Parameters, variables, constraints and the objective#

These four blocks carry the math, and given: names what the math reads from another file. Each takes an optional description:, free text that the typeset legend prints.

parameters#

A parameter declares a shape. The numbers arrive by name with the data.

dimensions:
  snapshot: { dtype: int }
parameters:
  load:
    dims: [snapshot]
  discount_rate:
    dims: [] # a scalar

The column has to match the dtype:

A null or NaN value is refused when the data is attached. A coordinate with no value has no row. inf and -inf are values.

Only float and int are values. A str parameter is a label and a bool parameter is a mask: each selects rows in a where, and writing either as a coefficient, a term or a divisor is a load error. A 0 or 1 that is meant to be multiplied by is declared dtype: int.

variables#

A variable is what the solver decides. There is one column per coordinate of dims.

dimensions:
  snapshot: { dtype: int }
  generator: { dtype: str }
parameters:
  capacity: { dims: [generator] }
variables:
  dispatch:
    dims: [snapshot, generator]
    where: "capacity > 0"
    bounds:
      lower: 0
      upper: capacity

An open side is null. A bound is never infinite: .inf and -.inf are refused, with null named as the rewrite.

A bound is a name or a number: upper: capacity is accepted, and upper: -rating is refused. Ship the negated column as data.

Equal bounds pin a variable (fix a quantity). A pinned variable is still a variable.

given#

given: holds what this file reads and does not build: data under parameters:, columns under variables:, named expressions under expressions:, and row families under constraints:. It takes those four keys and no other. A file with a given: block loads and prints on its own.

given: parameters#

A given parameter is data this file reads and another file declares.

dimensions:
  snapshot: { dtype: int }
  generator: { dtype: str }
given:
  parameters:
    gen_cost: { dims: [generator], description: what one unit of output costs }
    gen_on: { dims: [generator], dtype: bool }
  variables:
    gen_p: { dims: [snapshot, generator] }
constraints:
  off_units_idle:
    dims: [snapshot, generator]
    where: not gen_on
    expression: gen_p <= 0
objective:
  sense: minimize
  expression: sum(gen_p * gen_cost)

A given parameter is read wherever a parameter is: in an expression, a where and a bound. A name declared under both parameters: and given: parameters: is refused. The typeset legend lists a given parameter under Given.

given: variables#

A given variable is a column this file reads and another file introduces.

dimensions:
  snapshot: { dtype: int }
  port: { dtype: str }
  generator: { dtype: str }
relations:
  gen_port: { key: generator, values: port }
variables:
  gen_p: { dims: [snapshot, generator], bounds: { lower: 0 } }
given:
  variables:
    flow:
      dims: [snapshot, port]
      description: what a port puts into its bus
constraints:
  gen_injects:
    dims: [snapshot, generator]
    expression: at(flow, by=gen_port, over=port, into=generator) == gen_p

There is no bounds and no where. The file that introduces the column owns both.

An expression reads a given variable as it reads any other. A name declared under both variables: and given: variables: is refused. The typeset legend lists a given variable under Given, and prints no domain line for it.

merge folds a given declaration into the declaration of another fragment that introduces the name, so a composed library carries none of them. The folded declaration is the introducer's, and what the reader states has to say the same or less.

Where nothing in this language introduces the column, the program carries the declaration until a host model provides it (what a program does not build).

given: constraints#

A given constraint is a row family that another model builds. This file reads its dual.

dimensions:
  snapshot: { dtype: int }
  bus: { dtype: str }
given:
  constraints:
    balance:
      dims: [snapshot, bus]
      description: the host model clears each bus
expressions:
  price:
    expression: dual(balance)

There is no expression and no sense. dual(name) is the only place a given row family may be named, and the frame gives the reported expression its dimensions. A name declared under both constraints: and given: constraints: is refused.

given: expressions#

A given expression is a named expression this file reads and another file defines.

dimensions:
  snapshot: { dtype: int }
  bus: { dtype: str }
given:
  expressions:
    injection:
      dims: [snapshot, bus]
      description: what the components put into a bus
constraints:
  balance:
    dims: [snapshot, bus]
    expression: injection == 0

There is no body. A term of this file may add to the name. This file reads the name as it reads a given variable: a quantity over the frame, of degree one. A where does not read it, because a mask is built before any variable exists. A name declared under both expressions: and given: expressions: is refused. The typeset legend lists a given expression under Given.

merge folds a given expression into the definition of another fragment. The dims are an upper bound: a body that carries a dimension they do not name is refused. A body over fewer dimensions is folded, and the composed spec decides: it refuses a row that would repeat across the missing dimension, and accepts one where another term carries it. The composed spec holds the body to the rules of every place this file reads it: a square of a given expression that is quadratic is refused once folded.

