test_linear_coefficients.R
In validate: Data Validation Infrastructure

## linear coeffiencts can be derived ----
v <- validator(x>0)
expect_equivalent(v$linear_coefficients()$A , matrix(-1,1,1))
expect_equivalent(v$linear_coefficients()$b, matrix(0,1,1))
w <- validator(2*x > 0)
expect_equivalent(w$linear_coefficients()$A,matrix(-2,1,1))
expect_equivalent(w$linear_coefficients()$b,matrix(0,1,1))
x <- validator(2*x + y > 0)
expect_equivalent(x$linear_coefficients()$A,matrix(c(-2,-1),1,2))
expect_equivalent(x$linear_coefficients()$b,matrix(0,1,1))  
y <- validator(2*x + y > 3*z + w)  
expect_equivalent(y$linear_coefficients()$A,matrix(c(-2,-1,3,1),1,4))
expect_equivalent(y$linear_coefficients()$b,matrix(0,1,1))  
z <- validator(2*x + 1 +  y -2 > 4 + z - 8)
expect_equivalent(z$linear_coefficients()$A,matrix(c(-2,-1,1),1,3))
expect_equivalent(z$linear_coefficients()$b,matrix(3,1,1))


## linear equalities are detected ----
for ( op in c("<", "<=", "==", ">=", ">") ){
  expect_false(validate:::linear_call( parse(text=paste("x",op,'"a"'))[[1]]  ))
}
expect_false(validate:::linear_call(expression( "a"*x < 3)[[1]]))
expect_false(validate:::linear_call(expression( x < -"a")[[1]]))


## normalisation can be switched off ----
# check normalisation
z <- validator(2*x + 1 +  y -2 > 4 + z - 8)
expect_equivalent(z$linear_coefficients(normalize=FALSE)$A, -1*matrix(c(-2,-1,1),1,3))
expect_equivalent(z$linear_coefficients(normalize=FALSE)$b, -1*matrix(3,1,1))