Vectornaut
b92be312e8
This PR adds code for a Julia-language prototype of a configuration solver, in the `engine-proto` folder. It uses Julia version 1.10.0. ### Approaches Development of this PR tried two broad approaches to the constraint geometry problem. Each one suggested various solution techniques. The Gram matrix approach, with the low-rank factorization technique, seems the most promising. - **Algebraic** *(In the `alg-test` subfolder).* Write the constraints as polynomials in the inversive coordinates of the elements, and use computational algebraic geometry techniques to solve the resulting system. We tried the following techniques. - **Gröbner bases** *(`Engine.Algebraic.jl`).* Symbolic. Find a Gröbner basis for the ideal generated by the constraint equations. Information about the solution variety, like its codimension, is then relatively easy to extract. - **Homotopy continuation** *(`Engine.Numerical.jl`).* Numerical. Cut the solution set along a random hyperplane to get a generic zero-dimensional slice, and then use a fancy homotopy technique to approximate the points in that slice. A few notes about our experiences can be found on the [engine prototype](wiki/Engine-prototype) wiki page. - **Gram matrix** *(in the `gram-test` subfolder).* A construction is described completely, up to conformal transformations, by the Gram matrix of the vectors representing its elements. Express the constraints as fixed entries of the Gram matrix, and use numerical linear algebra techniques to find a list of vectors whose Gram matrix fits the bill. We tried the following techniques. - **LDL decomposition** *(`gram-test.sage`, `gram-test.jl`, `overlap-test.jl`).* Find a cluster of up to five elements whose Gram matrix is completely filled in by the constraints. Use LDL decomposition to find a list of vectors with that Gram matrix. This technique can be made algebraic, as seen in `overlap-test.jl`. - **Low-rank factorization** *(source files listed in findings section).* Write down a quadratic loss function that says how far a set of vectors is from meeting the Gram matrix constraints. Use a smooth optimization technique like Newton's method or gradient descent to find a zero of the loss function. In addition to the polished prototype described in the results section, we have an early prototype using an off-the-shelf factorization package (`low-rank-test.jl`) and an visualization of the loss function landscape near global minima (`basin-shapes.jl`). The [Gram matrix parameterization](wiki/Gram-matrix-parameterization) wiki page contains detailed notes on this approach. ### Findings With the algebraic approach, we hit a performance wall pretty quickly as our constructions grew. It was often hard to find real solutions of the polynomial system, since the techniques we use work most naturally in the complex world. With the Gram matrix approach, on the other hand, we could solve interesting problems in acceptably short times using the low-rank factorization technique. We put the optimization routine in its own module (`Engine.jl`) and used it to solve five example problems: - `overlapping-pyramids.jl` - `circles-in-triangle.jl` - `sphere-in-tetrahedron.jl` - `tetrahedron-radius-ratio.jl` - `irisawa-hexlet.jl` We plan to use low-rank factorization of the Gram matrix in our first app prototype. ### Visualizations We used the visualizer in the `ganja-test` folder to visually check our low-rank factorization results. The visualizer runs [Ganja.js](https://enkimute.github.io/ganja.js/) in an Electron app, made with [Blink](https://github.com/JuliaGizmos/Blink.jl). Although Ganja.js makes beautiful pictures under most circumstances, we found two obstacles to using it in production. - It seems to have precision problems with low-curvature spheres. - We couldn't figure out how to customize its clipping and transparency settings, and the default settings often obscure construction details. Co-authored-by: Aaron Fenyes <aaron.fenyes@fareycircles.ooo> Co-authored-by: Glen Whitney <glen@studioinfinity.org> Reviewed-on: #13 Co-authored-by: Vectornaut <vectornaut@nobody@nowhere.net> Co-committed-by: Vectornaut <vectornaut@nobody@nowhere.net>
450 lines
13 KiB
Julia
450 lines
13 KiB
Julia
module Engine
|
|
|
|
using LinearAlgebra
|
|
using GenericLinearAlgebra
|
|
using SparseArrays
|
|
using Random
|
|
using Optim
|
|
|
|
export
|
|
rand_on_shell, Q, DescentHistory,
|
|
realize_gram_gradient, realize_gram_newton, realize_gram_optim, realize_gram
|
|
|
|
# === guessing ===
|
|
|
|
sconh(t, u) = 0.5*(exp(t) + u*exp(-t))
|
|
|
|
function rand_on_sphere(rng::AbstractRNG, ::Type{T}, n) where T
|
|
out = randn(rng, T, n)
|
|
tries_left = 2
|
|
while dot(out, out) < 1e-6 && tries_left > 0
|
|
out = randn(rng, T, n)
|
|
tries_left -= 1
|
|
end
|
|
normalize(out)
|
|
end
|
|
|
|
##[TO DO] write a test to confirm that the outputs are on the correct shells
|
|
function rand_on_shell(rng::AbstractRNG, shell::T) where T <: Number
|
|
space_part = rand_on_sphere(rng, T, 4)
|
|
rapidity = randn(rng, T)
|
|
sig = sign(shell)
|
|
nullmix * [sconh(rapidity, sig)*space_part; sconh(rapidity, -sig)]
|
|
end
|
|
|
|
rand_on_shell(rng::AbstractRNG, shells::Array{T}) where T <: Number =
|
|
hcat([rand_on_shell(rng, sh) for sh in shells]...)
