def lnlikeHF(pars, samples, obs, u, extra=False):
    '''
    Generic likelihood function for importance sampling with any number of
    dimensions.
    Now with added jitter parameter (hierarchical)
    obs should be a 2d array of observations. shape = (ndims, nobs)
    u should be a 2d array of uncertainties. shape = (ndims, nobs)
    samples is a 3d array of samples. shape = (ndims, nobs, nsamp)
    if extra == True, the sigma has both a slope and intercept
    '''
    ndims, nobs, nsamp = samples.shape
    ll = np.zeros((nobs, nsamp*nobs))
    for i in range(nobs):
        if extra:
            inv_sigma2 = 1.0/(u[1, :][i]**2 + \
                    pars[2]**2 + pars[3] * obs[0, :][i])
        else:
            inv_sigma2 = 1.0/(u[1, :][i]**2 + \
                    (pars[2]*model1(pars, obs[0, :][i]))**2)
        ll[i, :] = -.5*((obs[1, :][i] - model(pars, samples))**2*inv_sigma2) \
                + np.log(inv_sigma2)
    loglike = np.sum(np.logaddexp.reduce(ll, axis=1))
    if np.isfinite(loglike):
        return loglike, loglike
    return -np.inf, None

	def init(self):
		self.y_d = self.parms.y_d.copy()

		# Load Models
		self.models = list()
		for nm in xrange(int(self.parms.num_models)):
			self.models.append(mdl.model(self.parms))

		# Load multimodel setup
		self.multimodels = mmdl.multimodel(float(self.parms.NinSoftmax),float(self.parms.Nout),self.parms.num_models,self.parms.eta_s, self.parms.eta_g)

		self.sdpos = np.zeros((self.parms.Nout,))

		# Reading and setting al the initial parameters
		t1 = self.parms.t1
		dt = self.parms.dt
		delta = self.parms.delta
	   
		self.sysInp = np.zeros((self.parms.sNin, t1 / dt + delta + 1))
		self.sysInp2 = np.zeros((self.parms.Nout, t1 / dt + delta + 1))
		self.sysOutp = np.zeros((self.parms.Nout, t1 / dt + delta + 1))

		self.eeVal = self.sysInp
		self.inited = 1

		self.model_e = np.zeros((self.parms.Nout, self.parms.num_models))
		self.model_outputs = np.zeros((self.parms.Nout, self.parms.num_models,t1 / dt + delta + 1))
		self.h = np.zeros((self.parms.num_models,t1 / dt + delta + 1))
		self.model_doutputs = np.zeros((self.parms.Nout, self.parms.num_models))

		self.i = delta
 
		self.t = 0

def keyring_present_type(**kwargs):
    """
    Check if keyring exists on disk

    CLI Example:

        salt '*' sesceph.keyring_admin_save \\
                '[mon.]\n\tkey = AQA/vZ9WyDwsKRAAxQ6wjGJH6WV8fDJeyzxHrg==\n\tcaps mon = \"allow *\"\n' \\
                'cluster_name'='ceph' \\
                'cluster_uuid'='cluster_uuid'
    Notes:

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.

    cluster_name
        Set the cluster name. Defaults to "ceph".
    keyring_type
        Set the keyring type
    """
    keyring_type = kwargs.get("keyring_type")
    if (keyring_type is None):
        raise Error("keyring_type is None")
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    try:
        u.defaults_refresh()
    except:
        pass
    keyobj = keyring.keyring_facard(m)
    keyobj.key_type = keyring_type
    return keyobj.present()

    def build_agg_snapshots(self, db='./database'):
        conn = sqlite3.connect(db)
        c = conn.cursor()

        # force model rebuild if created within last n days.
        created_within = 7
        # select models which require rebuilding.        
        cmd = c.execute('select mid, uid, version, created_on, last_updated_on, agg_fx, granularity \
                         from model \
                         where julianday(current_date) - julianday(last_updated_on) >= expires_in or \
                               rebuild = 1 and julianday(current_date) - julianday(last_updated_on) >= earliest_rebuild or \
                               julianday(current_date) - julianday(created_on) <= ?', (created_within,)) 
        r = cmd.fetchall()
        for m in r:
            mdl = model(m[0], m[2], m[6]) 
            # update model # && force deletion existing aggregrate snapshot.
            c.execute('delete from agg_location_snapshot where mid = ?', (mdl.id(), ))
            mdl.inc_version()
            c.execute('update model \
                       set version = ?, last_updated_on = current_date, \
                       nr_matches = 0, nr_successful = 0, streak = 0, last10 = 0 \
                       where mid = ?', (mdl.version(), mdl.id()))
            # rebuid aggregrate snapshot.
            self.__build_agg_snapshot(mdl, c)

        c.close()
        conn.commit()
        conn.close()

def pool_del(pool_name, **kwargs):
    """
    List all cephx authorization keys

    CLI Example:

        salt '*' sesceph.pool_del pool_name \\
                'cluster_name'='ceph' \\
                'cluster_uuid'='cluster_uuid'
    Notes:

    cluster_name
        Set the cluster name. Defaults to "ceph".

