Skip to content

use normal entrypoints into base's display pipeline #736

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
rokke-git wants to merge 1 commit into
base: main
Choose a base branch
from
Open

use normal entrypoints into base's display pipeline #736

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions docs/src/index.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,8 +113,11 @@ julia> I = [1, 4, 3, 5]; J = [4, 7, 18, 9]; V = [1, 2, -5, 3];

julia> S = sparse(I,J,V)
5×18 SparseMatrixCSC{Int64, Int64} with 4 stored entries:
⎡⠀⠈⠀⠀⠀⠀⠀⠀⢀⎤
⎣⠀⠀⠀⠂⡀⠀⠀⠀⠀⎦
⋅ ⋅ ⋅ 1 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ -5
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ 2 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ 3 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅

julia> R = sparsevec(I,V)
5-element SparseVector{Int64, Int64} with 4 stored entries:
Expand Down
12 changes: 6 additions & 6 deletions src/linalg.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2094,19 +2094,19 @@ julia> A[4:4:8] .= 1;

julia> A
3×3 SparseMatrixCSC{Float64, Int64} with 2 stored entries:
⋅ 1.0 ⋅
⋅ ⋅ 1.0
⋅ ⋅ ⋅
⋅ 1.0 ⋅
⋅ ⋅ 1.0
⋅ ⋅ ⋅

julia> C = spzeros(3,3);

julia> C[2:4:6] .= 2;

julia> C
3×3 SparseMatrixCSC{Float64, Int64} with 2 stored entries:
⋅ ⋅ ⋅
2.0 ⋅ ⋅
⋅ 2.0 ⋅
⋅ ⋅ ⋅
2.0 ⋅ ⋅
⋅ 2.0 ⋅

