Related
I am having trouble with random_state in scipy.stats.truncnorm. Here is my code:
from scipy.stats import truncnorm
from numpy.random import default_rng
rg = default_rng( 12345 )
truncnorm.rvs(0.0,1.0,size=10, random_state=rg)
I get the following error:
File "test2.py", line 4, in <module>
truncnorm.rvs(0.0,1.0,size=10, random_state=rg)
File "/opt/anaconda3/envs/newbase/lib/python3.8/site-packages/scipy/stats/_distn_infrastructure.py", line 1004, in rvs
vals = self._rvs(*args, size=size, random_state=random_state)
File "/opt/anaconda3/envs/newbase/lib/python3.8/site-packages/scipy/stats/_continuous_distns.py", line 7641, in _rvs
out = self._rvs_scalar(a.item(), b.item(), size, random_state=random_state)
File "/opt/anaconda3/envs/newbase/lib/python3.8/site-packages/scipy/stats/_continuous_distns.py", line 7697, in _rvs_scalar
U = random_state.random_sample(N)
AttributeError: 'numpy.random._generator.Generator' object has no attribute 'random_sample'
I am using numpy 1.19.1 and scipy 1.5.0. The problem does not occur with scipy.norm.rvs.
In scipy 1.7.1, the problem line has been changed to:
def _rvs_scalar(self, a, b, numsamples=None, random_state=None):
if not numsamples:
numsamples = 1
# prepare sampling of rvs
size1d = tuple(np.atleast_1d(numsamples))
N = np.prod(size1d) # number of rvs needed, reshape upon return
# Calculate some rvs
U = random_state.uniform(low=0, high=1, size=N)
x = self._ppf(U, a, b)
rvs = np.reshape(x, size1d)
return rvs
Both have uniform, but rg does not have random_sample:
In [221]: rg.uniform
Out[221]: <function Generator.uniform>
In [222]: np.random.uniform
Out[222]: <function RandomState.uniform>
np.random.random_sample has this note:
.. note::
New code should use the ``random`` method of a ``default_rng()``
instance instead; please see the :ref:`random-quick-start`.
Hi I just wanted to use my Module again which I defined in Python.
Here is the Code what I saved in Python.
import numpy as np
matrixSize = 10 Qa= np.random.rand(matrixSize, matrixSize) Q_1 = np.dot(Qa, Qa.transpose()) print(Q_1) np.linalg.cholesky(Q_1) ##check if a matrix is positive definite p_1= np.random.uniform(0, 9, size=10) G_1 = np.diag(-1*np.random.uniform(0, 9, size=10)) h_1 = np.random.uniform(-9, 9, size=10)
np.savez_compressed(r'C:\Users\skqkr\Desktop\Semesterarbeit/Chiwan_Q1', Q=Q_1, p=p_1, G=G_1, h=h_1)
and I want use this Code in Jupyter with Juliaprogramm so I typed like this.
using PyCall
#pyimport numpy
using OSQP
using Compat.SparseArrays
using NPZ
# Define problem data
Matrix10 = npzread("C:/Users/skqkr/Desktop/Semesterarbeit/Chiwan_Q1.npz")
P = sparse(Matrix10['Q'])
q = Matrix10['p']
A = sparse(Matrix10['G']))
u = [Matrix10['h'] ]
# Crate OSQP object
prob = OSQP.Model()
# Setup workspace and change alpha parameter
OSQP.setup!(prob; P=P, q=q, A=A, l=l, u=u, alpha=1)
# Solve problem
results = OSQP.solve!(prob)
But there is an Error
KeyError: key 'Q' not found
So how can I fix it? because it works very well when I use this Module in Pythonprogramm
Thanks!
System information
Have I written custom code (as opposed to using a stock example script provided in TensorFlow): Y
OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Ubuntu 16.04
TensorFlow installed from (source or binary): binary
TensorFlow version (use command below): r1.8
Python version: 2.7.14
GCC/Compiler version (if compiling from source): 5.4
CUDA/cuDNN version: 8.0/7.0
GPU model and memory: GTX1080, 8G
Bazel version: N/A
Exact command to reproduce: python test_script.py
Describe the problem
Hello, I'm trying to make a custom_gradient op using the function of tf.custom_gradient. I made my test code based on the API explanation online. However, it seems there is a problem in the custom_gradient function. Thanks!
