[R-SIG-Finance] fPortfolio (version 3011.81) - solveRglpk.CVAR - lower bound constraints (z_i >= 0) allows negative values

[Pedro Oliveira]

Hi,

I have been looking at the code of the solveRglpk.CVAR function, in order
to create a similar function but where the objective function will be a
trade-off between the expected return and a risk aversion factor times the
CVaR risk measure. In the hidden function .cvarRglpkArguments, where the
matrices, served as input for the Rglpk solver, are built, I think the
lower bounds are not properly defined. Namely, the optimization problem has
the following constraints z_i >= 0 where i = 1:nScenarios, thus the lower
bound should not be -Inf, but 0. Am I making any mistake?

Lines of code (164-166):
    bounds <- list(
        lower = list(ind = nInd, val = c(rep(-Inf, 1+nScenarios), minW)),
        upper = list(ind = nInd, val = c(rep( Inf, 1+nScenarios), maxW)) )


Note: z_i is the term being summed after we discretized the integral.

Thanks and regards

Pedro Oliveira

	[[alternative HTML version deleted]]