selectViolations: Select points from audit grid to add to the constraint grid

In ivmte: Instrumental Variables: Extrapolation by Marginal Treatment Effects

Description

This function selects which points from the audit grid should be included into the original grid. Both the constraint grid and audit grid are represented as constraints in an LP/QCQP problem. This function selects which points in the audit grid (i.e. which rows in the audit constraint matrix) should be added to the constraint grid (i.e. should be appended to the constraint matrix).

Usage

selectViolations(
  diffVec,
  audit.add,
  lb0seq,
  lb1seq,
  lbteseq,
  ub0seq,
  ub1seq,
  ubteseq,
  mono0seq,
  mono1seq,
  monoteseq,
  mbmap
)

Arguments

diffVec	numeric vector, with a positive value indicating a violation of a shape constraint.
audit.add	integer, the number of points from the audit grid to add to the initial for each constraint type. For instance, if there are 5 different kinds of constraints imposed, and audit.add = 5, then up to 30 points may be added to the constraint grid.
lb0seq	integer vector, indicates which rows in the audit constraint matrix correspond to the lower bound for m0.
lb1seq	integer vector, indicates which rows in the audit constraint matrix correspond to the lower bound for m1.
lbteseq	integer vector, indicates which rows in the audit constriant matrix correspond to the lower bound for the treatment effect.
ub0seq	integer vector, indicates which rows in the audit constraint matrix correspond to the upper bound for m0.
ub1seq	integer vector, indicates which rows in the audit constraint matrix correspond to the upper bound for m1.
ubteseq	integer vector, indicates which rows in the audit constriant matrix correspond to the upper bound for the treatment effect.
mono0seq	integer matrix, indicates which rows in the audit constraint matrix correspond to the monotonicity conditions for m0, and whether the constraint is increasing (+1) or decreasing (-1).
mono1seq	integer matrix, indicates which rows in the audit constraint matrix correspond to the monotonicity conditions for m1, and whether the constraint is increasing (+1) or decreasing (-1).
monoteseq	integer matrix, indicates which rows in the audit constraint matrix correspond to the monotonicity conditions for the treatment effect, and whether the constraint is increasing (+1) or decreasing (-1).
mbmap	integer vector, indexes the X-value associated with each row in the audit constraint matrix.

Value

The audit grid is represented using a set of constraint matrices. Each point in the audit grid corresponds to a set of rows in the constraint matrices. The function simply returns the vector of row numbers for the points from the audit grid whose corresponding constraints should be added to the original LP/QCQP problem (i.e. the points to add to the original grid).

ivmte documentation built on Sept. 17, 2021, 5:06 p.m.