Develop Reference

Optimization Problem with fast matrix-vector multiplication in Python / cvxpy

I want to solve the following (convex) minimization problem:
min ||x||_1 under the constraints sgn(A[x,R]=y) and ||x||_2 = 1
where A is a mx(N+1) matrix, x in R^N a vector, and \[x,R\] a vector that is created by appending a given number R. The objective is to find the optimal value for x.
A is a Fourier matrix and there are fast matrix-vector, inversion, etc. algorithms available. Since this matrix is really big, I need to use an optimization algorithm that utilizes this.
Currently, I use the following implementation in cvxpy, which is way too slow:
import cvxpy as cvx
# rewrite the problem in the form x = x^- + x^+
n = A.shape[1]-1
vx = cvx.Variable(2*n)
objective = cvx.Minimize(cvx.pnorm(vx, 1)) # min ||x||_1
constraints = [vx >= 0, cvx.multiply(A[:,:n] # vx[:n] - A[:,:n] # vx[n:] + A[:,n]*R, y) >= 0,
cvx.norm(vx, 2) <= R] # sgn(A[x,1]) = y, ||x||_2 <= R
x, solve_time = solve(vx, objective, constraints)
solution = x[:n] - x[n:]
Is there a way to use fast matrix computations in cvxpy? Or is there a better library? I found a few implementations that can do this for one special algorithm but not in the general case, so I was not able to implement my problem.

No. The solver will not call your matrix multiplication code. They do their own linear algebra, which is very different in many ways. In a sense your matrix multiplication is just notation for the problem statement.
Regarding performance, it depends heavily on where the bottleneck is. Is it in generating the model (in cvxpy itself) or in the solver? What solver are you using? Consider using a different solver. Obviously, we don't have enough information (and no reproducible example) to answer this question.

Double antiderivative computation in python

I have the following problem. I have a function f defined in python using numpy functions. The function is smooth and integrable on positive reals. I want to construct the double antiderivative of the function (assuming that both the value and the slope of the antiderivative at 0 are 0) so that I can evaluate it on any positive real smaller than 100.
Definition of antiderivative of f at x:
integrate f(s) with s from 0 to x
Definition of double antiderivative of f at x:
integrate (integrate f(t) with t from 0 to s) with s from 0 to x
The actual form of f is not important, so I will use a simple one for convenience. But please note that even though my example has a known closed form, my actual function does not.
import numpy as np
f = lambda x: np.exp(-x)*x
My solution is to construct the antiderivative as an array using naive numerical integration:
N = 10000
delta = 100/N
xs = np.linspace(0,100,N+1)
vs = f(xs)
avs = np.cumsum(vs)*delta
aavs = np.cumsum(avs)*delta
This of course works but it gives me arrays instead of functions. But this is not a big problem as I can interpolate aavs using a spline to get a function and get rid of the arrays.
from scipy.interpolate import UnivariateSpline
aaf = UnivariateSpline(xs, aavs)
The function aaf is approximately the double antiderivative of f.
The problem is that even though it works, there is quite a bit of overhead before I can get my function and precision is expensive.
My other idea was to interpolate f by a spline and take the antiderivative of that, however this introduces numerical errors that are too big for what I want to use the function.
Is there any better way to do that? By better I mean faster without sacrificing accuracy.
Edit: What I hope is possible is to use some kind of Fourier transform to avoid integrating twice. I hope that there is some convenient transform of vs that allows to multiply the values component-wise with xs and transform back to get the double antiderivative. I played with this a bit, but I got lost.
Edit: I figured out that by using the trapezoidal rule instead of a naive sum, increases the accuracy quite a bit. Using Simpson's rule should increase the accuracy further, but it's somewhat fiddly to do with numpy arrays.
Edit: As #user202729 rightfully complains, this seems off. The reason it seems off is because I have skipped some details. I explain here why what I say makes sense, but it does not affect my question.
My actual goal is not to find the double antiderivative of f, but to find a transformation of this. I have skipped that because I think it only confuses the matter.
The function f decays exponentially as x approaches 0 or infinity. I am minimizing the numerical error in the integration by starting the sum from 0 and going up to approximately the peak of f. This ensure that the relative error is approximately constant. Then I start from the opposite direction from some very big x and go back to the peak. Then I do the same for the antiderivative values.
Then I transform the aavs by another function which is sensitive to numerical errors. Then I find the region where the errors are big (the values oscillate violently) and drop these values. Finally I approximate what I believe are good values by a spline.
Now if I use spline to approximate f, it introduces an absolute error which is the dominant term in a rather large interval. This gets "integrated" twice and it ends up being a rather large relative error in aavs. Then once I transform aavs, I find that the 'good region' has shrunk considerably.
EDIT: The actual form of f is something I'm still looking into. However, it is going to be a generalisation of the lognormal distribution. Right now I am playing with the following family.
I start by defining a generalization of the normal distribution:
def pdf_n(params, center=0.0, slope=8):
scale, min, diff = params
if diff > 0:
r = min
l = min + diff
else:
r = min - diff
l = min
def retfun(m):
x = (m - center)/scale
E = special.expit(slope*x)*(r - l) + l
return np.exp( -np.power(1 + x*x, E)/2 )
return np.vectorize(retfun)
It may not be obvious what is happening here, but the result is quite simple. The function decays as exp(-x^(2l)) on the left and as exp(-x^(2r)) on the right. For min=1 and diff=0, this is the normal distribution. Note that this is not normalized. Then I define
g = pdf(params)
f = np.vectorize(lambda x:g(np.log(x))/x/area)
where area is the normalization constant.
Note that this is not the actual code I use. I stripped it down to the bare minimum.

