Minsk"	""	"Belarus"	53.89	27.57	"BLR"	280.00	410.00	550.00	720.00	930.00	1120.00	1320.00	1470.00	1610.00	1650.00	1700.00	1780.00	180

shoyer

shoyer

"""Proof of concept for implementing bottleneck style aggregators with numba

These functions aggregate over any number of axes, just like the built-in
numpy functions, e.g.,
- nansum(x) # aggregates over all axes
- nansum(x, axis=1) # aggregates over axis 1
- nansum(x, axis=(0, 2)) # aggregtes over axes 0 and 2
"""
import functools

shoyer

shoyer

def transform_state_vector(rho, U):
    N = len(U)
    if rho.shape[-1] == N:
        # rho is in Hilbert space
        pass
    elif rho.shape[-1] == N ** 2:
        # rho is in Liouville space
        U = np.kron(U, U)
    else:
        raise ValueError('basis transformation incompatible with '

shoyer

import numpy as np
from numpy.lib.stride_tricks import as_strided

def broadcast_to(array, shape):
    """Expand a numpy.ndarray to a new shape according to broadcasting rules
    """
    array = np.asarray(array)
   
    # will raise ValueError if shapes incompatible 
    np.nditer((array,), itershape=shape)