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.
    """
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    u.defaults_refresh()
    u.load_confg(m.cluster_name)
    u.mon_members_refresh()
    mur = mdl_updater_remote.model_updater_remote(m)
    can_connect = mur.connect()
    if not can_connect:
        raise Error("Cant connect to cluster.")
    mur.pool_list()
    return mur.pool_del(pool_name)

def keyring_auth_list(**kwargs):
    """
    List all cephx authorization keys

    CLI Example:

        salt '*' sesceph.auth_list \\
                'cluster_name'='ceph' \\
                'cluster_uuid'='cluster_uuid'
    Notes:

    cluster_name
        Set the cluster name. Defaults to "ceph".

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.
    """
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    try:
        u.defaults_refresh()
    except:
        return {}
    u.load_confg(m.cluster_name)
    u.mon_members_refresh()
    u.auth_list()
    p = presenter.mdl_presentor(m)
    return p.auth_list()

def lnlikeHFM(pars, samples, obs, u, extra=False):
    '''
    Generic likelihood function for importance sampling with any number of
    dimensions.
    Now with added jitter parameter (hierarchical)
    obs should be a 2d array of observations. shape = (ndims, nobs)
    u should be a 2d array of uncertainties. shape = (ndims, nobs)
    samples is a 3d array of samples. shape = (ndims, nobs, nsamp)
    if extra == True, the sigma has both a slope and intercept
    Now with a mixture model!
    '''
    ndims, nobs, nsamp = samples.shape
    ypred = model(pars, samples)
    yobs = obs[1, :]
    xobs = obs[0, :]
    yerr = u[1, :]
    ll1 = np.zeros((nobs, nsamp*nobs))
    ll2 = np.zeros((nobs, nsamp*nobs))
    Y, V, P = pars[3], pars[4], pars[5]
    for i in range(nobs):
        if extra:
            inv_sigma2 = 1.0/(yerr[i]**2 + \
                    (pars[2] + pars[3] * model1(pars, xobs[i]))**2 + V)
        else:
            inv_sigma2 = 1.0/(yerr[i]**2 + \
                    (pars[2]*model1(pars, xobs[i]))**2 + V)
        ll1[i, :] = -.5*((yobs[i] - ypred)**2*inv_sigma2) + np.log(inv_sigma2)
        ll2[i, :] = -.5*((yobs[i] - Y)**2*inv_sigma2) + np.log(inv_sigma2)
    lnlike1 = np.logaddexp.reduce(ll1, axis=1)
    lnlike2 = np.logaddexp.reduce(ll2, axis=1)
    loglike = np.sum(np.logaddexp(np.log(1-P) + lnlike1, np.log(P) + lnlike2))
    if np.isfinite(loglike):
        return loglike
    return -np.inf

    def __raxml_to_model(self, model_name, param_names,args,s_size):
        raxmlmodelstr=self.__model_to_raxmodelstr(model_name,param_names)
        ##think about changing this so it catches crashes and continues
        rax_cmd=Popen('{raxpath} -f e -m {modelstr} -s {phy_file} -t {starting_tree} -n {output_names} -w {out_path}/{model_name}'.format(
                raxpath=self.raxml_path,
                modelstr=raxmlmodelstr,
                output_names=param_names,
                out_path=self.output,
                model_name=model_name,
                phy_file=self.ifile_name,
                starting_tree=self.stree),
                      
            shell=True,
            stdout=PIPE,
            stderr=PIPE)
				                                      
        rax_out = rax_cmd.communicate()
        full_name=model_name+param_names
        self.raxstdout+='='*80+'\n'+full_name.center(80)+'\n'+'='*80+'\n'
        self.raxstdout+=rax_out[0]

        self.raxstderr+='='*80+'\n'+full_name.center(80)+'\n'+'='*80+'\n'
        self.raxstderr+=rax_out[0]
        
        new_model=model(model_name=model_name,
                        param_names=param_names,
                        rax_name=raxmlmodelstr,
                        args=self.args,
                        s_size=s_size
                        )
        
        return new_model

def main(hps):

