Terminology word list

LP

Linear programming (LP), a mathematical method for finding the best possible outcome, from a linear model.

MIP

Mixed integer problem (MIP), optimization technique for solving problems with both continuous (real-valued) and discrete (integer or binary) decision variables.

SDP

Stochastic dynamic programming (SDP), a mathematical optimization technique for sequential decision-making problems where outcomes are affected by random uncertainty over time, aiming to maximize expected future return.

SDDP

Stochastic dual dynamic programming (SDDP), similar in concept to SDP, but uses sampling to avoid curse of dimensionality.

curse of dimensionality

When an increase in problem size exponentially increases solution time (or solution space).

MPI

Message parsing interface (MPI) - framework to communicate across computer processes, used to parallelize work over distributed memory. Can be used across CPUs, computer nodes or computers. Gold standard for distributed memory parallelization.

deterministic

Having a clear cause-effect relatioship.

stochastic

Opposed to deterministic, there is an uncertainty present. Etymology: greek, “guess”.

YAML

File format, file ending .yaml or .yml. .yml pronounced as “yaml”.

Markov model

A model describing a system which is changing over time but the next state is only dependent on the current state, not previous states.

future cost function

A function or variable containing the estimate of the cost of operating as planned from a given time and up to a defined future, ex. from today up until end of simulation horizon.

K-cost

Cost estimate from backward phase of iteration. Upper bound of optimal cost.

F-cost

Cost estimate from forward phase of iteration. Lower bound of optimal cost.

water value

The opportunity cost of not using water at a given time and state of reservoirs. Dual value of reservoir volume.

price taker

A market actor which does not influence the price. Large actors whose actions can influence or control the market price can be price makers.

convex

For functions, the property that the function is continous and any line between two points on the graph of the function lie on or above the graph; “smiling”. Important property because it guarantees a solution, i.e., the existence of a global minimum which can be found by moving along the function in any direction.

concave

Opposite to convex.