[Feature Request] Support Boolean Indexing for Constraints

[RoyiAvital]

It seems there is no support for using BitVector in constraints.

A simple (Stupid) example:

using Convex;
using ECOS;

numSamples = 100;
vY = rand(numSamples);

valThr = sort(vY, rev = true)[5];
vP = vY .>= valThr; #<! Set of boolean indices

vX = Variable(numSamples);
sConvProb = minimize( 0.5 * sumsquares(vX - vY), [vX[vP] == vY[vP]] ); #<! Will fail
solve!(sConvProb, ECOS.Optimizer; silent = true);

If one define vPi = findall(vP); then [vX[vPi] == vY[vPi]] will work.

[odow]

PRs accepted.

It's pretty much just this method, but for BitVector:

Convex.jl/src/atoms/IndexAtom.jl

Lines 112 to 118 in cc71b51

	function Base.getindex(x::AbstractExpr, I::AbstractMatrix{Bool})
	return [xi for (xi, ii) in zip(x, I) if ii]
	end
	function Base.getindex(x::AbstractExpr, I::AbstractVector{Bool})
	return [xi for (xi, ii) in zip(x, I) if ii]
	end

[RoyiAvital]

I can add those with BitVector and BitMatrix.
I tried looking at the tests, should a specific test be added?

[odow]

Add a new function after:

Convex.jl/test/test_atoms.jl

Lines 470 to 544 in cc71b51

	### affine/IndexAtom
	function test_IndexAtom()
	target = """
	variables: x1, x2
	minobjective: [1.0 * x1, 1.0 * x2]
	"""
	_test_atom(target) do context
	return Variable(2)[:, 1]
	end
	x = Variable(2)
	@test x[:, :] === x
	target = """
	variables: x1, x2, x3
	minobjective: [1.0 * x1, 1.0 * x3]
	"""
	_test_atom(target) do context
	return Variable(3)[[1, 3]]
	end
	target = """
	variables: x1, x2, x3
	minobjective: [1.0 * x1, 1.0 * x3]
	"""
	_test_atom(target) do context
	return Variable(2, 2)[1, :]
	end
	target = """
	variables: x1, x2, x3, x4
	minobjective: [1.0 * x2, 1.0 * x4]
	"""
	_test_atom(target) do context
	return Variable(2, 2)[2, :]
	end
	_test_atom(target) do context
	return Variable(2, 2)[2:2, :]
	end
	target = """
	variables: x1, x2
	minobjective: [1.0 * x1, 1.0 * x2]
	"""
	_test_atom(target) do context
	return Variable(2, 2)[:, 1]
	end
	target = """
	variables: x1, x2, x3, x4
	minobjective: [1.0 * x3, 1.0 * x4]
	"""
	_test_atom(target) do context
	return Variable(2, 2)[:, 2]
	end
	y = [true, false, true]
	x = Variable(3)
	@test string(x[y]) == string([x[1], x[3]])
	return
	end
	function test_IndexAtom_issue_509()
	# *Scalar* indexing creates a column vector
	# https://github.com/jump-dev/Convex.jl/issues/509
	x = rand(2, 3)
	ϕ = Variable(2, 3)
	Convex.set_value!(ϕ, rand(2, 3))
	i = j = 1
	# "column-vector" in Convex.jl is a 1 column matrix
	@test size(ϕ[i, :]) == (3, 1)
	@test size(ϕ[i, 1:2]) == (2, 1)
	@test size(ϕ[i, [2, 1]]) == (2, 1)
	@test size(ϕ[i:i, :]) == (1, 3)
	@test size(evaluate(ϕ[i, :])) == (3,)
	@test size(evaluate(ϕ[i:i, :])) == (1, 3)
	# "Column vector", not 3x3 matrix!
	atom = broadcast(*, ϕ[i, :] - ϕ[j, :], x[i, :] - x[j, :])
	@test size(evaluate(atom)) == (3, 1)
	return
	end

I assume you want something like:

     target = """ 
     variables: x1, x2, x3 
     minobjective: [1.0 * x1, 1.0 * x3] 
     """ 
     _test_atom(target) do context 
         x = Variable(3)
         y = BitVector([true, false, true])
         return x[y] 
     end