Terms#

adds_to: adds this expression as a term to the sum it names. The sum is a given expression of the same file: the given entry is the read, and adds_to: is the write. The term is an ordinary named expression, so it takes cases:, a description and every other field a named expression takes. The file reads the name as the whole sum, alone and composed, and the term is its part of it.

# fleet.yaml adds a term
dimensions:
  snapshot: { dtype: int }
  bus: { dtype: str }
  generator: { dtype: str }
relations:
  gen_bus: { key: generator, values: bus }
variables:
  gen_p: { dims: [snapshot, generator], bounds: { lower: 0 } }
given:
  expressions:
    injection: { dims: [snapshot, bus] }
expressions:
  generation:
    description: what the fleet puts into a bus
    expression: sum(gen_p, by=gen_bus, over=generator, into=bus)
    adds_to: injection

# balance.yaml reads the sum, and adds nothing to it
dimensions:
  snapshot: { dtype: int }
  bus: { dtype: str }
given:
  expressions:
    injection:
      dims: [snapshot, bus]
      description: what the components put into a bus
constraints:
  balance:
    dims: [snapshot, bus]
    expression: injection == 0

Each file loads alone. adds_to: names a given: expressions: entry of the same file, and a name the file does not read there is refused with the near miss. A name the same file defines is refused too: write the term into that body instead. The term does not read the name it adds to, directly or through another name. It carries no dimension the entry does not state, and it is held to degree two, as what reads the sum is. All of this is checked at load. The typeset legend lists the entry under Given and names the term, and the math prints the term under Definitions as its own line.

merge adds every term by its name, in the order the files are given in, and keeps each term as a named expression without its adds_to:. A cased term is added like any other, by its name.

Where one file defines the name with one expression:, the merged body is that body followed by the terms. The file that defines injection as slack gives slack + generation once fleet.yaml is merged with it. The body keeps its dims: and its description. A merged spec defines its sums, so it takes further terms the same way in a later merge. A merge defines every sum it has terms for, so the step that first merges the terms of a sum also merges a file that reads it. A definition written as cases: takes no term. Name the cased body as its own expression, and define the sum as that name.

Where no file defines the name, the merged spec defines it as its terms, over the frame the readers state. Every reader writes the dims in one order, since no file's order wins. The sum takes the first description a reader wrote. Some file has to read the name for more than adding to it: read it and add nothing, or use it in its math. Terms that only their own files read are refused, with the near miss, since that is what a misspelt given: entry looks like. A name one file alone reads is refused the same way, even where that file uses it in its math.

A name one file declares as a variable, a parameter or a constraint takes no term. Two files that both define the name collide, as any two declarations do. A term that reads its own sum through a name another file defines is refused with both files named:

fragment '#2' adds 't' to 'injection', and 't' reads 'injection' back through 'x' of '#1', so the sum would define itself. A term may not read what reads its sum: write 't' from something else, or define 'x' without 'injection'.

constraints#

One block is one rule. The name of the block is the name of the constraint.

dimensions:
  snapshot: { dtype: int }
  generator: { dtype: str }
parameters:
  load: { dims: [snapshot] }
variables:
  dispatch: { dims: [snapshot, generator] }
constraints:
  power_balance:
    dims: [snapshot]
    expression: sum(dispatch, over=generator) == load

The dimensions of the expression must equal its dims (how dimensions combine).

At least one side of the comparator carries a variable. A comparison between numbers and parameters alone is refused at load.

dims: [] gives one scalar row. A scalar variable may not carry a where; put the condition on the constraints that use it.

Two regimes of one rule are two blocks, each under its own where: (state a rule that differs by regime).

objective#

The objective is a single block with no name.

dimensions:
  generator: { dtype: str }
parameters:
  cost: { dims: [generator] }
variables:
  dispatch: { dims: [generator] }
objective:
  sense: minimize
  expression: sum(dispatch * cost)

The expression must be scalar. Nothing is summed for you: sum(x * a) + sum(y * b) and sum(x * a + y * b) are both allowed, and they state different objectives.

There is one objective block. To pursue several goals, weight them into one expression.