julia> SparseArrays.mergeinds!(C, A)
3×3 SparseMatrixCSC{Float64, Int64} with 4 stored entries:
Expand Down
12 changes: 6 additions & 6 deletions src/solvers/spqr.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,8 +192,8 @@ Q factor:
4×4 SparseArrays.SPQR.QRSparseQ{Float64, Int64}
R factor:
2×2 SparseMatrixCSC{Float64, Int64} with 2 stored entries:
-1.41421 ⋅
⋅ -1.41421
-1.41421 ⋅
⋅ -1.41421
Row permutation:
4-element Vector{Int64}:
1
Expand Down Expand Up @@ -471,10 +471,10 @@ when the problem is underdetermined.
```jldoctest
julia> A = sparse([1,2,4], [1,1,1], [1.0,1.0,1.0], 4, 2)
4×2 SparseMatrixCSC{Float64, Int64} with 3 stored entries:
1.0 ⋅
1.0 ⋅
⋅ ⋅
1.0 ⋅
1.0 ⋅
1.0 ⋅
⋅ ⋅
1.0 ⋅

julia> qr(A)\\fill(1.0, 4)
2-element Vector{Float64}:
Expand Down
222 changes: 120 additions & 102 deletions src/sparsematrix.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -334,10 +334,7 @@ function Base.isstored(A::AdjOrTrans{<:Any,<:AbstractSparseMatrixCSC}, i::Intege
return false
end

Base.replace_in_print_matrix(A::AbstractSparseMatrixCSCInclAdjointAndTranspose, i::Integer, j::Integer, s::AbstractString) =
Base.isstored(A, i, j) ? s : Base.replace_with_centered_mark(s)

function Base.array_summary(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose, dims::Tuple{Vararg{Base.OneTo}})
function Base.summary(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
_checkbuffers(S)

xnnz = nnz(S)
Expand All @@ -347,12 +344,26 @@ function Base.array_summary(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTran
nothing
end

# called by `show(io, MIME("text/plain"), ::AbstractSparseMatrixCSCInclAdjointAndTranspose)`
function Base.print_array(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
if max(size(S)...) < 16
Base.print_matrix(io, S)
using Base: show_circular
function Base.show(io::IO, ::MIME"text/plain", S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
isempty(S) && get(io, :compact, false) && return show(io, S)
summary(io, S)
isempty(S) && return

if get(io, :limit, false)
screen = get(io, :displaysize, displaysize(io))::Tuple{Int, Int}
screen[1] <= 4 && return print(io, ": …")
else
screen = typemax(Int), typemax(Int)
end

show_circular(io, S) && return
io = IOContext(io, :compact=>true, :typeinfo=>eltype(S), :SHOWN_SET=>S)

if !get(io, :limit, false) || screen[1] < size(S, 1) + 4 || screen[2] < 3size(S, 2)
_show_with_braille_patterns(io, S)
else
_show_with_dotted_zeros(io, S)
end
end

Expand All @@ -376,6 +387,8 @@ size(C::ColumnIndices) = (nnz(C.arr),)
eltype(C)(ind)
end

colvals(C::AbstractSparseMatrixCSC{Tv,Ti}) where {Tv,Ti} = Vector{Ti}(ColumnIndices(C))

# always show matrices as `sparse(I, J, K)`
function Base.show(io::IO, _S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
_checkbuffers(_S)
Expand All @@ -398,97 +411,102 @@ function Base.show(io::IO, _S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
end

const brailleBlocks = UInt16['⠁', '⠂', '⠄', '⡀', '⠈', '⠐', '⠠', '⢀']
function _show_with_braille_patterns(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose)
m, n = size(S)
(m == 0 || n == 0) && return show(io, MIME("text/plain"), S)

# The maximal number of characters we allow to display the matrix
local maxHeight::Int, maxWidth::Int
maxHeight = displaysize(io)[1] - 4 # -4 from [Prompt, header, newline after elements, new prompt]
maxWidth = displaysize(io)[2] ÷ 2

# In the process of generating the braille pattern to display the nonzero
# structure of `S`, we need to be able to scale the matrix `S` to a
# smaller matrix with the same aspect ratio as `S`, but fits on the
# available screen space. The size of that smaller matrix is stored
# in the variables `scaleHeight` and `scaleWidth`. If no scaling is needed,
# we can use the size `m × n` of `S` directly.
# We determine if scaling is needed and set the scaling factors
# `scaleHeight` and `scaleWidth` accordingly. Note that each available
# character can contain up to 4 braille dots in its height (⡇) and up to
# 2 braille dots in its width (⠉).
if get(io, :limit, true) && (m > 4maxHeight || n > 2maxWidth)
s = min(2maxWidth / n, 4maxHeight / m)
scaleHeight = floor(Int, s * m)
scaleWidth = floor(Int, s * n)
function _show_with_braille_patterns(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose,

@stevengj stevengj Aug 6, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It looks like this function was mostly changed to compactify the code, but are there changes in behavior mixed in?

@rokke-git rokke-git Aug 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes, quite a few. it no longer scales by fractional (or sometimes uneven!) amounts, removing funky aliasing effects

see e.g. the changed tests for how it looks different

