Small bug in ValueFnFromPolicy_InfHorz

aledinola · July 11, 2026, 3:33pm

In the interpolation branch of [ValueFnFromPolicy_InfHorz.m (line 101)]: itercount is not incremented when gridinterplayer = 1, so maxiter is ineffective.

github.com/vfitoolkit/VFIToolkit-matlab

ValueFnFromPolicy/ValueFnFromPolicy_InfHorz.m

master

function V=ValueFnFromPolicy_InfHorz(Policy,n_d,n_a,n_z,d_grid,a_grid,z_grid, pi_z, ReturnFn, Parameters, DiscountFactorParamNames, vfoptions)

if ~exist('vfoptions','var')
    vfoptions.gridinterplayer=0;
    vfoptions.tolerance=10^(-9);
    vfoptions.maxiter=10^4; % Can be used to stop the VFI after a finite number of iterations
    % divide-and-conquer is not relevant for ValueFnFromPolicy
else
    if gpuDeviceCount==0
        error('ValueFnFromPolicy_InfHorz is only available on GPU')
    end
    if ~isfield(vfoptions,'gridinterplayer')
        vfoptions.gridinterplayer=0;
    end
    if ~isfield(vfoptions,'tolerance')
        vfoptions.tolerance=10^(-9);
    end
    if ~isfield(vfoptions,'maxiter')
        vfoptions.maxiter=10^4; % Can be used to stop the VFI after a finite number of iterations
    end

This file has been truncated. show original

robertdkirkby · July 12, 2026, 2:38am

One line fix, thanks for spotting!