Related
I am trying to minimize a nonlinear function with nonlinear inequality constraints with NLopt and JuMP.
In my test code below, I am minimizing a function with a known global minima.
Local optimizers such as LD_MMA fails to find this global minima, so I am trying to use global optimizers of NLopt that allow nonlinear inequality constraintes.
However, when I check my termination status, it says “termination_status(model) = MathOptInterface.OTHER_ERROR”. I am not sure which part of my code to check for this error.
What could be the cause?
I am using JuMP since in the future I plan to use other solvers such as KNITRO as well, but should I rather use the NLopt syntax?
Below is my code:
# THIS IS A CODE TO SOLVE FOR THE TOYMODEL
# THE EQUILIBRIUM IS CHARACTERIZED BY A NONLINEAR SYSTEM OF ODEs OF INCREASING FUCTIONS B(x) and S(y)
# THE GOAL IS TO APPROXIMATE B(x) and S(y) WITH POLYNOMIALS
# FIND THE POLYNOMIAL COEFFICIENTS THAT MINIMIZE THE LEAST SQUARES OF THE EQUILIBRIUM EQUATIONS
# load packages
using Roots, NLopt, JuMP
# model primitives and other parameters
k = .5 # equal split
d = 1 # degree of polynomial
nparam = 2*d+2 # number of parameters to estimate
m = 10 # number of grids
m -= 1
vGrid = range(0,1,m) # discretize values
c1 = 0 # lower bound for B'() and S'()
c2 = 2 # lower and upper bounds for offers
c3 = 1 # lower and upper bounds for the parameters to be estimated
# objective function to be minimized
function obj(α::T...) where {T<:Real}
# split parameters
αb = α[1:d+1] # coefficients for B(x)
αs = α[d+2:end] # coefficients for S(y)
# define B(x), B'(x), S(y), and S'(y)
B(v) = sum([αb[i] * v .^ (i-1) for i in 1:d+1])
B1(v) = sum([αb[i] * (i-1) * v ^ (i-2) for i in 2:d+1])
S(v) = sum([αs[i] * v .^ (i-1) for i in 1:d+1])
S1(v) = sum([αs[i] * (i-1) * v ^ (i-2) for i in 2:d+1])
# the equilibrium is characterized by the following first order conditions
#FOCb(y) = B(k * y * S1(y) + S(y)) - S(y)
#FOCs(x) = S(- (1-k) * (1-x) * B1(x) + B(x)) - B(x)
function FOCb(y)
sy = S(y)
binv = find_zero(q -> B(q) - sy, (-c2, c2))
return k * y * S1(y) + sy - binv
end
function FOCs(x)
bx = B(x)
sinv = find_zero(q -> S(q) - bx, (-c2, c2))
return (1-k) * (1-x) * B1(x) - B(x) + sinv
end
# evaluate the FOCs at each grid point and return the sum of squares
Eb = [FOCb(y) for y in vGrid]
Es = [FOCs(x) for x in vGrid]
E = [Eb; Es]
return E' * E
end
# this is the actual global minimum
αa = [1/12, 2/3, 1/4, 2/3]
obj(αa...)
# do optimization
model = Model(NLopt.Optimizer)
set_optimizer_attribute(model, "algorithm", :GN_ISRES)
#variable(model, -c3 <= α[1:nparam] <= c3)
#NLconstraint(model, [j = 1:m], sum(α[i] * (i-1) * vGrid[j] ^ (i-2) for i in 2:d+1) >= c1) # B should be increasing
#NLconstraint(model, [j = 1:m], sum(α[d+1+i] * (i-1) * vGrid[j] ^ (i-2) for i in 2:d+1) >= c1) # S should be increasing
register(model, :obj, nparam, obj, autodiff=true)
#NLobjective(model, Min, obj(α...))
println("")
println("Initial values:")
for i in 1:nparam
set_start_value(α[i], αa[i]+rand()*.1)
println(start_value(α[i]))
end
JuMP.optimize!(model)
println("")
#show termination_status(model)
#show objective_value(model)
println("")
println("Solution:")
sol = [value(α[i]) for i in 1:nparam]
My output:
Initial values:
0.11233072522513032
0.7631843020124309
0.3331559403539963
0.7161240026812674
termination_status(model) = MathOptInterface.OTHER_ERROR
objective_value(model) = 0.19116585196576466
Solution:
4-element Vector{Float64}:
0.11233072522513032
0.7631843020124309
0.3331559403539963
0.7161240026812674
I answered on the Julia forum: https://discourse.julialang.org/t/mathoptinterface-other-error-when-trying-to-use-isres-of-nlopt-through-jump/87420/2.
