v0.2.0

Release Notes

• Implements single-node parallelism (thread-based) for both training and simulation steps
• Applies L1 dominance cut selection while locking the future cost function on each thread
• Number of parallel threads is capped either on the logical core count or number of forward passes / simulated scenarios
• Generalizes the underlying data structure for the SDDP algorithm from a vector to a graph
• Creates submodules for scenario generation, risk measure, state, and stochastic process
• Generalizes recourse input for defining inflow and load scenarios, with branchings per node
• Uses trait objects for dynamic risk measure, state and stochastic process definition

Install powers-rs 0.2.0

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/rjmalves/powers/releases/download/v0.2.0/powers-rs-installer.sh | sh

Download powers-rs 0.2.0

File	Platform	Checksum
powers-rs-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
powers-rs-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
powers-rs-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.1.1

Release Notes

• Better handles memory allocation during the SAA generation step in simulation
• Fixes doctests from renaming the crate

Install powers-rs 0.1.1

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/rjmalves/powers/releases/download/v0.1.1/powers-rs-installer.sh | sh

Download powers-rs 0.1.1

File	Platform	Checksum
powers-rs-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
powers-rs-x86_64-apple-darwin.tar.xz	Intel macOS	checksum
powers-rs-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

v0.1.0

• Initial release
• Solves a simples hydrothermal dispatch problem with hydro storages as state variables using SDDP
• Inflows are sampled from LogNormal distributions, considered the same for all stages
• Loads are constant, given by the user in the input data
• Implements a custom interface to the HiGHS solver
• Contains cut selection and basis reuse for improving performance
• Performs simulation by sampling from the same distributions used for training