Fixed deprecation (#98)

* Fixed deprecation

* Put through latest juliaformatter

* Testing on 1.8

* Version bump

Author: jlperla

.github/workflows/CI.yml

@@ -14,7 +14,7 @@ jobs:
         group:
           - All
         version:
-          - "1.7"
+          - "1.8"
     steps:
       - uses: actions/checkout@v3
       - uses: julia-actions/setup-julia@v1

Project.toml

@@ -1,7 +1,7 @@
 name = "DifferenceEquations"
 uuid = "e0ca9c66-1f9e-11ec-127a-1304ce62169c"
 authors = ["various contributors"]
-version = "0.5.1"
+version = "0.5.2"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
@@ -23,4 +23,4 @@ PDMats = "0.11"
 RecursiveArrayTools = "2.25"
 SciMLBase = "1.39"
 UnPack = "1"
-julia = "1.6"
+julia = "1.8"

src/algorithms/linear.jl

@@ -216,7 +216,7 @@ function DiffEqBase.__solve(prob::LinearStateSpaceProblem, alg::KalmanFilter, ar
             lmul!(0.5, V[t].mat)
 
             copy!(V[t].chol.factors, V[t].mat) # copy over to the factors for the cholesky and do in place
-            cholesky!(V[t].chol.factors, Val(false); check = false) # inplace uses V_t with cholesky.  Now V[t]'s chol is upper-triangular        
+            cholesky!(V[t].chol.factors, NoPivot(); check = false) # inplace uses V_t with cholesky.  Now V[t]'s chol is upper-triangular        
             innovation[t] .= prob.observables[:, t - 1] - z[t]
             loglik += logpdf(MvNormal(V[t]), innovation[t])  # no allocations since V[t] is a PDMat
 
@@ -334,7 +334,7 @@ function ChainRulesCore.rrule(::typeof(DiffEqBase.solve), prob::LinearStateSpace
             lmul!(0.5, V[t].mat)
 
             copy!(V[t].chol.factors, V[t].mat) # copy over to the factors for the cholesky and do in place
-            cholesky!(V[t].chol.factors, Val(false); check = false) # inplace uses V_t with cholesky.  Now V[t]'s chol is upper-triangular        
+            cholesky!(V[t].chol.factors, NoPivot(); check = false) # inplace uses V_t with cholesky.  Now V[t]'s chol is upper-triangular        
             innovation[t] .= prob.observables[:, t - 1] - z[t]
             loglik += logpdf(MvNormal(V[t]), innovation[t])  # no allocations since V[t] is a PDMat
 

src/utilities.jl

@@ -135,7 +135,6 @@ maybe_zero(A::Nothing) = nothing
 maybe_zero(A::AbstractArray, i::Int64) = zero(A[i])
 maybe_zero(A::Nothing, i) = nothing
 
-
 # old quad and adjoint replaced by inplace accumulation versions.
 # function quad(A::AbstractArray{<:Number,3}, x)
 #     return map(j -> dot(x, view(A, j, :, :), x), 1:size(A, 1))

test/kalman_likelihood.jl

@@ -263,7 +263,8 @@ R = [0.01 0.0 0.0 0.0;
     loglik = solve_manual_cov_lik(A_kalman, B_kalman, C_kalman, u0_mean, u0_var_vech,
                                   observables_kalman,
                                   R, [0, T])
-    sol = solve_kalman_cov(A_kalman, B_kalman, C_kalman, u0_mean, u0_var_vech, observables_kalman,
+    sol = solve_kalman_cov(A_kalman, B_kalman, C_kalman, u0_mean, u0_var_vech,
+                           observables_kalman,
                            R)
     @test sol.logpdf ≈ loglik
 end