Remove all Evaluators from ExaPF

.github/workflows/action.yml

@@ -22,7 +22,6 @@ jobs:
       - uses: julia-actions/setup-julia@latest
         with:
           version: ${{ matrix.julia-version }}
-      - run: julia --project deps/deps.jl
       - uses: julia-actions/julia-buildpkg@latest
       - uses: julia-actions/julia-runtest@latest
 
@@ -44,6 +43,5 @@ jobs:
         with:
           version: ${{ matrix.julia-version }}
           arch: ${{ matrix.julia-arch }}
-      - run: julia --project deps/deps.jl
       - uses: julia-actions/julia-buildpkg@latest
       - uses: julia-actions/julia-runtest@latest

Project.toml

@@ -12,7 +12,6 @@ KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
 Krylov = "ba0b0d4f-ebba-5204-a429-3ac8c609bfb7"
 LightGraphs = "093fc24a-ae57-5d10-9952-331d41423f4d"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
-MathOptInterface = "b8f27783-ece8-5eb3-8dc8-9495eed66fee"
 Metis = "2679e427-3c69-5b7f-982b-ece356f1e94b"
 Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
@@ -27,7 +26,6 @@ ForwardDiff = "0.10"
 KernelAbstractions = "0.6, 0.7"
 Krylov = "~0.7.3"
 LightGraphs = "1.3"
-MathOptInterface = "0.9"
 Metis = "1"
 SparseDiffTools = "1"
 TimerOutputs = "0.5"

README.md

@@ -1,6 +1,6 @@
 # ExaPF
 
-[![][docs-latest-img]][docs-latest-url] ![CI](https://github.com/exanauts/ExaPF.jl/workflows/Run%20tests/badge.svg?branch=master) 
+[![][docs-latest-img]][docs-latest-url] ![CI](https://github.com/exanauts/ExaPF.jl/workflows/Run%20tests/badge.svg?branch=master)
 
 [docs-latest-img]: https://img.shields.io/badge/docs-latest-blue.svg
 [docs-latest-url]: https://exanauts.github.io/ExaPF.jl/
@@ -49,49 +49,6 @@ Iteration 4. Residual norm: 3.9111e-12.
 ExaPF.ConvergenceStatus(true, 4, 3.911102241031109e-12, 0)
 ```
 
-### How to solve the optimal power flow in the reduced space?
-
-ExaPF implements a wrapper to [MathOptInterface](https://github.com/jump-dev/MathOptInterface.jl)
-that allows to solve the optimal power flow problem directly in the reduced space
-induced by the power flow equations:
-
-```julia
-julia> case = "case57.m"
-# Instantiate a ReducedSpaceEvaluator object
-julia> nlp = ExaPF.ReducedSpaceEvaluator(datafile)
-# MOI optimizer
-julia> optimizer = Ipopt.Optimizer()
-# Use LBFGS algorithm, as reduced Hessian is not available by default!
-julia> MOI.set(optimizer, MOI.RawParameter("hessian_approximation"), "limited-memory")
-julia> MOI.set(optimizer, MOI.RawParameter("tol"), 1e-4)
-julia> solution = ExaPF.optimize!(optimizer, nlp)
-Total number of variables............................:       10
-                     variables with only lower bounds:        0
-                variables with lower and upper bounds:       10
-                     variables with only upper bounds:        0
-Total number of equality constraints.................:        0
-Total number of inequality constraints...............:       58
-        inequality constraints with only lower bounds:        0
-   inequality constraints with lower and upper bounds:       58
-        inequality constraints with only upper bounds:        0
-
-
-Number of Iterations....: 9
-
-                                   (scaled)                 (unscaled)
-Objective...............:   1.9630480251946040e+03    3.7589338203438238e+04
-Dual infeasibility......:   2.5545890554923290e-05    4.8916435433709606e-04
-Constraint violation....:   4.7695181137896725e-13    4.7695181137896725e-13
-Complementarity.........:   1.0270912626531211e-11    1.9667211572084318e-10
-Overall NLP error.......:   2.5545890554923290e-05    4.8916435433709606e-04
-
-[...]
-Total CPU secs in IPOPT (w/o function evaluations)   =      0.049
-Total CPU secs in NLP function evaluations           =      0.023
-
-EXIT: Optimal Solution Found.
-```
-
 ## Development
 
 We welcome any contribution to ExaPF! Bug fixes or feature requests

benchmark/benchmarks.jl

@@ -53,15 +53,18 @@ function run_benchmark(datafile, device, linsolver)
 
     algo = linsolver(J; P=precond)
     powerflow_solver = NewtonRaphson(tol=ntol)
-    nlp = ExaPF.ReducedSpaceEvaluator(polar;
-                                      powerflow_solver=powerflow_solver)
-    nlp.linear_solver = algo
-    convergence = ExaPF.update!(nlp, u0)
-    ExaPF.reset!(nlp)
-    convergence = ExaPF.update!(nlp, u0)
-    ExaPF.reset!(nlp)
+
+    # Init variables
+    buffer = get(polar, ExaPF.PhysicalState())
+    jx = AutoDiff.Jacobian(polar, ExaPF.power_balance, State())
+
+    # Warmstart
+    ExaPF.init_buffer!(polar, buffer)
+    ExaPF.powerflow(polar, jx, buffer, powerflow_solver; linear_solver=algo)
+
     TimerOutputs.reset_timer!(ExaPF.TIMER)
-    convergence = ExaPF.update!(nlp, u0)
+    ExaPF.init_buffer!(polar, buffer)
+    convergence = ExaPF.powerflow(polar, jx, buffer, powerflow_solver; linear_solver=algo)
 
     # Make sure we are converged
     @assert(convergence.has_converged)