CanonicalIndexError with TrendFilter

[ptaitAtMcMaster]

Hi,
I wanted to raise awareness of an error I have been getting with trend filtering. It looks like it could be an edge case of some kind. Here is a reproducible example:

`using Lasso, LinearAlgebra, Random
rng = Xoshiro(45);
n = 100;
y = rand(rng, n) .- 0.5;

f_fl = fit(FusedLasso, y, 0.5);
f_tf = fit(TrendFilter, y, 2, 0.5);
`

The trend filter fit results in the following error:

TrendFilter{Float64, Lasso.FusedLassoMod.NormalCoefs{Float64}, SubArray{Float64, 1, Vector{Float64}, Tuple{UnitRange{Int64}}, true}}([ERROR: CanonicalIndexError: getindex not defined for Lasso.TrendFiltering.DifferenceMatrix{Float64}
Stacktrace:
[1] error_if_canonical_getindex(::IndexCartesian, ::Lasso.TrendFiltering.DifferenceMatrix{Float64}, ::Int64, ::Int64)
@ Base ./abstractarray.jl:1327
[2] getindex
@ ./abstractarray.jl:1311 [inlined]
[3] isassigned(::Lasso.TrendFiltering.DifferenceMatrix{Float64}, ::Int64, ::Int64)
@ Base ./multidimensional.jl:1612
[4] _show_nonempty(io::IOContext{Base.TTY}, X::AbstractMatrix, prefix::String, drop_brackets::Bool, axs::Tuple{Base.OneTo{Int64}, Base.OneTo{Int64}})
@ Base ./arrayshow.jl:441
[5] _show_nonempty(io::IOContext{Base.TTY}, X::Lasso.TrendFiltering.DifferenceMatrix{Float64}, prefix::String)
@ Base ./arrayshow.jl:413
[6] show(io::IOContext{Base.TTY}, X::Lasso.TrendFiltering.DifferenceMatrix{Float64})
@ Base ./arrayshow.jl:491
[7] _show_default(io::IOContext{Base.TTY}, x::Any)
@ Base ./show.jl:496
[8] show_default
@ ./show.jl:479 [inlined]
[9] show
@ ./show.jl:474 [inlined]
[10] show(io::IOContext{Base.TTY}, ::MIME{Symbol("text/plain")}, x::TrendFilter{Float64, Lasso.FusedLassoMod.NormalCoefs{Float64}, SubArray{Float64, 1, Vector{Float64}, Tuple{UnitRange{Int64}}, true}})
@ Base.Multimedia ./multimedia.jl:47
[11] (::REPL.var"#68#69"{REPL.REPLDisplay{REPL.LineEditREPL}, MIME{Symbol("text/plain")}, Base.RefValue{Any}})(io::Any)
@ REPL /snap/julia/113/share/julia/stdlib/v1.11/REPL/src/REPL.jl:367
[12] with_repl_linfo(f::Any, repl::REPL.LineEditREPL)
@ REPL /snap/julia/113/share/julia/stdlib/v1.11/REPL/src/REPL.jl:661
[13] display(d::REPL.REPLDisplay, mime::MIME{Symbol("text/plain")}, x::Any)
@ REPL /snap/julia/113/share/julia/stdlib/v1.11/REPL/src/REPL.jl:353
[14] display(d::REPL.REPLDisplay, x::Any)
@ REPL /snap/julia/113/share/julia/stdlib/v1.11/REPL/src/REPL.jl:372
[15] display(x::Any)
@ Base.Multimedia ./multimedia.jl:340
[16] #invokelatest#2
@ ./essentials.jl:1055 [inlined]
[17] invokelatest
@ ./essentials.jl:1052 [inlined]

The error is generated regardless of the value of the order parameter. I have attached a plot of the data and the fused lasso coefficients.

In the next example, trend filtering dose work:

`fs = 250.0 # Sampling frequency in Hz
t = 0:1/fs:10 # 10 seconds of data
signal = sin.(2π1.0.t) + 0.5sin.(2π50.0.*t)

lambda = norm(signal, Inf)/2
fit_fl = fit(FusedLasso, signal, lambda);
fit_tf = fit(TrendFilter, signal, 2, lambda);
`
A plot of the results is attached.

I have been using trend filtering often and run into this indexing error ~1/4 of the time.
Cheers
Peter

(@v1.11) pkg> st Lasso
Status ~/.julia/environments/v1.11/Project.toml
[b4fcebef] Lasso v0.7.2

julia> versioninfo()
Julia Version 1.11.1
Build Info:
Official https://julialang.org/ release
Platform Info:
OS: Linux (x86_64-linux-gnu)
LLVM: libLLVM-16.0.6 (ORCJIT, goldmont)