Source code / logs
import tensorflow as tf
import numpy as np
#tf.custom_gradient
def log1pexp(x):
e = tf.exp(x)
def grad(dy):
return dy * (1 - 1 / (1 + e))
return tf.log(1 + e), grad
x = tf.constant(100.)
f = tf.custom_gradient(log1pexp)
y, dy = f(x)
sess = tf.Session()
print (y.eval(session=sess), y.eval(session=sess).shape)
File "/home/local/home/research/DL/unit_tests/tf_test_custom_grad.py", line 14, in <module>
y, dy = f(x)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/custom_gradient.py", line 111, in decorated
return _graph_mode_decorator(f, *args, **kwargs)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/custom_gradient.py", line 132, in _graph_mode_decorator
result, grad_fn = f(*args)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 439, in __iter__
"Tensor objects are not iterable when eager execution is not "
TypeError: Tensor objects are not iterable when eager execution is not enabled. To iterate over this tensor use tf.map_fn.
If you just want to test the code in the documentation, here is the way.
The following code will give the instable [nan] result:
import tensorflow as tf
def log1pexp(x):
return tf.log(1 + tf.exp(x))
x = tf.constant(100.)
y = log1pexp(x)
dy = tf.gradients(y, x)
with tf.Session() as sess:
print(sess.run(dy))
And the following code will give the correct result [1.0]:
import tensorflow as tf
#tf.custom_gradient
def log1pexp(x):
e = tf.exp(x)
def grad(dy):
return dy * (1 - 1 / (1 + e))
return tf.log(1 + e), grad
x = tf.constant(100.)
y = log1pexp(x)
dy = tf.gradients(y, x)
with tf.Session() as sess:
print(sess.run(dy))
Details:
The main problem here is that you are trying to decorate log1pexp twice in your code: once with #tf.custom_gradient and once with f = tf.custom_gradient(log1pexp). In python, #tf.custom_gradient here is equivalent to log1pexp = tf.custom_gradient(log1pexp). You should do this only once, especially here for the following reason.
tf.custom_gradient needs to call the function being pass to it to get both the function output and the gradient, i.e. expecting two returns. During decoration, everything works as expected because log1pexp returns tf.log(1 + e) and grad. After decorating log1pexp, log1pexp (returned by tf.custom_gradient) becomes a new function which returns only one tensor tf.log(1 + e). When you do f = tf.custom_gradient(log1pexp) after decorating log1pexp, tf.custom_gradient can only get one return which is the single tensor tf.log(1 + e). It will try to split this tensor into two by iterating this returned tensor. But it is wrong and is not allowed as the error message stated:
Tensor objects are not iterable when eager execution is not enabled.
You should not decorate log1pexp twice anyway. But this is why you got this error. One more thing to mention, your code will trigger another error for the same reason even if you removed #tf.custom_gradient. After removing #tf.custom_gradient, the line f = tf.custom_gradient(log1pexp) should work as expected. But f is a function returning only one tensor. y, dy = f(x) is wrong and will not work.
Below is some sample code straight from the MayaVI website on using sliders. Try putting in a number outside of the slider range for a fatal error:
from numpy import arange, pi, cos, sin
from traits.api import HasTraits, Range, Instance, \
on_trait_change
from traitsui.api import View, Item, Group
from mayavi.core.api import PipelineBase
from mayavi.core.ui.api import MayaviScene, SceneEditor, \
MlabSceneModel
dphi = pi/1000.
phi = arange(0.0, 2*pi + 0.5*dphi, dphi, 'd')
def curve(n_mer, n_long):
mu = phi*n_mer
x = cos(mu) * (1 + cos(n_long * mu/n_mer)*0.5)
y = sin(mu) * (1 + cos(n_long * mu/n_mer)*0.5)
z = 0.5 * sin(n_long*mu/n_mer)
t = sin(mu)
return x, y, z, t
class MyModel(HasTraits):
n_meridional = Range(0, 30, 6, )#mode='spinner')
n_longitudinal = Range(0, 30, 11, )#mode='spinner')
scene = Instance(MlabSceneModel, ())
plot = Instance(PipelineBase)