Posting my answer for posterity:
You have multiple issues:
range(0,1,m) should be range(0,1; length = m) (how did this work otherwise?) This is true for Julia 1.6. The range(start, stop, length) method was added for Julia v1.8
Sometimes your objective function errors because the root doesn't exist. If I run with Ipopt, I get
ERROR: ArgumentError: The interval [a,b] is not a bracketing interval.
You need f(a) and f(b) to have different signs (f(a) * f(b) < 0).
Consider a different bracket or try fzero(f, c) with an initial guess c.
Here's what I would do:
using JuMP
import Ipopt
import Roots
function main()
k, d, c1, c2, c3, m = 0.5, 1, 0, 2, 1, 10
nparam = 2 * d + 2
m -= 1
vGrid = range(0, 1; length = m)
function obj(α::T...) where {T<:Real}
αb, αs = α[1:d+1], α[d+2:end]
B(v) = sum(αb[i] * v^(i-1) for i in 1:d+1)
B1(v) = sum(αb[i] * (i-1) * v^(i-2) for i in 2:d+1)
S(v) = sum(αs[i] * v^(i-1) for i in 1:d+1)
S1(v) = sum(αs[i] * (i-1) * v^(i-2) for i in 2:d+1)
function FOCb(y)
sy = S(y)
binv = Roots.fzero(q -> B(q) - sy, zero(T))
return k * y * S1(y) + sy - binv
end
function FOCs(x)
bx = B(x)
sinv = Roots.fzero(q -> S(q) - bx, zero(T))
return (1-k) * (1-x) * B1(x) - B(x) + sinv
end
return sum(FOCb(x)^2 + FOCs(x)^2 for x in vGrid)
end
αa = [1/12, 2/3, 1/4, 2/3]
model = Model(Ipopt.Optimizer)
#variable(model, -c3 <= α[i=1:nparam] <= c3, start = αa[i]+ 0.1 * rand())
#constraints(model, begin
[j = 1:m], sum(α[i] * (i-1) * vGrid[j]^(i-2) for i in 2:d+1) >= c1
[j = 1:m], sum(α[d+1+i] * (i-1) * vGrid[j]^(i-2) for i in 2:d+1) >= c1
end)
register(model, :obj, nparam, obj; autodiff = true)
#NLobjective(model, Min, obj(α...))
optimize!(model)
print(solution_summary(model))
return value.(α)
end
main()
I have written a surrogate function called GEKPLS and I am trying to make the code work with an optimizer to find the optimal theta parameter values.
As a first step, I'm trying to make it work with Zygote and have the following code:
function min_rlfv(theta)
g = GEKPLS(X, y, grads, n_comp, delta_x, xlimits, extra_points, theta)
return -g.reduced_likelihood_function_value
end
Zygote.gradient(min_rlfv, [0.01, 0.1])
Running the above code results in the following error message:
ERROR: Need an adjoint for constructor LinearAlgebra.QRCompactWYQ{Float64, Matrix{Float64}}. Gradient is of type LinearAlgebra.Transpose{Float64, Matrix{Float64}}
The stacktrace leads to the following function. The highlighted line in my code editor is Q, G = qr(Ft) from the following code block:
function _reduced_likelihood_function(theta, kernel_type, d, nt, ij, y_norma, noise = 0.0)
reduced_likelihood_function_value = -Inf
nugget = 1000000.0 * eps() #a jitter for numerical stability;
if kernel_type == "squar_exp"
r = squar_exp(theta, d)
end
R = (I + zeros(nt, nt)) .* (1.0 + nugget + noise)
for k in 1:size(ij)[1]
R[ij[k, 1], ij[k, 2]] = r[k]
R[ij[k, 2], ij[k, 1]] = r[k]
end
C = cholesky(R).L
F = ones(nt, 1)
Ft = C \ F
Q, G = qr(Ft)
Q = Array(Q)
Yt = C \ y_norma
beta = G \ [(transpose(Q) ⋅ Yt)]
rho = Yt .- (Ft .* beta)
gamma = transpose(C) \ rho
sigma2 = sum((rho) .^ 2, dims = 1) / nt
detR = prod(diag(C) .^ (2.0 / nt))
reduced_likelihood_function_value = -nt * log10(sum(sigma2)) - nt * log10(detR)
return beta, gamma, reduced_likelihood_function_value
end
Any pointers on how this can be fixed?