|
|
|
|
rand_on_shell(shells::Array{<:Number}) = rand_on_shell(Random.default_rng(), shells)
|
|
|
|
# === elements ===
|
|
|
|
point(pos) = [pos; 0.5; 0.5 * dot(pos, pos)]
|
|
|
|
plane(normal, offset) = [-normal; 0; -offset]
|
|
|
|
function sphere(center, radius)
|
|
dist_sq = dot(center, center)
|
|
[
|
|
center / radius;
|
|
0.5 / radius;
|
|
0.5 * (dist_sq / radius - radius)
|
|
]
|
|
end
|
|
|
|
# === Gram matrix realization ===
|
|
|
|
# basis changes
|
|
nullmix = [Matrix{Int64}(I, 3, 3) zeros(Int64, 3, 2); zeros(Int64, 2, 3) [-1 1; 1 1]//2]
|
|
unmix = [Matrix{Int64}(I, 3, 3) zeros(Int64, 3, 2); zeros(Int64, 2, 3) [-1 1; 1 1]]
|
|
|
|
# the Lorentz form
|
|
Q = [Matrix{Int64}(I, 3, 3) zeros(Int64, 3, 2); zeros(Int64, 2, 3) [0 -2; -2 0]]
|
|
|
|
# project a matrix onto the subspace of matrices whose entries vanish away from
|
|
# the given indices
|
|
function proj_to_entries(mat, indices)
|
|
result = zeros(size(mat))
|
|
for (j, k) in indices
|
|
result[j, k] = mat[j, k]
|
|
end
|
|
result
|
|
end
|
|
|
|
# the difference between the matrices `target` and `attempt`, projected onto the
|
|
# subspace of matrices whose entries vanish at each empty index of `target`
|
|
function proj_diff(target::SparseMatrixCSC{T, <:Any}, attempt::Matrix{T}) where T
|
|
J, K, values = findnz(target)
|
|
result = zeros(size(target))
|
|
for (j, k, val) in zip(J, K, values)
|
|
result[j, k] = val - attempt[j, k]
|
|
end
|
|
result
|
|
end
|
|
|
|
# a type for keeping track of gradient descent history
|
|
struct DescentHistory{T}
|
|
scaled_loss::Array{T}
|
|
neg_grad::Array{Matrix{T}}
|
|
base_step::Array{Matrix{T}}
|
|
hess::Array{Hermitian{T, Matrix{T}}}
|
|
slope::Array{T}
|
|
stepsize::Array{T}
|
|
positive::Array{Bool}
|
|
backoff_steps::Array{Int64}
|
|
last_line_L::Array{Matrix{T}}
|
|
last_line_loss::Array{T}
|
|
|
|
function DescentHistory{T}(
|
|
scaled_loss = Array{T}(undef, 0),
|
|
neg_grad = Array{Matrix{T}}(undef, 0),
|
|
hess = Array{Hermitian{T, Matrix{T}}}(undef, 0),
|
|
base_step = Array{Matrix{T}}(undef, 0),
|
|
slope = Array{T}(undef, 0),
|
|
stepsize = Array{T}(undef, 0),
|
|
positive = Bool[],
|
|
backoff_steps = Int64[],
|
|
last_line_L = Array{Matrix{T}}(undef, 0),
|
|
last_line_loss = Array{T}(undef, 0)
|
|
) where T
|
|
new(scaled_loss, neg_grad, hess, base_step, slope, stepsize, positive, backoff_steps, last_line_L, last_line_loss)
|
|
end
|
|
end
|
|
|
|
# seek a matrix `L` for which `L'QL` matches the sparse matrix `gram` at every
|
|
# explicit entry of `gram`. use gradient descent starting from `guess`
|
|
function realize_gram_gradient(
|
|
gram::SparseMatrixCSC{T, <:Any},
|
|
guess::Matrix{T};
|
|
scaled_tol = 1e-30,
|
|
min_efficiency = 0.5,
|
|
init_stepsize = 1.0,
|
|
backoff = 0.9,
|
|
max_descent_steps = 600,
|
|
max_backoff_steps = 110
|
|
) where T <: Number
|
|
# start history
|
|
history = DescentHistory{T}()
|
|
|
|
# scale tolerance
|
|
scale_adjustment = sqrt(T(nnz(gram)))
|
|
tol = scale_adjustment * scaled_tol
|
|
|
|
# initialize variables
|
|
stepsize = init_stepsize
|
|
L = copy(guess)
|
|
|
|
# do gradient descent
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
