FRA_diving_heuristic.m

function divingOutput = FRA_diving_heuristic(model, params)
%FRA_DIVING_HEURISTIC Implements the feasible rounding approaches diving
%heuristic. Initially runs feasiblility diving and then optimality diving.
%Mode can be 'RA' (random) or 'MC'(greedy), where maxIter is fixed within
%the individual diving methods.
if ~isfield(params,'mode')
    fprintf('WARNING: SETTING MODE TO RANDOM');
    params.mode = 'RA';
end
if ~isfield(params,'maxIter')
    fprintf('WARNING: SETTING maxIter TO 30');
    params.maxIter = 30;
end


model = preProcessModel(model);
depth0 = 0; 
objValPP = inf;
node_granular = false;
if ~isfield(model,'objcon')
    objCon = 0;
else
    objCon = model.objcon;
end
fprintf('Running Feasibility diving\n');
[xyFeasible, fixedIndices, fixingValues, alpha, iterF, timeFD] = feasibilityDiving(model,params);
time = timeFD;
if isnan(xyFeasible)
   objVal = inf;
elseif alpha>0
    objVal = transpose(model.obj)*xyFeasible + objCon;
else
    fprintf('Running Optimality diving\n');
    node_granular = true;
    [xyFeasible, depth0, v_check, v_pp, ~, timeOD] = optimalityDiving(model,params,fixedIndices,fixingValues);
    time = time+timeOD;
    objVal = v_check + objCon;
    objValPP = v_pp + objCon;
end
divingOutput.node_granular = node_granular;
divingOutput.objVal = objVal;
divingOutput.objValPP = objValPP;
divingOutput.time = time;
divingOutput.iterF = iterF;
divingOutput.depth0 = depth0;
divingOutput.xy = xyFeasible;
end