Merge pull request #46 from JuliaDiffEq/revert_false_asserts

Minor tweaks in the asserts and new release

Author: AlCap23

src/dmdc.jl

@@ -7,7 +7,8 @@ mutable struct DMDc{K, B, Q, P} <: abstractKoopmanOperator
 end
 
 
-function DMDc(X::AbstractArray, Y::AbstractArray, U::AbstractArray; B::AbstractArray = [], dt::Float64 = 0.0)
+function DMDc(X::AbstractArray, Y::AbstractArray, U::AbstractArray; B::AbstractArray = [], dt::T = 0.0) where T <: Real
+    @assert dt >= zero(dt)
     @assert all(size(X) .== size(Y))
     @assert size(X)[2] .== size(U)[2]
 
@@ -44,7 +45,7 @@ function DMDc(X::AbstractArray, Y::AbstractArray, U::AbstractArray; B::AbstractA
 end
 
 
-function DMDc(X::AbstractArray, U::AbstractArray; B::AbstractArray = [], dt::Float64 = 0.0)
+function DMDc(X::AbstractArray, U::AbstractArray; B::AbstractArray = [], dt::T = 0.0) where T <: Real
     @assert size(X)[2]-1 == size(U)[2] "Provide consistent input data."
     return DMDc(X[:, 1:end-1], X[:, 2:end], U, B = B, dt = dt)
 end

src/extended_dmd.jl

@@ -12,11 +12,11 @@ LinearAlgebra.eigen(m::ExtendedDMD) = eigen(m.koopman)
 LinearAlgebra.eigvals(m::ExtendedDMD) = eigvals(m.koopman)
 LinearAlgebra.eigvecs(m::ExtendedDMD) = eigvecs(m.koopman)
 
-function ExtendedDMD(X::AbstractArray, Ψ::abstractBasis; p::AbstractArray = [],  B::AbstractArray = reshape([], 0,0), dt::T = 1.0)  where T <: Real
+function ExtendedDMD(X::AbstractArray, Ψ::abstractBasis; p::AbstractArray = [],  B::AbstractArray = reshape([], 0,0), dt::T = 0.0)  where T <: Real
     return ExtendedDMD(X[:, 1:end-1], X[:, 2:end], Ψ, p = p, B = B, dt = dt)
 end
 
-function ExtendedDMD(X::AbstractArray, Y::AbstractArray, Ψ::abstractBasis; p::AbstractArray = [], B::AbstractArray = reshape([], 0,0), dt::T = 1.0) where T <: Real
+function ExtendedDMD(X::AbstractArray, Y::AbstractArray, Ψ::abstractBasis; p::AbstractArray = [], B::AbstractArray = reshape([], 0,0), dt::T = 0.0) where T <: Real
     @assert dt >= zero(typeof(dt)) "Provide positive dt"
     @assert size(X)[2] .== size(Y)[2] "Provide consistent dimensions for data"
     @assert size(Y)[1] .<= size(Y)[2] "Provide consistent dimensions for data"

src/isindy.jl

@@ -5,6 +5,7 @@
 
 # TODO preallocation
 function ISInDy(X::AbstractArray, Ẋ::AbstractArray, Ψ::Basis; maxiter::Int64 = 10, rtol::Float64 = 0.99, p::AbstractArray = [], opt::T = ADM()) where T <: DataDrivenDiffEq.Optimise.AbstractSubspaceOptimiser
+    @assert size(X)[end] == size(Ẋ)[end]
     nb = length(Ψ.basis)
 
     # Compute the library and the corresponding nullspace

src/sindy.jl

@@ -35,7 +35,7 @@ end
 
 # Returns a basis for the differential state
 function SInDy(X::AbstractArray, Ẋ::AbstractArray, Ψ::Basis; p::AbstractArray = [], maxiter::Int64 = 10, opt::T = Optimise.STRRidge()) where T <: Optimise.AbstractOptimiser
-    @assert size(X) == size(Ẋ)
+    @assert size(X)[end] == size(Ẋ)[end]
     nx, nm = size(X)
 
     Ξ = zeros(eltype(X), length(Ψ), nx)
@@ -50,7 +50,7 @@ end
 
 # Returns an array of basis for all values of lambda
 function SInDy(X::AbstractArray, Ẋ::AbstractArray, Ψ::Basis, thresholds::AbstractArray ; p::AbstractArray = [], maxiter::Int64 = 10, opt::T = Optimise.STRRidge()) where T <: Optimise.AbstractOptimiser
-    @assert size(X) == size(Ẋ)
+    @assert size(X)[end] == size(Ẋ)[end]
     nx, nm = size(X)
 
     θ = Ψ(X, p = p)