edit: "yes, quite a few" was meant to be sarcasm, sorry. I didn't mean to go off that hard but it's a complete rewrite; about the only thing remaining is the fact that is uses prints braille characters stored in an array

@rokke-git rokke-git Aug 6, 2026

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

it also uses a considerably faster algorithm; instead of looping over every single possible index we only check the indices with an item in them. also, it now tells you exactly how scaled the output actually is

@rokke-git rokke-git Aug 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(you have to view in a terminal since github's "fixed" width is not really fixed)
for example, #21 (also here):
old:

julia> laplacian_2d(11)
121×121 SparseMatrixCSC{Float64, Int64} with 561 stored entries:
⠻⣦⡀⠀⠀⠘⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⣀⠀⠀⠈⠻⢆⡀⠀⠀⠑⢄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠱⡀⠀⠀⠈⠱⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⡀⠀⠀⠘⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⠀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⣀⠀⠀⠀⠻⣦⡀⠀⠀⠑⢄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠛⣤⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠱⡀⠀⠀⠈⠻⣦⡀⠀⠀⠈⢢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠱⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠒⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⢆⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⡀⠀⠀⠈⢢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠛⣤⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠒⢄⠀⠀⠈⠻⣦⠀⠀⠀⠑⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠀⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠈⠢⡀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⢆⡀⠀⠀⠘⢄⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⡀⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠢⣀⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⡀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠱⣦⡀⠀⠀⠈
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠱⡀⠀⠀⠈⠻⣦

new

julia> laplacian_2d(10)
100×100 SparseMatrixCSC{Float64, Int64} with 460 stored entries:
⎡⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎤
⎢⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⢄⠀⠀⠈⠛⣤⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⠀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠑⢄⠀⠀⠀⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠛⣤⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⠀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠀⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠛⣤⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⠀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠀⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠛⣤⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⠀⠀⠀⠑⢄⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠀⠻⣦⡀⠀⠀⠑⢄⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⠀⠑⢄⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠛⣤⡀⠀⠀⠑⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⡀⠀⎥
⎣⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠑⢄⠀⠀⠈⠻⣦⎦

or when it needs to shrink to screen:


julia> laplacian_2d(13)
169×169 SparseMatrixCSC{Float64, Int64} with 793 stored entries (displaying at 1/2 size):
⎡⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎤
⎢⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠈⠳⣄⠈⠱⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠛⣤⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⢆⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⠀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠀⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠱⣦⡀⠙⢦⡀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⠙⢦⡀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠛⣤⡀⠙⠂⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠳⣄⠈⠻⣦⡀⎥
⎣⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠀⠀⠈⠁⎦

the much bigger example:

old:

julia> laplacian_2d(1000)
1000000×1000000 SparseMatrixCSC{Float64, Int64} with 4996000 stored entries:
⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦

new:

julia> laplacian_2d(1000)
1000000×1000000 SparseMatrixCSC{Float64, Int64} with 4996000 stored entries (displaying at 1/10870 size):
⎡⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎤
⎢⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⠀⠀⎥
⎢⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⡀⠀⎥
⎣⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣦⎦

note that the brackets on the edges were added since then but the aliasing effects had not changed. also, the time it takes to print laplacian_2d(1000) used to be quite bad; now we only check 4996000 indices instead of 1000000000000

rinds=rowvals(parent(S)), cinds=colvals(parent(S)))
# The maximum number of characters we allow to display the matrix
h, w = if get(io, :limit, false)::Bool
