v1.10.1

SDDP v1.10.1

Diff since v1.10.0

Merged pull requests:

• Update air conditioning example (#801) (@Thuener)
• Fix thread safety of RegularizedForwardPass (#806) (@odow)
• Drop exception for regularized forward pass in algorithm.jl (#807) (@FSchmidtDIW)
• Fix thread safety of AlternativeForwardPass (#808) (@odow)
• Prep for v1.10.1 (#809) (@odow)

Closed issues:

• Recording primal/dual iterates (#791)
• Restructure the examples in the documentation (#798)

v1.10.0

SDDP v1.10.0

Diff since v1.9.0

Merged pull requests:

• Add tutorial Example: inventory management (#795) (@odow)
• Fix cut sharing in a graph with zero-probability arcs (#797) (@odow)
• Add stochastic lead times example to docs (#800) (@Thuener)
• Add root_node_risk_measure kwarg to SDDP.train (#804) (@odow)
• Prep for v1.10 (#805) (@odow)

Closed issues:

• Help with Visualizing Cuts (#794)
• Cut sharing with Markov policy graph (#796)
• Cycle policy condition on t (#799)
• Periodical SDDP (#802)
• Value of bound output during training with risk aversion and first stage uncertainty (#803)

v1.9.0

SDDP v1.9.0

Diff since v1.8.1

Merged pull requests:

• Skip threading tests if running in serial (#770) (@odow)
• Fix here-and-now typo (#771) (@joaquimg)
• Update simulate_using_a_different_sampling_scheme.md (#772) (@Thuener)
• Fix failing doctest because of JuMP's printing (#773) (@odow)
• Change BanditDuality to use a prior and softmax to randomize arms (#779) (@odow)
• Cut selection flags for write and read cuts functions (#781) (@Thuener)
• Add simple cut selection flags test (#784) (@Thuener)
• Improve error for lagged state variables in MSPFormat (#786) (@odow)
• Support publication plot with replications of different lengths (#788) (@odow)
• Use disable_sigint to allow interrupts only between iterations (#789) (@odow)
• Add a callback to modify the subproblem on numerical difficulty (#790) (@odow)
• Prep for v1.9.0 (#793) (@odow)

Closed issues:

• Pass non-SDDP related decision rules (#696)
• Central Path Cutting Plane (#755)
• Manually stopping training SIGINT Gurobi (#767)
• Add a callback to let the user modify a subproblem when numerical difficulties found (#774)
• Correct Modeling Technique for Capacity Expansion (#775)
• Question regarding calculate_bound with Bin variables. (#776)
• Getting out of local optimal SDDiP (#777)
• Performance Improvement on read/write cuts (#778)
• the solver encountered numerical issues (#782)
• [MSPFormat] Can't read Problem Instance 10 (Inventory Problem with Lead Time) from MSPLib-Library (#785)
• Counting the number of cuts (#787)
• Recording primal/dual iterates (#791)

v1.8.1

SDDP v1.8.1

Diff since v1.8.0

Merged pull requests:

• DNM: Fix various Threading() bugs (#761) (@odow)
• Update copyright year (#762) (@odow)
• Fix deprecation warning for sort(::Dict) (#763) (@odow)
• Add separate_kwargs_with_semicolon = true to .JuliaFormatter.toml (#764) (@odow)
• Prep for v1.8.1 (#769) (@odow)

Closed issues:

• TTFX (#685)
• Which would be the correct way to model the uncertainty from historical data? (#765)
• Performance enhance (Direct Model) (#768)

v1.8.0

SDDP v1.8.0

Diff since v1.7.0

Merged pull requests:

• [docs] clarify monotonicity of bound in first_steps.jl (#747) (@odow)
• Add Threaded parallel scheme (#758) (@odow)
• Misc maintenance fixes and updates (#759) (@odow)
• Prep for v1.8.0 (#760) (@odow)

Closed issues:

• Multithreading parallelism (#599)
• Multiplying random variables will result in an error. This is related to nonlinear programming. (#739)
• Multiply two variables (#744)
• Does SDDP.jl support modeling with endogenous uncertainty and custom decision rules? (#745)
• Clarify documentation of non monotonic bound (#746)
• Cuts with really small coefficents (#748)
• Can I define a constraint that correlates state variables from multiple nodes? (#749)
• Does SDDP.jl support polytopic uncertainty? (#750)
• UnicyclicGraph with Integer variables (#751)
• Runtime specified variable naming (#752)
• Some questions on parameterizing the subproblem and numerical issues (#753)
• ERROR: TypeError: in keyword argument incoming_state, expected Dict{Symbol, Float64}, got a value of type Dict{Symbol, Any} (#754)
• Some questions about model debugging (#756)
• How to save a trained model? (#757)

v1.7.0

SDDP v1.7.0

Diff since v1.6.7

Merged pull requests:

• Add programatic logo constructor (#730) (@odow)
• Add a test for non-finite data in publication_plot (#738) (@odow)
• Add sample_backward_noise_terms_with_state (#742) (@arthur-brigatto)
• Prep for v1.7.0 (#743) (@odow)

Closed issues:

• First-stage regularization (#697)
• Out of Sample Simulation for the infinite horizon problem. (#728)
• trouble with sum (#731)
• Add multi-dimensional random variables (#732)
• Question Regarding State-Dependent Paramether Updates in SDDP.jl (#733)
• Error when trying to train a policy (#734)
• Error when using nonlinear function of state.in (#735)
• Cyclic Markov Policy Graph (#736)
• PublicationPlots do not check for non-finite data (#737)
• Best Practices for custom cut selection heuristic (#741)

v1.6.7

SDDP v1.6.7

Diff since v1.6.6

Merged pull requests:

• [docs] clarify simulate using a different sampling scheme (#692) (@odow)
• Add kelley example to example newsvendor (#694) (@odow)
• Fix non-constant state dimension in MSPFormat reader (#695) (@odow)
• Update to StochOptFormat v1.0 (#705) (@odow)
• [docs] update choose_a_stopping_rule.md (#706) (@odow)
• Add OuterApproximation algorithm to LocalImprovementSearch (#709) (@odow)
• Fix SimulatorSamplingScheme for deterministic nodes and update docs (#710) (@odow)
• Fix line search in BFGS (#711) (@odow)
• Fix doc build (#716) (@odow)
• Update .gitignore (#717) (@odow)
• [docs] fix typo in first_steps.jl (#719) (@odow)
• Add code for MDP paper (#720) (@odow)
• [docs] replace chheese_producer.pdf (#721) (@odow)
• Add Google analytics (#723) (@odow)
• Support NEARLY_FEASIBLE_POINT (#726) (@odow)
• [docs] fix Julia 1.10 printing (#727) (@odow)
• Prep for v1.6.7 (#729) (@odow)

Closed issues:

• Undefined variable in out-of-sample-sampling documentation (#691)
• MSPFormat: dimension of state variable can change between nodes (#693)
• Impact of the number of stages on solving time (#698)
• Free up resources in rolling horizon (#699)
• Is it mandatory to create state variables for all nodes? (#700)
• Inconsistency (#701)
• Rolling Horizon (#702)
• Historical Simulation (#707)
• Compute dual variables with continous state variables and mixed control variables (#708)
• Sources regarding Integrality (#712)
• multiple lead time (#713)
• Regarding file saving issues of SDDP model (#714)
• Probability of Noise depending on Control Variable (#715)
• Typo (#718)
• Markov graph + Risk aversion (#722)
• Is it applicable when the local problems of some stages are mixed interger linear programming? (#724)
• Support NEARLY_FEASIBLE_POINT (#725)

v1.6.6

SDDP v1.6.6

Diff since v1.6.5

Merged pull requests:

• Add warning to Statistical stopping rule (#687) (@odow)
• [docs] add more L-shaped theory (#689) (@odow)
• Prep for v1.6.6 (#690) (@odow)

Closed issues:

• Here and now solution (#686)
• First Stage Stopping Rule Issue (#688)

v1.6.5

SDDP v1.6.5

Diff since v1.6.4

Merged pull requests:

• [docs] update docs/src/tutorial/example_reservoir.jl (#677) (@odow)
• [docs] update docs/src/tutorial/example_newsvendor.jl (#678) (@odow)
• [docs] re-add GITHUB_TOKEN (#679) (@odow)
• Fix duplicate nodes in MarkovianGraph (#681) (@odow)
• [docs] remove Binder links (#682) (@odow)
• [docs] add link to pglib_opf_cases5_pjm.m (#683) (@odow)
• Prep for v1.6.5 (#684) (@odow)

Closed issues:

• Accessing Optimal Decision (#675)
• the notation lower_bound (#676)
• Duplicate support in MarkovianGraph (#680)

v1.6.4

SDDP v1.6.4

Diff since v1.6.3

Merged pull requests:

• [docs] update Two-stage newsvendor example (#658) (@odow)
• Throw error for invalid log_frequency values (#665) (@odow)
• [docs] remove example_reservoir.csv (#666) (@odow)
• Create static.yml for deploying GitHub pages (#667) (@odow)
• Update README.md links (#668) (@odow)
• [docs] update to Documenter@1 (#669) (@odow)
• Use DOCUMENTER_KEY to deploy docs (#670) (@odow)
• [docs] update index.md (#671) (@odow)
• Fix support for objective_sense in deterministic_equivalent (#673) (@odow)
• Prep for v1.6.4 (#674) (@odow)

Closed issues:

• MSPFormat issues (#603)
• Provide more flexible options for stopping rules (#664)
• FEASIBILITY_SENSE returned by SDDP.deterministic_equivalent() (#672)