davidwhogg

0

davidwhogg

import pyfits
import csv

if __name__ == "__main__":
    hdulist = pyfits.open("foo.fits")
    blob = hdulist[-1].data
    with open('foo.csv', 'wb') as f:
        writer = csv.writer(f)
        writer.writerows(blob)

davidwhogg

echo Hello World

davidwhogg

First install the Python module:

pip install casjobs

Then set the environment variables:

export CAS_DR9_WSID="1234567890" # get yours here: http://skyserver.sdss3.org/CasJobs/ChangeDetails.aspx

davidwhogg

<img width="20" height="20" align="absmiddle" src="http://octodex.github.com/images/nyantocat.gif" class="emoji" alt=":nyantocat:">

davidwhogg

import numpy as np
import time as tt

def main():
    timing = np.array([])
    log2N = np.arange(6) + 8
    for N in 2 ** log2N:
        uu = np.random.normal(size=(N, N+2))
        matrix = np.dot(uu, uu.T)
        print matrix.shape

davidwhogg

import numpy as np
import scipy.linalg as la

# create data matrix and design matrices A (for sinusoids) and B (for nuisances) and M (for both)
y = np.random.normal(size=(100)) # data
A = np.random.normal(size=(2, 100)) # components we care about
B = np.random.normal(size=(10, 100)) # nuisance components
M = np.vstack((B, A))
C = np.diag(0.25 * np.ones(100))
Cinv = np.diag(4.00 * np.ones(100))

davidwhogg

import numpy as np

def int_string(x, base, digits):
  xx = x + 0
  str = ""
  for i in range(digits):
    str = (xx % base).__str__() + str
    xx /= base
  return str

davidwhogg

import numpy as np

def hoggsumexp(qns, dqn_dams, diag=False):
    """
    # purpose:
    - Computes L = log(sum(exp(qns, axis=-1))).
    - Also computes its M-dimensional gradient components dL / da_m.

    # input
    - qns: ndarray of shape [n1, n2, n3, ..., nD, N]

davidwhogg

import numpy as np
from scipy.interpolate import interp1d

def weighted_median(xs, ws=None):
    if ws is None:
        ws = np.ones_like(xs)
    totalw = np.sum(ws)
    sindx = np.argsort(xs) # expensive                                                                                                 
    cs = np.cumsum(ws[sindx]) / totalw
    symcumsum = 0.5 * (cs + np.append(0., cs[0:-1]))