get(io, :displysize, displaysize(io)) .- (4, 2)
else
scaleHeight = m
scaleWidth = n
end

# Make sure that the matrix size is big enough to be able to display all
# the corner border characters
if scaleHeight < 8
scaleHeight = 8
end
if scaleWidth < 4
scaleWidth = 4
end

# `brailleGrid` is used to store the needed braille characters for
# the matrix `S`. Each row of the braille pattern to print is stored
# in a column of `brailleGrid`.
brailleGrid = fill(UInt16(10240), (scaleWidth - 1) ÷ 2 + 4, (scaleHeight - 1) ÷ 4 + 1)
brailleGrid[1,:] .= '⎢'
brailleGrid[end-1,:] .= '⎥'
brailleGrid[1,1] = '⎡'
brailleGrid[1,end] = '⎣'
brailleGrid[end-1,1] = '⎤'
brailleGrid[end-1,end] = '⎦'
brailleGrid[end, :] .= '\n'

rvals = rowvals(parent(S))
rowscale = max(1, scaleHeight - 1) / max(1, m - 1)
colscale = max(1, scaleWidth - 1) / max(1, n - 1)
if isa(S, AbstractSparseMatrixCSC)
@inbounds for j in axes(S,2)
# Scale the column index `j` to the best matching column index
# of a matrix of size `scaleHeight × scaleWidth`
sj = round(Int, (j - 1) * colscale + 1)
for x in nzrange(S, j)
# Scale the row index `i` to the best matching row index
# of a matrix of size `scaleHeight × scaleWidth`
si = round(Int, (rvals[x] - 1) * rowscale + 1)

# Given the index pair `(si, sj)` of the scaled matrix,
# calculate the corresponding triple `(k, l, p)` such that the
# element at `(si, sj)` can be found at position `(k, l)` in the
# braille grid `brailleGrid` and corresponds to the 1-dot braille
# character `brailleBlocks[p]`
k = (sj - 1) ÷ 2 + 2
l = (si - 1) ÷ 4 + 1
p = ((sj - 1) % 2) * 4 + ((si - 1) % 4 + 1)

brailleGrid[k, l] |= brailleBlocks[p]
end
typemax(Int)÷4, typemax(Int)÷2
end::Tuple{Int, Int}

warn = false
try
zero(eltype(S))
catch
warn = true
h -= 1
end

# In order to prevent aliasing, we only scale down by full integers.
# Note each character has 4 dots of height and 2 dots of width.
scale = maximum(cld.(size(S), (4h, 2w)))
char_h, char_w = cld.(size(S), (4scale, 2scale))

scale != 1 && print(io, " (displaying at 1/$scale scale)")
println(io, ":")
warn && printstyled(stderr, "WARNING: could not find generic zero for given elements. expect errors and wrong results\n", color=:red)

# Rows of output are cols of `brailleGrid` since julia is column-major
brailleGrid = fill(UInt16(10240), char_w + 3, char_h)
brailleGrid[[1,end-1,end],:] .= ['⎢', '⎥', '\n']
brailleGrid[[1,end-1],[1,end]] .= ['⎡';'⎤';;'⎣';'⎦']
char_h == 1 && (brailleGrid[[1,end-1],1] .= ['[', ']'])

if S isa AbstractSparseMatrixCSC
for cords in zip(rinds, cinds)
row, col = cld.(cords, scale) |>x-> fldmod1.(x, (4, 2))
brailleGrid[col[1]+1, row[1]] |= brailleBlocks[row[2] + 4(col[2]-1)]
end
else
# If `S` is a adjoint or transpose of a sparse matrix we invert the
# roles of the indices `i` and `j`
@inbounds for i = 1:m
si = round(Int, (i - 1) * rowscale + 1)
for x in nzrange(parent(S), i)
sj = round(Int, (rvals[x] - 1) * colscale + 1)
k = (sj - 1) ÷ 2 + 2
l = (si - 1) ÷ 4 + 1
p = ((sj - 1) % 2) * 4 + ((si - 1) % 4 + 1)
brailleGrid[k, l] |= brailleBlocks[p]
end
else # swap rows / cols for adj and transpose
for cords in zip(rinds, cinds)
col, row = cld.(cords, scale) |>x-> fldmod1.(x, (2, 4))
brailleGrid[col[1]+1, row[1]] |= brailleBlocks[row[2] + 4(col[2]-1)]
end
end
foreach(c -> print(io, Char(c)), @view brailleGrid[1:end-1])
end

using Base: alignment
function _show_with_dotted_zeros(io::IO, S::AbstractSparseMatrixCSCInclAdjointAndTranspose, P=parent(S))

@stevengj stevengj Aug 6, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why not call Base.print_matrix after defining Base.replace_in_print_matrix, as we did before? It seems like this duplicates a lot of code from Base?

@rokke-git rokke-git Aug 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I personally feel Base.replace_in_print_matrix was a mistake for a number of reasons, some of which I will try to summarize here:

  • it makes the alignment of zeros the width of a zero, so printing an array with an eltype of Any (or any eltype without a defined zero()) errors (extra benefit: rows with zeros in them no longer depend on the textwidth of zero(eltype))
  • finding the full alignment of the matrix allows us to decide between showing the items in the array or just the spy plot, making it not wrap (or elide) on small screens and not hide items on big screens
  • it allows us to handle cases where multiple items are positioned at the same location, as requested in struct SparseMatrixCSC cannot be constructed correctly with "repeating" entries #618

@rokke-git rokke-git Aug 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

stats note: from a quick count rn, only around ~10 lines out of ~40 here represent functionality that exists in base's print_matrix, which is including lines like for row in axes(S,1) and end (excluding lines which are end puts it closer to ~5). the majority of the logic of this function is specific to sparse arrays, while Base.print_matrix is mostly about eliding rows / cols, functionality we don't need since we have braille plots