I am trying to practice fitting with the Lsq-Fit-function in Julia.
The derivative of a Cauchy-distribution with parameters \gamma and x_0.
Following this manual I tried
f(x, x_0, γ) = -2*(x - x_0)*(π * γ^3 * (1 + ((x - x_0)/γ)^2)^2)^(-1)
x_0 = 3350
γ = 50
xarr = range(3000, length = 5000, stop = 4000)
yarr = [f(x, x_0, γ) for x in xarr]
using LsqFit
# p ≡ [x_0, γ]
model(x, p) = -2*(x - p[1])*(π * (p[2])^3 * (1 + ((x - p[1])/p[2])^2)^2)^(-1)
p0 = [3349, 49]
curve_fit(model, xarr, yarr, p0)
param = fit.param
... and it does not work, giving a MethodError: no method matching -(::StepRangeLen[...], leaving me confused.
Can please somebody tell me what I am doing wrong?
There are a few issues with what you've written:
the model function is meant to be called with its first argument (x) being the full vector of independent variables, not just one value. This is where the error you mention comes from:
julia> model(x, p) = -2*(x - p[1])*(π * (p[2])^3 * (1 + ((x - p[1])/p[2])^2)^2)^(-1);
julia> p0 = [3349, 49];
julia> model(xarr, p0);
ERROR: MethodError: no method matching -(::StepRangeLen{Float64,Base.TwicePrecision{Float64},Base.TwicePrecision{Float64}}, ::Float64)
One way to fix this is to use the dot notation to broadcast all operators so that they work elementwise:
julia> model(x, p) = -2*(x .- p[1]) ./ (π * (p[2])^3 * (1 .+ ((x .- p[1])/p[2]).^2).^2);
julia> model(xarr, p0); # => No error
but if this is too tedious you can let the #. macro do the work for you:
# just put #. in front of the expression to transform every
# occurrence of a-b into a.-b (and likewise for all operators)
# which means to compute the operation elementwise
julia> model(x, p) = #. -2*(x - p[1])*(π * (p[2])^3 * (1 + ((x - p[1])/p[2])^2)^2)^(-1);
julia> model(xarr, p0); # => No error
Another issue is that the parameters you're looking for are meant to be floating-point values. But your initial guess p0 is initialized with integers, which confuses curve_fit. There are two ways of fixing this. Either put floating-point values in p0:
julia> p0 = [3349.0, 49.0]
2-element Array{Float64,1}:
3349.0
49.0
or use a typed array initializer to specify explicitly the element type:
julia> p0 = Float64[3349, 49]
2-element Array{Float64,1}:
3349.0
49.0
This is not really an error, but I would find it more intuitive to compute a/b instead of a*b^(-1). Also, yarr can be computed with a simple broadcast using dot notation instead of a comprehension.
Wrapping all this together:
f(x, x_0, γ) = -2*(x - x_0)*(π * γ^3 * (1 + ((x - x_0)/γ)^2)^2)^(-1)
(x_0, γ) = (3350, 50)
xarr = range(3000, length = 5000, stop = 4000);
# use dot-notation to "broadcast" f and map it
# elementwise to elements of xarr
yarr = f.(xarr, x_0, γ);
using LsqFit
model(x, p) = #. -2*(x - p[1]) / (π * (p[2])^3 * (1 + ((x - p[1])/p[2])^2)^2)
p0 = Float64[3300, 10]
fit = curve_fit(model, xarr, yarr, p0)
yields:
julia> fit.param
2-element Array{Float64,1}:
3349.999986535933
49.99999203625603
I am trying to use NLsolve.jl to solve a set of 6 equations with known Jacobian matrix.
