diff --git a/Project.toml b/Project.toml
new file mode 100644
index 0000000..d2a5219
--- /dev/null
+++ b/Project.toml
@@ -0,0 +1,23 @@
+name = "HALeqO"
+uuid = "8c1b84d6-365e-44ef-b36c-195f7cd27d2c"
+authors = ["Alberto De Marchi <aldmarchi@gmail.com>"]
+version = "0.2.1"
+
+[deps]
+CSV = "336ed68f-0bac-5ca0-87d4-7b16caf5d00b"
+CUTEst = "1b53aba6-35b6-5f92-a507-53c67d53f819"
+DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
+KNITRO = "67920dd8-b58e-52a8-8622-53c4cffbe346"
+LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
+NCL = "af6b844e-5feb-47ff-8350-2ae20a94a0cf"
+NLPModels = "a4795742-8479-5a88-8948-cc11e1c8c1a6"
+NLPModelsIpopt = "f4238b75-b362-5c4c-b852-0801c9a21d71"
+NLPModelsKnitro = "bec4dd0d-7755-52d5-9a02-22f0ffc7efcb"
+Percival = "01435c0c-c90d-11e9-3788-63660f8fbccc"
+Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80"
+PositiveFactorizations = "85a6dd25-e78a-55b7-8502-1745935b8125"
+Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
+QDLDL = "bfc457fd-c171-5ab7-bd9e-d5dbfc242d63"
+SolverBenchmark = "581a75fa-a23a-52d0-a590-d6201de2218a"
+SolverCore = "ff4d7338-4cf1-434d-91df-b86cb86fb843"
+SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
diff --git a/README.md b/README.md
index 676594c..eec480d 100644
--- a/README.md
+++ b/README.md
@@ -22,15 +22,34 @@ HALeqO.jl uses the free [QDLDL.jl](https://github.com/osqp/QDLDL.jl) routines as
 
 ### Citing
 
-If you are using HALeqO for your work, we encourage you to
+If you find this code useful, you can [cite](CITATION.bib) the related [paper](https://doi.org/10.1109/CDC45484.2021.9683199) as
+
+    @inproceedings{demarchi2021augmented,
+        author      = {De~Marchi, Alberto},
+        title       = {Augmented {L}agrangian methods as dynamical systems for constrained optimization},
+        year        = {2021},
+        month       = {12},
+        pages       = {6533--6538},
+        booktitle   = {2021 60th {IEEE} {C}onference on {D}ecision and {C}ontrol ({CDC})},
+        doi         = {10.1109/CDC45484.2021.9683199},
+    }
 
-* [Cite](CITATION.bib) the related [paper](https://doi.org/10.1109/CDC45484.2021.9683199),
-* Put a star on this repository.
+### Benchmarks
 
-### Bug reports and support
+We compared HALeqO against [Ipopt](https://coin-or.github.io/Ipopt/), via the wrapper provided by [NLPModelsIpopt](https://github.com/JuliaSmoothOptimizers/NLPModelsIpopt.jl), and [NCL.jl](https://github.com/JuliaSmoothOptimizers/NCL.jl) invoking Ipopt.
+There is also [Percival](https://github.com/JuliaSmoothOptimizers/Percival.jl) now.
+See `run_benchmarks.jl` in the `tests` folder.
 
-Please report any issues via the [issue tracker](https://github.com/aldma/HALeqO.jl/issues). All types of issues are welcome including bug reports, typos, feature requests and so on.
+To use the provided test codes (originally compatible with Julia version 1.8.5, tested on Linux x86_64, August 2021):
 
-### Benchmarks
+* start Julia from this directory with `julia --project=.` (or with the relative path to the `HALeqO` directory from somewhere else);
+
+* do `]instantiate` to download all dependencies (only required the first time) and go back to standard Julia prompt with backspace afterwards;
+
+* load the package with `using HALeqO`
+
+To run the CUTEst benchmarks:
+
+* do `include("tests/run_benchmarks.jl")` to invoke the solvers, generate the results, print statistics and save figures.
 
-We compared HALeqO against [IPOPT](https://coin-or.github.io/Ipopt/), via the wrapper provided by [NLPModelsIpopt](https://github.com/JuliaSmoothOptimizers/NLPModelsIpopt.jl), and [NCL.jl](https://github.com/JuliaSmoothOptimizers/NCL.jl) invoking IPOPT. See `run_benchmarks.jl` in the `tests` folder.
+Options for benchmarking (e.g., max time and tolerance) can be modified at the beginning of the `tests/run_benchmarks.jl` file.
\ No newline at end of file
diff --git a/tests/run_benchmarks.jl b/tests/run_benchmarks.jl
index 9a95342..575290e 100644
--- a/tests/run_benchmarks.jl
+++ b/tests/run_benchmarks.jl
@@ -1,7 +1,7 @@
 # load solvers
 using HALeqO
 using NLPModelsIpopt
-using NCL # from https://github.com/JuliaSmoothOptimizers/NCL.jl, using IPOPT
+#using NCL # from https://github.com/JuliaSmoothOptimizers/NCL.jl, using Ipopt
 using Percival
 
 # load problems
@@ -29,7 +29,7 @@ MAXITER = 3000 # default 3000
 problems = (CUTEstModel(probname) for probname in probnames)
 solvers = Dict{Symbol,Function}(
     :HALeqO => prob -> haleqo(prob; tol = TOL, max_iter = MAXITER),
-    :NCL => prob -> NCLSolve(prob, opt_tol = TOL, feas_tol = TOL, max_iter = MAXITER),
+    #:NCL => prob -> NCLSolve(prob, opt_tol = TOL, feas_tol = TOL, max_iter = MAXITER),
     :IPOPT => prob -> ipopt(prob; tol = TOL, print_level = 0, max_iter = MAXITER),
     :Percival => prob -> percival(prob; atol = TOL, rtol = 0.0, ctol = TOL, max_iter = MAXITER),
 )
@@ -68,3 +68,6 @@ tprof = time_profile(stats, cost)
 for solver ∈ keys(stats)
     CSV.write("data/" * filename * "_" * String(solver) * ".csv", stats[solver], header = true)
 end
+
+savefig(pprof,"data/" * filename * "_" * "perfprof" * ".pdf")
+savefig(tprof,"data/" * filename * "_" * "timeprof" * ".pdf")
\ No newline at end of file
diff --git a/tests/run_tests.jl b/tests/run_tests.jl
index bf12666..554885b 100644
--- a/tests/run_tests.jl
+++ b/tests/run_tests.jl
@@ -15,12 +15,16 @@ probnames = CUTEst.select(
     only_equ_con = true,
 )
 
+#nlp = CUTEstModel("BOXBOD")
+#out = haleqo(nlp, tol=1e-6, max_iter=1000)
+#finalize(nlp)
+
 # setup testing
 problems = (CUTEstModel(probname) for probname in probnames)
 solver = prob -> haleqo(prob)
 
 # run solver!
-stats = solve_problems(solver, problems)
+stats = solve_problems(solver, "haleqo", problems)
 
 # get statistics
 @info "HALeqO statuses" count_unique(stats.status)