You can compute the two np.cumsum (and the divisions) at once more efficiently using Numba. This is significantly faster since there is no need for several temporary arrays to be allocated, filled, read again and freed. Here is a naive implementation:
import numba as nb
#nb.njit('float64[::1](float64[::1], float64)') # Assume vs is contiguous
def doubleAntiderivative_naive(vs, delta):
res = np.empty(vs.size, dtype=np.float64)
sum1, sum2 = 0.0, 0.0
for i in range(vs.size):
sum1 += vs[i] * delta
sum2 += sum1 * delta
res[i] = sum2
return res
However, the sum is not very good in term of numerical stability. A Kahan summation is needed to improve the accuracy (or possibly the alternative Kahan–Babuška-Klein algorithm if you are paranoid about the accuracy and performance do not matter so much). Note that Numpy use a pair-wise algorithm which is quite good but far from being prefect in term of accuracy (this is a good compromise for both performance and accuracy).
Moreover, delta can be factorized during in the summation (ie. the result just need to be premultiplied by delta**2).
Here is an implementation using the more accurate Kahan summation:
#nb.njit('float64[::1](float64[::1], float64)')
def doubleAntiderivative_accurate(vs, delta):
res = np.empty(vs.size, dtype=np.float64)
delta2 = delta * delta
sum1, sum2 = 0.0, 0.0
c1, c2 = 0.0, 0.0
for i in range(vs.size):
# Kahan summation of the antiderivative of vs
y1 = vs[i] - c1
t1 = sum1 + y1
c1 = (t1 - sum1) - y1
sum1 = t1
# Kahan summation of the double antiderivative of vs
y2 = sum1 - c2
t2 = sum2 + y2
c2 = (t2 - sum2) - y2
sum2 = t2
res[i] = sum2 * delta2
return res
Here is the performance of the approaches on my machine (with an i5-9600KF processor):
Numpy cumsum: 51.3 us
Naive Numba: 11.6 us
Accutate Numba: 37.2 us
Here is the relative error of the approaches (based on the provided input function):
Numpy cumsum: 1e-13
Naive Numba: 5e-14
Accutate Numba: 2e-16
Perfect precision: 1e-16 (assuming 64-bit numbers are used)
If f can be easily computed using Numba (this is the case here), then vs[i] can be replaced by calls to f (inlined by Numba). This helps to reduce the memory consumption of the computation (N can be huge without saturating your RAM).
As for the interpolation, the splines often gives good numerical result but they are quite expensive to compute and AFAIK they require the whole array to be computed (each item of the array impact all the spline although some items may have a negligible impact alone). Regarding your needs, you could consider using Lagrange polynomials. You should be careful when using Lagrange polynomials on the edges. In your case, you can easily solve the numerical divergence issue on the edges by extending the array size with the border values (since you know the derivative on each edges of vs is 0). You can apply the interpolation on the fly with this method which can be good for both performance (typically if the computation is parallelized) and memory usage.