    # Initialize Horovod.
    hvd.init()

    # Create tensorflow session
    sess = tensorflow_session()

    # Download and load dataset.
    tf.set_random_seed(hvd.rank() + hvd.size() * hps.seed)
    np.random.seed(hvd.rank() + hvd.size() * hps.seed)

    # Get data and set train_its and valid_its
    train_iterator, test_iterator, data_init = get_data(hps, sess)
    hps.train_its, hps.test_its, hps.full_test_its = get_its(hps)

    # Create log dir
    logdir = os.path.abspath(hps.logdir) + "/"
    if not os.path.exists(logdir):
        os.mkdir(logdir)

    # Create model
    import model
    model = model.model(sess, hps, train_iterator, test_iterator, data_init)

    # Initialize visualization functions
    visualise = init_visualizations(hps, model, logdir)

    if not hps.inference:
        # Perform training
        train(sess, model, hps, logdir, visualise)
    else:
        infer(sess, model, hps, test_iterator)

    def __init__(self):

        self.M = model()
        self.V = view()

        self.M.setPaintCallback(self.Paint)
        self.M.setMessageCallback(self.systemMessage)

        self.mkLoad()
        self.mkSelect()
        self.mkFunction()
        self.mkDatabus()
        self.mkAddressbus()
        self.mkMemory()
        self.mkPGM()
        self.mkProgramControls()
        self.mkHALT()

        self.Paint()

        m = "\n\n                    WELCOME TO VIRTUAL MACHINE\n"
        m = m + "                       Memory: " + str(len(self.M.MEMORY.values())) + " bytes\n"
        m = m + "  ==========================================================\n"
        m = m + "Press README on left...."


        self.systemMessage(m)

        self.V.mainloop()

def cluster_quorum(**kwargs):
    """
    Get the cluster status

    CLI Example:

        salt '*' sesceph.cluster_status \\
                'cluster_name'='ceph' \\
                'cluster_uuid'='cluster_uuid'
    Notes:
    Get the cluster quorum status.

    Scope:
    Cluster wide

    Arguments:

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.

    cluster_name
        Set the cluster name. Defaults to "ceph".
    """
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    u.defaults_refresh()
    u.load_confg(m.cluster_name)
    u.mon_members_refresh()
    mur = mdl_updater_remote.model_updater_remote(m)
    can_connect = mur.connect()
    if not can_connect:
        return False
    q = mdl_query.mdl_query(m)
    return q.cluster_quorum()

def main(weights_path, base_path, base_file, style_path, style_file,
         combo_path, img_width, img_height, iterations):
    result_prefix = base_file[:-4] + '_' + style_file[:-4]
    base_img_path = base_path + base_file
    style_img_path = style_path + style_file
    # get tensor representations of images
    base_img = K.variable(preprocess_image(base_img_path,
                                           img_width,
                                           img_height))
    style_img = K.variable(preprocess_image(style_img_path,
                                            img_width,
                                            img_height))
    combo_img = K.placeholder((1, 3, img_width, img_height))
    # combine the 3 images into a single Keras tensor
    input_tensor = K.concatenate([base_img, style_img, combo_img],
                                 axis=0)

    print('Creating painting of {} in the style of {}'.format(base_file[:-4],
                                                              style_file[:-4]))
    print('Loading model with VGG16 network weights...')
    nn = model(weights_path, input_tensor, img_width, img_height)
    loss, grads = calc_loss_grad(nn, combo_img, img_width, img_height)
    evaluate = Evaluator(loss, grads, combo_img, img_width, img_height)
    return optimizer(evaluate, img_width, img_height, combo_path,
                     result_prefix, iterations=iterations)

def main():
    sample_size = int(sys.argv[1])
    train = data_io.read_train()
    print("Data Size:")
    print(train.shape)
    feature_eng(train)
    ## originally sample size = 100000
    train_sample = train[:sample_size]