loss = dot(Δ_proj, Δ_proj)
|
|
for _ in 1:max_descent_steps
|
|
# stop if the loss is tolerably low
|
|
if loss < tol
|
|
break
|
|
end
|
|
|
|
# find negative gradient of loss function
|
|
neg_grad = 4*Q*L*Δ_proj
|
|
slope = norm(neg_grad)
|
|
dir = neg_grad / slope
|
|
|
|
# store current position, loss, and slope
|
|
L_last = L
|
|
loss_last = loss
|
|
push!(history.scaled_loss, loss / scale_adjustment)
|
|
push!(history.neg_grad, neg_grad)
|
|
push!(history.slope, slope)
|
|
|
|
# find a good step size using backtracking line search
|
|
push!(history.stepsize, 0)
|
|
push!(history.backoff_steps, max_backoff_steps)
|
|
empty!(history.last_line_L)
|
|
empty!(history.last_line_loss)
|
|
for backoff_steps in 0:max_backoff_steps
|
|
history.stepsize[end] = stepsize
|
|
L = L_last + stepsize * dir
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
loss = dot(Δ_proj, Δ_proj)
|
|
improvement = loss_last - loss
|
|
push!(history.last_line_L, L)
|
|
push!(history.last_line_loss, loss / scale_adjustment)
|
|
if improvement >= min_efficiency * stepsize * slope
|
|
history.backoff_steps[end] = backoff_steps
|
|
break
|
|
end
|
|
stepsize *= backoff
|
|
end
|
|
|
|
# [DEBUG] if we've hit a wall, quit
|
|
if history.backoff_steps[end] == max_backoff_steps
|
|
break
|
|
end
|
|
end
|
|
|
|
# return the factorization and its history
|
|
push!(history.scaled_loss, loss / scale_adjustment)
|
|
L, history
|
|
end
|
|
|
|
function basis_matrix(::Type{T}, j, k, dims) where T
|
|
result = zeros(T, dims)
|
|
result[j, k] = one(T)
|
|
result
|
|
end
|
|
|
|
# seek a matrix `L` for which `L'QL` matches the sparse matrix `gram` at every
|
|
# explicit entry of `gram`. use Newton's method starting from `guess`
|
|
function realize_gram_newton(
|
|
gram::SparseMatrixCSC{T, <:Any},
|
|
guess::Matrix{T};
|
|
scaled_tol = 1e-30,
|
|
rate = 1,
|
|
max_steps = 100
|
|
) where T <: Number
|
|
# start history
|
|
history = DescentHistory{T}()
|
|
|
|
# find the dimension of the search space
|
|
dims = size(guess)
|
|
element_dim, construction_dim = dims
|
|
total_dim = element_dim * construction_dim
|
|
|
|
# list the constrained entries of the gram matrix
|
|
J, K, _ = findnz(gram)
|
|
constrained = zip(J, K)
|
|
|
|
# scale the tolerance
|
|
scale_adjustment = sqrt(T(length(constrained)))
|
|
tol = scale_adjustment * scaled_tol
|
|
|
|
# use Newton's method
|
|
L = copy(guess)
|
|
for step in 0:max_steps
|
|
# evaluate the loss function
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
loss = dot(Δ_proj, Δ_proj)
|
|
|
|
# store the current loss
|
|
push!(history.scaled_loss, loss / scale_adjustment)
|
|
|
|
# stop if the loss is tolerably low
|
|
if loss < tol || step > max_steps
|
|
break
|
|
end
|
|
|
|
# find the negative gradient of loss function
|
|
neg_grad = 4*Q*L*Δ_proj
|
|
|
|
# find the negative Hessian of the loss function
|
|
hess = Matrix{T}(undef, total_dim, total_dim)
|
|
indices = [(j, k) for k in 1:construction_dim for j in 1:element_dim]
|
|
for (j, k) in indices
|
|
basis_mat = basis_matrix(T, j, k, dims)
|
|
neg_dΔ = basis_mat'*Q*L + L'*Q*basis_mat
|
|