First, I created a version of the code I found more intuitive. Here I multiply cumulative sum values by bin widths. I believe there is a small error in the original version of the code related to the bin width issue.
import numpy as np
f = lambda x: np.exp(-x)*x
N = 1000
xs = np.linspace(0,100,N+1)
domainwidth = ( np.max(xs) - np.min(xs) )
binwidth = domainwidth / N
vs = f(xs)
avs = np.cumsum(vs)*binwidth
aavs = np.cumsum(avs)*binwidth
Next, for visualization here is some very simple plotting code:
import matplotlib
import matplotlib.pyplot as plt
plt.figure()
plt.scatter( xs, vs )
plt.figure()
plt.scatter( xs, avs )
plt.figure()
plt.scatter( xs, aavs )
plt.show()
The first integral matches the known result of the example expression and can be seen on wolfram
Below is a simple function that extracts an element from the second derivative. Note that int is a bad rounding function. I assume this is what you have implemented already.
def extract_double_antideriv_value(x):
return aavs[int(x/binwidth)]
singleresult = extract_double_antideriv_value(50.24)
print('singleresult', singleresult)
Whatever full computation steps are required, we need to know them before we can start optimizing. Do you have a million different functions to integrate? If you only need to query a single double anti-derivative many times, your original solution should be fairly ideal.
Symbolic Approximation:
Have you considered approximations to the original function f, which can have closed form integration solutions? You have a limited domain on which the function lives. Perhaps approximate f with a Taylor series (which can be constructed with known maximum error) then integrate exactly? (consider Pade, Taylor, Fourier, Cheby, Lagrange(as suggested by another answer), etc...)
Log Tricks:
Another alternative to dealing with spiky errors, would be to take the log of your original function. Is f always positive? Is the integration error caused because the neighborhood around the max is very small? If so, you can study ln(f) or even ln(ln(f)) instead. It would really help to understand what f looks like more.
Approximation Integration Tricks
There exist countless integration tricks in general, which can make approximate closed form solutions to undo-able integrals. A very common one when exponetnial functions are involved (I think yours is expoential?) is to use Laplace's Method. But which trick to pull out of the bag is highly dependent upon the conditions which f satisfies.

Faster Multigrid Poission Equation Solver?

I am trying to make my own CFD solver and one of the most computationally expensive parts is solving for the pressure term. One way to solve Poisson differential equations faster is by using a multigrid method. The basic recursive algorithm for this is:
function phi = V_Cycle(phi,f,h)
% Recursive V-Cycle Multigrid for solving the Poisson equation (\nabla^2 phi = f) on a uniform grid of spacing h
% Pre-Smoothing
phi = smoothing(phi,f,h);
% Compute Residual Errors
r = residual(phi,f,h);
% Restriction
rhs = restriction(r);
eps = zeros(size(rhs));
% stop recursion at smallest grid size, otherwise continue recursion
if smallest_grid_size_is_achieved
eps = smoothing(eps,rhs,2*h);
else
eps = V_Cycle(eps,rhs,2*h);
end
% Prolongation and Correction
phi = phi + prolongation(eps);
% Post-Smoothing
phi = smoothing(phi,f,h);
end
I've attempted to implement this algorithm myself however it is very slow and doesn't give good results so evidently it is doing something wrong. I've been trying to find why for too long and I think it's just worthwhile seeing if anyone can help me.
If I use a grid size of 2^5 by 2^5 points, then it can solve it and give reasonable results. However, as soon as I go above this it takes exponentially longer to solve and basically get stuck at some level of inaccuracy, no matter how many V-Loops are performed. at 2^7 by 2^7 points, the code takes way too long to be useful.
I think my main issue is that my implementation of a jacobian iteration is using linear algebra to calculate the update at each step. This should, in general, be fast however, the update matrix A is an n*m sized matrix, and calculating the dot product of a 2^7 * 2^7 sized matrix is expensive. As most of the cells are just zeros, should I calculate the result using a different method?
if anyone has any experience in multigrid methods, I would appreciate any advice!
Thanks