    ## Train the booking model
    for i in range(0,2):
        if i==0:
            model_name = "Booking"
            response_name = "booking_bool"
            isBook = True
        else:
            model_name = "Click"
            response_name = "click_bool"
            isBook = False
        print("Training the "+model_name+" Classifier...")
        tstart = datetime.now()
        feature_names = get_features(train_sample, isBook)
        print("Using "+str(len(feature_names))+" features...")
        features = train_sample[feature_names].values
        target = train_sample[response_name].values
        classifier = model.model()
        classifier.fit(features, target)
        # print the time interval
        print("Time used,")
        print datetime.now() - tstart
        print("Saving the classifier...")
        tstart = datetime.now()
        data_io.save_model(classifier, isBook)
        print("Time used,")
        print datetime.now() - tstart

 def __init__(self, **kwargs):
     self.model = model.model(**kwargs)
     self._clear_implementation()
     u = mdl_updater.model_updater(self.model)
     u.ceph_version_refresh()
     q = mdl_query.mdl_query(self.model)
     self.ceph_daemon_user = q.ceph_daemon_user()

def evaluate(image_path, label_path):
	""" Loads network, reads image and returns IOU."""
	# Load image and label
	image = load_image(image_path)
	label = scipy.misc.imread(label_path)
	
	# Define the model
	prediction = model.model(image, drop=False)
	
	# Get a saver
	saver = tf.train.Saver()

	# Launch the graph
	with tf.Session() as sess:
		# Restore variables
		checkpoint_path = tf.train.latest_checkpoint(checkpoint_dir)
		saver.restore(sess, checkpoint_path)
		model.log("Variables restored from:", checkpoint_path)
	
		logits = prediction.eval()
		
		segmentation = post(logits, label, threshold = -1)
		
		iou = IOU(segmentation, label)
		
		print("iou =", iou)
		
	return iou

def pool_del(pool_name, **kwargs):
    """
    List all cephx authorization keys

    CLI Example:

        salt '*' sesceph.pool_del pool_name \
                'cluster_name'='ceph' \
                'cluster_uuid'='cluster_uuid' \
    Notes:

    cluster_name
        Set the cluster name. Defaults to "ceph".

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.
    """
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    try:
        u.defaults_refresh()
    except:
        return {}
    u.load_confg(m.cluster_name)
    u.mon_members_refresh()
    u.pool_list()
    u.pool_del(pool_name)
    return True

def setup():
    import model
    models =[] 
    tmpM = model.model()
    # funcs: (nextCond, prevCond, handleTag, handleData)
    tmpM.push(
        lambda *parg: parg[0] == "div" and ('class', 'threadpost') in parg[1], None, None, None)
    
    tmpM.push(
        lambda *parg: parg[0] == "img" and 1 or ( (parg[0]=="div" and ('class','quote') in parg[1]) and 2),
        lambda *parg: parg[0] == "hr",
        lambda *parg: [v for k,v in parg[1] if k == "id" and ( ('class', 'threadpost') in parg[1] or ('class', 'reply') in parg[1])], 
        None)
    
    tmpM.push(
        None,
        lambda *parg: parg[0] == "img",
        lambda *parg: apply(MyParser.getPicIdFromTURL, [v for k, v in parg[1] if k=="src"]),
        None)
    
    tmpM.push(
        None,
        lambda *parg: parg[0] == "div" and 2,
        None,
        lambda *parg: parg)
    
    models.append(tmpM)
    
    return models

def cluster_status(**kwargs):
    """
    Get the cluster status

    CLI Example:

        salt '*' sesceph.cluster_status \\
                'cluster_name'='ceph' \\
                'cluster_uuid'='cluster_uuid'
    Notes:
    Get the cluster status including health if in quorum.

    Scope:
    Cluster wide

    Arguments:

    cluster_uuid
        Set the cluster UUID. Defaults to value found in ceph config file.

    cluster_name
        Set the cluster name. Defaults to "ceph".
    """
    m = model.model(**kwargs)
    u = mdl_updater.model_updater(m)
    u.hostname_refresh()
    u.defaults_refresh()
    u.load_confg(m.cluster_name)
    u.mon_members_refresh()
    mur = mdl_updater_remote.model_updater_remote(m)
    can_connect = mur.connect()
    if not can_connect:
        raise Error("Cant connect to cluster.")
    p = presenter.mdl_presentor(m)
    return p.cluster_status()

    def __init__(self, config):
        self.config = config
        self.model = model(config)
        self.criterion = nn.PairwiseDistance()
        self.optimizer = optim.Adam(self.model.parameters(), lr=self.config['lr'])

        self.generate_dataset()

 def __init__(self, tf_ckpt_path):
     #init the tensorflow model
     self.x, _ = model.place_holders()
     self.y_conv = model.model(self.x)
     self.sess = tf.InteractiveSession()
     self.sess.run(tf.initialize_all_variables())
     saver = tf.train.Saver(tf.all_variables())
     saver.restore(self.sess, tf_ckpt_path)