# When the scene is activated, or when the parameters are changed, we
# update the plot.
#on_trait_change('n_meridional,n_longitudinal,scene.activated')
def update_plot(self):
x, y, z, t = curve(self.n_meridional, self.n_longitudinal)
if self.plot is None:
self.plot = self.scene.mlab.plot3d(x, y, z, t,
tube_radius=0.025, colormap='Spectral')
else:
self.plot.mlab_source.set(x=x, y=y, z=z, scalars=t)
# The layout of the dialog created
view = View(Item('scene', editor=SceneEditor(scene_class=MayaviScene),
height=250, width=300, show_label=False),
Group(
'_', 'n_meridional', 'n_longitudinal',
),
resizable=True,
)
my_model = MyModel()
my_model.configure_traits()
How can I improve this code to disallow users from triggering this fatal error? I think a line that could deactivate the entry box (such as setDisabled(True)) could work, or remove it entirely - but I'm not sure how to implement it within the traitsui methods.
After lots of trial and error, this appears to be a bug in the default Range() mode of Traitsui, at least in the case for Mac OS X (I'm running High Sierra, 10.13.3).
The solution is to alter the default mode to one that looks and acts identical, minus crashing the program:
n_meridional = Range(0, 30, 6, mode='slider')
Recently I want to try some open source solvers instead of CPLEX. I found that PICOS + zibopt may be a good choice. However, I can merely find instruction on how to make zibopt work with python under windows properly. I downloaded the windows libraries (.dll file) of scip, and I try to install python-zibopt according to the command "python setup.py install". The error " blockmemshell/memory.h no such file" always popped out. I felt that it is because my compiler, which is VS120COMNTOOL, doecn't find the scip solver. Is there any chance that I can make scip work under windows now?
Did you have a look at the current python interface of SCIP 3.1.0? It uses the library from the SCIP Optimization Suite so you don't have to link another LP solver to SCIP.
On Windows, please try this modified setup.py file:
import sys, os, readline, glob, platform
from distutils.core import setup
from distutils.extension import Extension
from Cython.Distutils import build_ext
from Cython.Build import cythonize
BASEDIR = os.path.dirname(os.path.abspath(__file__))
BASEDIR = os.path.dirname(BASEDIR)
BASEDIR = os.path.dirname(BASEDIR)
INCLUDEDIR = os.path.join(BASEDIR,'src')
BASEDIR = os.path.dirname(BASEDIR)
#identify compiler version
prefix = "MSC v."
i = sys.version.find(prefix)
if i == -1:
raise Exception('cannot determine compiler version')
i = i + len(prefix)
s, rest = sys.version[i:].split(" ", 1)
majorVersion = int(s[:-2]) - 6
minorVersion = int(s[2:3]) / 10.0
if platform.architecture()[0].find('64')>=0:
LIBDIR = os.path.join(BASEDIR,'vc'+str(majorVersion),'scip_spx','x64','Release')
else:
LIBDIR = os.path.join(BASEDIR,'vc'+str(majorVersion),'scip_spx','Release')
print('BASEDIR='+ BASEDIR)
print('INCLUDEDIR='+ INCLUDEDIR)
print('LIBDIR='+ LIBDIR)
def complete(text, state):
return (glob.glob(text+'*')+[None])[state]
readline.set_completer_delims(' \t\n;')
readline.parse_and_bind("tab: complete")
readline.set_completer(complete)
libscipopt = 'lib/libscipopt.so'
includescip = 'include/scip'
ext_modules = []
ext_modules += [Extension('pyscipopt.scip', [os.path.join('pyscipopt', 'scip.pyx')],
#extra_compile_args=['-g', '-O0', '-UNDEBUG'],
include_dirs=[INCLUDEDIR],
library_dirs=[LIBDIR],
#runtime_library_dirs=[os.path.abspath('lib')],
libraries=['spx', 'scip_spx'])]
#libraries=['scipopt', 'readline', 'z', 'gmp', 'ncurses', 'm'])]
setup(
name = 'pyscipopt',
version = '0.1',
description = 'wrapper for SCIP in Python',
author = 'Zuse Institute Berlin',
author_email = 'scip#zib.de',
license = 'MIT',
cmdclass = {'build_ext' : build_ext},
ext_modules = ext_modules,
packages=['pyscipopt']
)