FFT polynomial multiplication in Python using inbuilt Numpy.fft

I want to multiply two polynomials fast in python. As my polynomials are rather large (> 100000) elements and I have to multiply lots of them. Below, you will find my approach,
from numpy.random import seed, randint
from numpy import polymul, pad
from numpy.fft import fft, ifft
from timeit import default_timer as timer
length=100
def test_mul(arr_a,arr_b): #inbuilt python multiplication
c=polymul(arr_a,arr_b)
return c
def sb_mul(arr_a,arr_b): #my schoolbook multiplication
c=[0]*(len(arr_a) + len(arr_b) - 1 )
for i in range( len(arr_a) ):
for j in range( len(arr_b) ):
k=i+j
c[k]=c[k]+arr_a[i]*arr_b[j]
return c
def fft_test(arr_a,arr_b): #fft based polynomial multuplication
arr_a1=pad(arr_a,(0,length),'constant')
arr_b1=pad(arr_b,(0,length),'constant')
a_f=fft(arr_a1)
b_f=fft(arr_b1)
c_f=[0]*(2*length)
for i in range( len(a_f) ):
c_f[i]=a_f[i]*b_f[i]
return c_f
if __name__ == '__main__':
seed(int(timer()))
random=1
if(random==1):
x=randint(1,1000,length)
y=randint(1,1000,length)
else:
x=[1]*length
y=[1]*length
start=timer()
res=test_mul(x,y)
end=timer()
print("time for built in pol_mul", end-start)
start=timer()
res1=sb_mul(x,y)
end=timer()
print("time for schoolbook mult", end-start)
res2=fft_test(x,y)
print(res2)
#########check############
if( len(res)!=len(res1) ):
print("ERROR");
for i in range( len(res) ):
if( res[i]!=res1[i] ):
print("ERROR at pos ",i,"res[i]:",res[i],"res1[i]:",res1[i])
Now, here are my approach in detail,
1. First, I tried myself with a naive implementation of Schoolbook with complexity O(n^2). But as you may expect it turned out to be very slow.
Second, I came to know polymul in the Numpy library. This function is a lot faster than the previous one. But I realized this is also a O(n^2) complexity. You can see, if you increase the length k the time increases by k^2 times.
My third approach is to try a FFT based multiplication using the inbuilt FFT functions. I followed the the well known approach also described here but Iam not able to get it work.
Now my questions are,
Where am I going wrong in my FFT based approach? Can you please tell me how can I fix it?
Is my observation that polymul function has O(n^2) complexity correct?
Please, let me know if you have any question.
Thanks in advance.

Where am I going wrong in my FFT based approach? Can you please tell me how can I fix it?
The main problem is that in the FFT based approach, you should be taking the inverse transform after the multiplication, but that step is missing from your code. With this missing step your code should look like the following:
def fft_test(arr_a,arr_b): #fft based polynomial multiplication
arr_a1=pad(arr_a,(0,length),'constant')
arr_b1=pad(arr_b,(0,length),'constant')
a_f=fft(arr_a1)
b_f=fft(arr_b1)
c_f=[0]*(2*length)
for i in range( len(a_f) ):
c_f[i]=a_f[i]*b_f[i]
return ifft(c_f)
Note that there may also a few opportunities for improvements:
The zero padding can be handled directly by passing the required FFT length as the second argument (e.g. a_f = fft(arr_a, length))
The coefficient multiplication in your for loop may be directly handled by numpy.multiply.
If the polynomial coefficients are real-valued, then you can use numpy.fft.rfft and numpy.fft.irfft (instead of numpy.fft.fft and numpy.fft.ifft) for some extra performance boost.
So an implementation for real-valued inputs may look like:
from numpy.fft import rfft, irfft
def fftrealpolymul(arr_a, arr_b): #fft based real-valued polynomial multiplication
L = len(arr_a) + len(arr_b)
a_f = rfft(arr_a, L)
b_f = rfft(arr_b, L)
return irfft(a_f * b_f)
Is my observation that polymul function has O(n2) complexity correct?
That also seem to be the performance I am observing, and matches the available code in my numpy installation (version 1.15.4, and there doesn't seem any change in that part in the more recent 1.16.1 version).