rows, cols = S isa Adjoint || S isa Transpose ? (colvals(P), rowvals(P)) : (rowvals(P), colvals(P))
vals = nonzeros(P)

align = [isassigned(vals, ind) ? alignment(io, vals[ind]) : (3, 3) for ind in eachindex(vals)]

colwidths = [maximum.((first,last), Ref(align[findall(==(col), cols)]);init=0) for col in axes(S,2)]
displaysize(io)[2] < sum(sum.(colwidths) .+ 2) && return _show_with_braille_patterns(io, S, rows, cols)

println(io, ":")

try
zero(eltype(S))
catch
printstyled(stderr, "WARNING: could not find generic zero for given elements. expect errors and wrong results\n", color=:red)
end

for row in axes(S,1)
for col in axes(S,2)
index = findall(==(col), cols)
index = index[findall(==(row), rows[index])]
l, r = colwidths[col]
if isempty(index) # no value here, print an aligned dot
l, r = cld(l+r-1, 2) + 1, div(l+r-1, 2) + 1
print(io, " "^l * "⋅" * (col==axes(S,2)[end] ? "" : " "^r))
elseif length(index) == 1 # print the element with 1 space of buffer on each side
l, r = (l+1, r+1) .- align[index[]]
print(io, " "^l)
isassigned(vals, index[]) ? show(io, vals[index[]]) : print(io, "#undef")
col == axes(S,2)[end] || print(io, " "^r)
else # default to summing entries, but print red to warn user that something's wrong
elm = any(!isassigned(vals, ind) for ind in index) ? "#undef" :
try repr(sum(vals[index]); context=io) catch _ "#NaN" end

l, r = if textwidth(elm) <= l+r+2 # give up on alignment, just center item
cld(l+r-textwidth(elm), 2) + 1, fld(l+r-textwidth(elm), 2) + 1
else # new item does not fit in column
elm = "▒"^(l+r)
1, 1
end

printstyled(io, " "^l, elm, color=:red)
col == axes(S,2)[end] || print(io, " "^r)
end
end
row == axes(S,1)[end] || println(io)
end
end

for QT in (:LinAlgLeftQs, :LQPackedQ)
@eval (*)(Q::$QT, B::AbstractSparseMatrixCSC) = Q * Matrix(B)
@eval (*)(Q::$QT, B::AdjOrTrans{<:Any,<:AbstractSparseMatrixCSC}) = Q * copy(B)
Expand Down Expand Up @@ -631,7 +649,7 @@ function _sparse_copyto!(dest::AbstractMatrix, src::AbstractSparseMatrixCSC)
@inbounds for col in axes(src, 2), ptr in nzrange(src, col)
row = rowvals(src)[ptr]
val = nonzeros(src)[ptr]
dest[isrc[row, col]] = val
dest[isrc[row, col]] += val
end
return dest
end
Expand All @@ -655,7 +673,7 @@ function copyto!(dest::AbstractMatrix, Rdest::CartesianIndices{2},
if row in rows
val = nonzeros(src′)[ptr]
I = Rdest[lin[row, col]]
dest[I] = val
dest[I] += val

@stevengj stevengj Aug 6, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why is copyto! being changed to use +=?

@rokke-git rokke-git Aug 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I mentioned it above:

this last push just changed the Matrix(x) conversion call as specified in #618 from dest[isrc[row, col]] = val to dest[isrc[row, col]] += val, which technically adds a get to a potentially hot loop (ie, not display, actual computation).

if this should instead be actually benchmarked I can remove this change; or if we don't care it can remain fixed like this. the dest matrix was already initialized with basically zero(eltype(x)), so Matrix(x) would have already been failing for non-numeric types, no change there

#618 was fixed on the display end by a number of things I added to _show_with_dotted_zeros, so the other ask was just changing 3 =s to +=; if we want to remove those for a different pr that's fine

end
end
return dest
Expand All @@ -680,7 +698,7 @@ function Base.copyto!(A::Array{T}, S::SparseMatrixCSC{<:Number}) where {T<:Numbe
for i in nzrange(S, col)
row = rowval[i]
val = nzval[i]
A[linear_index_col0+row] = val
A[linear_index_col0+row] += val
end
linear_index_col0 += num_rows
end
Expand Down Expand Up @@ -1905,9 +1923,9 @@ julia> A = sparse([1, 2, 3], [1, 2, 3], [1.0, 0.0, 1.0])

julia> dropzeros(A)
3×3 SparseMatrixCSC{Float64, Int64} with 2 stored entries:
1.0 ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ 1.0
1.0 ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ 1.0
```
"""
dropzeros(A::AbstractSparseMatrixCSC) = dropzeros!(copy(A))
Expand Down Expand Up @@ -2077,7 +2095,7 @@ argument specifies a random number generator, see [Random Numbers](@ref).
```jldoctest; setup = :(using Random; Random.seed!(0))
julia> sprandn(2, 2, 0.75)
2×2 SparseMatrixCSC{Float64, Int64} with 3 stored entries:
-1.20577 ⋅
-1.20577 ⋅
0.311817 -0.234641
```
"""
Expand All @@ -2104,9 +2122,9 @@ specified.
```jldoctest
julia> spzeros(3, 3)
3×3 SparseMatrixCSC{Float64, Int64} with 0 stored entries:
⋅ ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ ⋅
⋅ ⋅ ⋅

julia> spzeros(Float32, 4)
4-element SparseVector{Float32, Int64} with 0 stored entries
Expand Down
Loading
Loading