The unknowns are in the array U, so following the repository example I created a function fun! as
function fun!(F, U, k)
W = anotherFunction(U, k)
F[1] = U[1] + 4*k[1]*U[4] - W[1]
F[2] = U[2] - 4*k[2]*U[5] - W[2]
F[3] = U[3] - 4*k[3]*U[6] - W[3]
F[4] = U[1]*(U[4]*U[4]*U[4]*U[4]) -
U[2]*(U[5]*U[5]*U[5]*U[5]) -
U[3]*(U[6]*U[6]*U[6]*U[6])
F[5] = k[7]*(U[4]*U[4]*k[4] - 1 ) -
(k[8]*(U[5]*U[5]*k[5] - 1))
F[6] = k[7]*(U[4]*U[4]*k[4] - 1 ) -
(k[9]*(U[6]*U[6]*k[6] - 1))
end
where the values of W are computed throuhg another function. Similarly, the Jacobian reads
function jac!(J, U, k)
J = eye(6)
J[1,4] = 4*k[1]
J[2,5] = -4*k[2]
J[3,6] = -4*k[3]
J[4,1] = (U[4]*U[4]*U[4]*U[4])
J[4,2] = -(U[5]*U[5]*U[5]*U[5])
J[4,3] = -(U[6]*U[6]*U[6]*U[6])
J[4,4] = 4*U[1]*(U[4]*U[4]*U[4])
J[4,5] = -4*U[2]*(U[5]*U[5]*U[5])
J[4,6] = -4*U[3]*(U[6]*U[6]*U[6])
J[5,1] = 0.
J[5,2] = 0.
J[5,4] = 2*k[7]*U[4]*k[4]
J[5,5] = -2*k[8]*U[5]*k[5]
J[6,1] = 0.
J[6,3] = 0.
J[6,4] = 2*k[7]*U[4]*k[4]
J[6,6] = -2*k[9]*U[6]*k[6]
end
The k array is constant and defined before calling nlsove, therefore I create two more functions to enclose fun! and jac! as described here, here, and here
f_closure!(U) = fun!(F, U, k)
j_closure!(U) = jac!(J, U, k)
However, calling res = nlsolve(f_closure!, j_closure!, U) results in this error
ERROR: LoadError: MethodError: no method matching
(::#f_closure!#5{Array{Float64,1}})(::Array{Float64,1},
::Array{Float64,1}) Closest candidates are: f_closure!(::Any)
Is there a different way to pass the k array? Thanks!
I’m trying to optimize a function using one of the algorithms that require a gradient. Basically I’m trying to learn how to optimize a function using a gradient in Julia. I’m fairly confident that my gradient is specified correctly. I know this because the similarly defined Matlab function for the gradient gives me the same values as in Julia for some test values of the arguments. Also, the Matlab version using fminunc with the gradient seems to optimize the function fine.