neg_dΔ_proj = proj_to_entries(neg_dΔ, constrained)
|
|
deriv_grad = 4*Q*(-basis_mat*Δ_proj + L*neg_dΔ_proj)
|
|
hess[:, (k-1)*element_dim + j] = reshape(deriv_grad, total_dim)
|
|
end
|
|
hess = Hermitian(hess)
|
|
push!(history.hess, hess)
|
|
|
|
# compute the Newton step
|
|
step = hess \ reshape(neg_grad, total_dim)
|
|
L += rate * reshape(step, dims)
|
|
end
|
|
|
|
# return the factorization and its history
|
|
L, history
|
|
end
|
|
|
|
LinearAlgebra.eigen!(A::Symmetric{BigFloat, Matrix{BigFloat}}; sortby::Nothing) =
|
|
eigen!(Hermitian(A))
|
|
|
|
function convertnz(type, mat)
|
|
J, K, values = findnz(mat)
|
|
sparse(J, K, type.(values))
|
|
end
|
|
|
|
function realize_gram_optim(
|
|
gram::SparseMatrixCSC{T, <:Any},
|
|
guess::Matrix{T}
|
|
) where T <: Number
|
|
# find the dimension of the search space
|
|
dims = size(guess)
|
|
element_dim, construction_dim = dims
|
|
total_dim = element_dim * construction_dim
|
|
|
|
# list the constrained entries of the gram matrix
|
|
J, K, _ = findnz(gram)
|
|
constrained = zip(J, K)
|
|
|
|
# scale the loss function
|
|
scale_adjustment = length(constrained)
|
|
|
|
function loss(L_vec)
|
|
L = reshape(L_vec, dims)
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
dot(Δ_proj, Δ_proj) / scale_adjustment
|
|
end
|
|
|
|
function loss_grad!(storage, L_vec)
|
|
L = reshape(L_vec, dims)
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
storage .= reshape(-4*Q*L*Δ_proj, total_dim) / scale_adjustment
|
|
end
|
|
|
|
function loss_hess!(storage, L_vec)
|
|
L = reshape(L_vec, dims)
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
indices = [(j, k) for k in 1:construction_dim for j in 1:element_dim]
|
|
for (j, k) in indices
|
|
basis_mat = basis_matrix(T, j, k, dims)
|
|
neg_dΔ = basis_mat'*Q*L + L'*Q*basis_mat
|
|
neg_dΔ_proj = proj_to_entries(neg_dΔ, constrained)
|
|
deriv_grad = 4*Q*(-basis_mat*Δ_proj + L*neg_dΔ_proj) / scale_adjustment
|
|
storage[:, (k-1)*element_dim + j] = reshape(deriv_grad, total_dim)
|
|
end
|
|
end
|
|
|
|
optimize(
|
|
loss, loss_grad!, loss_hess!,
|
|
reshape(guess, total_dim),
|
|
Newton()
|
|
)
|
|
end
|
|
|
|
# seek a matrix `L` for which `L'QL` matches the sparse matrix `gram` at every
|
|
# explicit entry of `gram`. use gradient descent starting from `guess`
|
|
function realize_gram(
|
|
gram::SparseMatrixCSC{T, <:Any},
|
|
guess::Matrix{T},
|
|
frozen = nothing;
|
|
scaled_tol = 1e-30,
|
|
min_efficiency = 0.5,
|
|
init_rate = 1.0,
|
|
backoff = 0.9,
|
|
reg_scale = 1.1,
|
|
max_descent_steps = 200,
|
|
max_backoff_steps = 110
|
|
) where T <: Number
|
|
# start history
|
|
history = DescentHistory{T}()
|
|
|
|
# find the dimension of the search space
|
|
dims = size(guess)
|
|
element_dim, construction_dim = dims
|
|
total_dim = element_dim * construction_dim
|
|
|
|
# list the constrained entries of the gram matrix
|
|
J, K, _ = findnz(gram)
|
|
constrained = zip(J, K)
|
|
|
|
# scale the tolerance
|
|
scale_adjustment = sqrt(T(length(constrained)))
|
|
tol = scale_adjustment * scaled_tol
|
|
|
|
# list the un-frozen indices
|
|
has_frozen = !isnothing(frozen)
|
|
if has_frozen
|
|
is_unfrozen = fill(true, size(guess))