High frequency noise at solving differential equation

I'm trying to simulate a simple diffusion based on Fick's 2nd law.
from pylab import *
import numpy as np
gridpoints = 128
def profile(x):
range = 2.
straggle = .1576
dose = 1
return dose/(sqrt(2*pi)*straggle)*exp(-(x-range)**2/2/straggle**2)
x = linspace(0,4,gridpoints)
nx = profile(x)
dx = x[1] - x[0] # use np.diff(x) if x is not uniform
dxdx = dx**2
figure(figsize=(12,8))
plot(x,nx)
timestep = 0.5
steps = 21
diffusion_coefficient = 0.002
for i in range(steps):
coefficients = [-1.785714e-3, 2.539683e-2, -0.2e0, 1.6e0,
-2.847222e0,
1.6e0, -0.2e0, 2.539683e-2, -1.785714e-3]
ccf = (np.convolve(nx, coefficients) / dxdx)[4:-4] # second order derivative
nx = timestep*diffusion_coefficient*ccf + nx
plot(x,nx)
for the first few time steps everything looks fine, but then I start to get high frequency noise, do to build-up from numerical errors which are amplified through the second derivative. Since it seems to be hard to increase the float precision I'm hoping that there is something else that I can do to suppress this? I already increased the number of points that are being used to construct the 2nd derivative.

I don't have the time to study your solution in detail, but it seems that you are solving the partial differential equation with a forward Euler scheme. This is pretty easy to implement, as you show, but this can become numerical instable if your timestep is too small. Your only solution is to reduce the timestep or to increase the spatial resolution.
The easiest way to explain this is for the 1-D case: assume your concentration is a function of spatial coordinate x and timestep i. If you do all the math (write down your equations, substitute the partial derivatives with finite differences, should be pretty easy), you will probably get something like this:
C(x, i+1) = [1 - 2 * k] * C(x, i) + k * [C(x - 1, i) + C(x + 1, i)]
so the concentration of a point on the next step depends on its previous value and the ones of its two neighbors. It is not too hard to see that when k = 0.5, every point gets replaced by the average of its two neighbors, so a concentration profile of [...,0,1,0,1,0,...] will become [...,1,0,1,0,1,...] on the next step. If k > 0.5, such a profile will blow up exponentially. You calculate your second order derivative with a longer convolution (I effectively use [1,-2,1]), but I guess that does not change anything for the instability problem.
I don't know about normal diffusion, but based on experience with thermal diffusion, I would guess that k scales with dt * diffusion_coeff / dx^2. You thus have to chose your timestep small enough so that your simulation does not become instable. To make the simulation stable, but still as fast as possible, chose your parameters so that k is a bit smaller than 0.5. Something similar can be derived for 2-D and 3-D cases. The easiest way to achieve this is to increase dx, since your total calculation time will scale with 1/dx^3 for a linear problem, 1/dx^4 for 2-D problems, and even 1/dx^5 for 3-D problems.
There are better methods to solve diffusion equations, I believe that Crank Nicolson is at least standard for solving heat-equations (which is also a diffusion problem). The 'problem' is that this is an implicit method, which means that you have to solve a set of equations to calculate your 'concentration' at the next timestep, which is a bit of a pain to implement. But this method is guaranteed to be numerical stable, even for big timesteps.