However when I run the Julia script, I seem to get the following error:
julia> include("ex2b.jl")
ERROR: `g!` has no method matching g!(::Array{Float64,1}, ::Array{Float64,1})
while loading ...\ex2b.jl, in ex
pression starting on line 64
I'm running Julia 0.3.2 on a windows 7 32bit machine. Here is the code (basically a translation of some Matlab to Julia):
using Optim
function mapFeature(X1, X2)
degrees = 5
out = ones(size(X1)[1])
for i in range(1, degrees+1)
for j in range(0, i+1)
term = reshape( (X1.^(i-j) .* X2.^(j)), size(X1.^(i-j))[1], 1)
out = hcat(out, term)
end
end
return out
end
function sigmoid(z)
return 1 ./ (1 + exp(-z))
end
function costFunc_logistic(theta, X, y, lam)
m = length(y)
regularization = sum(theta[2:end].^2) * lam / (2 * m)
return sum( (-y .* log(sigmoid(X * theta)) - (1 - y) .* log(1 - sigmoid(X * theta))) ) ./ m + regularization
end
function costFunc_logistic_gradient!(theta, X, y, lam, m)
grad= X' * ( sigmoid(X * theta) .- y ) ./ m
grad[2:end] = grad[2:end] + theta[2:end] .* lam / m
return grad
end
data = readcsv("ex2data2.txt")
X = mapFeature(data[:,1], data[:,2])
m, n = size(data)
y = data[:, end]
theta = zeros(size(X)[2])
lam = 1.0
f(theta::Array) = costFunc_logistic(theta, X, y, lam)
g!(theta::Array) = costFunc_logistic_gradient!(theta, X, y, lam, m)
optimize(f, g!, theta, method = :l_bfgs)
And here is some of the data:
0.051267,0.69956,1
-0.092742,0.68494,1
-0.21371,0.69225,1
-0.375,0.50219,1
-0.51325,0.46564,1
-0.52477,0.2098,1
-0.39804,0.034357,1
-0.30588,-0.19225,1
0.016705,-0.40424,1
0.13191,-0.51389,1
0.38537,-0.56506,1
0.52938,-0.5212,1
0.63882,-0.24342,1
0.73675,-0.18494,1
0.54666,0.48757,1
0.322,0.5826,1
0.16647,0.53874,1
-0.046659,0.81652,1
-0.17339,0.69956,1
-0.47869,0.63377,1
-0.60541,0.59722,1
-0.62846,0.33406,1
-0.59389,0.005117,1
-0.42108,-0.27266,1
-0.11578,-0.39693,1
0.20104,-0.60161,1
0.46601,-0.53582,1
0.67339,-0.53582,1
-0.13882,0.54605,1
-0.29435,0.77997,1
-0.26555,0.96272,1
-0.16187,0.8019,1
-0.17339,0.64839,1
-0.28283,0.47295,1
-0.36348,0.31213,1
-0.30012,0.027047,1
-0.23675,-0.21418,1
-0.06394,-0.18494,1
0.062788,-0.16301,1
0.22984,-0.41155,1
0.2932,-0.2288,1
0.48329,-0.18494,1
0.64459,-0.14108,1
0.46025,0.012427,1
0.6273,0.15863,1
0.57546,0.26827,1
0.72523,0.44371,1
0.22408,0.52412,1
0.44297,0.67032,1
0.322,0.69225,1
0.13767,0.57529,1
-0.0063364,0.39985,1
-0.092742,0.55336,1
-0.20795,0.35599,1
-0.20795,0.17325,1
-0.43836,0.21711,1
-0.21947,-0.016813,1
-0.13882,-0.27266,1
0.18376,0.93348,0
0.22408,0.77997,0
Let me know if you guys need additional details. Btw, this relates to a coursera machine learning course if curious.
The gradient should not be a function to compute the gradient,
but a function to store it
(hence the exclamation mark in the function name, and the second argument in the error message).
The following seems to work.
function g!(theta::Array, storage::Array)
storage[:] = costFunc_logistic_gradient!(theta, X, y, lam, m)
end
optimize(f, g!, theta, method = :l_bfgs)
The same using closures and currying (version for those who got used to a function that returns the cost and gradient):
function cost_gradient(θ, X, y, λ)
m = length(y);
return (θ::Array) -> begin
h = sigmoid(X * θ); #(m,n+1)*(n+1,1) -> (m,1)
J = (1 / m) * sum(-y .* log(h) .- (1 - y) .* log(1 - h)) + λ / (2 * m) * sum(θ[2:end] .^ 2);
end, (θ::Array, storage::Array) -> begin
h = sigmoid(X * θ); #(m,n+1)*(n+1,1) -> (m,1)
storage[:] = (1 / m) * (X' * (h .- y)) + (λ / m) * [0; θ[2:end]];
end
end
Then, somewhere in the code:
initialθ = zeros(n,1);
f, g! = cost_gradient(initialθ, X, y, λ);
res = optimize(f, g!, initialθ, method = :cg, iterations = your_iterations);
θ = res.minimum;