|
|
is_unfrozen[frozen] .= false
|
|
unfrozen = findall(is_unfrozen)
|
|
unfrozen_stacked = reshape(is_unfrozen, total_dim)
|
|
end
|
|
|
|
# initialize variables
|
|
grad_rate = init_rate
|
|
L = copy(guess)
|
|
|
|
# use Newton's method with backtracking and gradient descent backup
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
loss = dot(Δ_proj, Δ_proj)
|
|
for step in 1:max_descent_steps
|
|
# stop if the loss is tolerably low
|
|
if loss < tol
|
|
break
|
|
end
|
|
|
|
# find the negative gradient of loss function
|
|
neg_grad = 4*Q*L*Δ_proj
|
|
|
|
# find the negative Hessian of the loss function
|
|
hess = Matrix{T}(undef, total_dim, total_dim)
|
|
indices = [(j, k) for k in 1:construction_dim for j in 1:element_dim]
|
|
for (j, k) in indices
|
|
basis_mat = basis_matrix(T, j, k, dims)
|
|
neg_dΔ = basis_mat'*Q*L + L'*Q*basis_mat
|
|
neg_dΔ_proj = proj_to_entries(neg_dΔ, constrained)
|
|
deriv_grad = 4*Q*(-basis_mat*Δ_proj + L*neg_dΔ_proj)
|
|
hess[:, (k-1)*element_dim + j] = reshape(deriv_grad, total_dim)
|
|
end
|
|
hess = Hermitian(hess)
|
|
push!(history.hess, hess)
|
|
|
|
# regularize the Hessian
|
|
min_eigval = minimum(eigvals(hess))
|
|
push!(history.positive, min_eigval > 0)
|
|
if min_eigval <= 0
|
|
hess -= reg_scale * min_eigval * I
|
|
end
|
|
|
|
# compute the Newton step
|
|
neg_grad_stacked = reshape(neg_grad, total_dim)
|
|
if has_frozen
|
|
hess = hess[unfrozen_stacked, unfrozen_stacked]
|
|
neg_grad_compressed = neg_grad_stacked[unfrozen_stacked]
|
|
else
|
|
neg_grad_compressed = neg_grad_stacked
|
|
end
|
|
base_step_compressed = hess \ neg_grad_compressed
|
|
if has_frozen
|
|
base_step_stacked = zeros(total_dim)
|
|
base_step_stacked[unfrozen_stacked] .= base_step_compressed
|
|
else
|
|
base_step_stacked = base_step_compressed
|
|
end
|
|
base_step = reshape(base_step_stacked, dims)
|
|
push!(history.base_step, base_step)
|
|
|
|
# store the current position, loss, and slope
|
|
L_last = L
|
|
loss_last = loss
|
|
push!(history.scaled_loss, loss / scale_adjustment)
|
|
push!(history.neg_grad, neg_grad)
|
|
push!(history.slope, norm(neg_grad))
|
|
|
|
# find a good step size using backtracking line search
|
|
push!(history.stepsize, 0)
|
|
push!(history.backoff_steps, max_backoff_steps)
|
|
empty!(history.last_line_L)
|
|
empty!(history.last_line_loss)
|
|
rate = one(T)
|
|
step_success = false
|
|
for backoff_steps in 0:max_backoff_steps
|
|
history.stepsize[end] = rate
|
|
L = L_last + rate * base_step
|
|
Δ_proj = proj_diff(gram, L'*Q*L)
|
|
loss = dot(Δ_proj, Δ_proj)
|
|
improvement = loss_last - loss
|
|
push!(history.last_line_L, L)
|
|
push!(history.last_line_loss, loss / scale_adjustment)
|
|
if improvement >= min_efficiency * rate * dot(neg_grad, base_step)
|
|
history.backoff_steps[end] = backoff_steps
|
|
step_success = true
|
|
break
|
|
end
|
|
rate *= backoff
|
|
end
|
|
|
|
# if we've hit a wall, quit
|
|
if !step_success
|
|
return L_last, false, history
|
|
end
|
|
end
|
|
|
|
# return the factorization and its history
|
|
push!(history.scaled_loss, loss / scale_adjustment)
|
|
L, loss < tol, history
|
|
end
|
|
|
|
end |