GlobalSearch Not Analyzing all Trial Points for a Specific Objective Function

I am currently having an issue when running GlobalSearch using the Global Optimization toolbox. I have found that if I use specific objective functions such as functions that find the spectral radius of a matrix i.e.

 max(abs(eig(A)))

then GlobalSearch will not analyze all of the trial points and instead it will only run one local solver. My optimization problem is defined as:

options = optimoptions(@fmincon, 'Algorithm', 'interior-point', ... 
'Display', 'final', 'OptimalityTolerance', 1e-10, ...
'SpecifyObjectiveGradient', false, 'CheckGradients', false, ... 
'stepTolerance', 1e-14', 'MaxFunctionEvaluations', 100000, ... 
'MaxIterations', 1000, 'ScaleProblem', 'obj-and-constr',...
'ScaleProblem', 'obj-and-constr', 'ConstraintTolerance', 0);

 % Define optimization problem
 problem = createOptimProblem('fmincon', 'objective', ...
 @(dn) objFuncs(dn,iteration,op), 'x0', op.d0{1},'ub', op.ub{1}, 'lb',...
 op.lb{1}, 'nonlcon', @(dn) nonlcons(dn,op), 'options', options);

 % Define GlobalSearch structure
 gs = GlobalSearch('Display', 'iter', 'FunctionTolerance', 0, 'NumTrialPoints', 2000);

 % Run optimization

 [dsol, errv, exitflag, output, solutions] = run(gs, problem)

An example of an objective function that works is:

 function [fval] = objective_weight2(H,D1,x,op_accorder)

 e1 = D1*x.^(op_accorder+1) - (op_accorder+1).*x.^(op_accorder);
 e2 = D1*x.^(op_accorder+2) - (op_accorder+2).*x.^(op_accorder+1);

 fval = (e1+e2)'*abs(H)*(e1+e2);
 end

which will give me outputs like:

 fmincon stopped because the size of the current step is less than
 the selected value of the step size tolerance and constraints are 
 satisfied to within the selected value of the constraint tolerance.

 <stopping criteria details>

   Num Pts                 Best       Current    Threshold        Local        Local
  Analyzed  F-count        f(x)       Penalty      Penalty         f(x)     exitflag        Procedure
    1510     8963   3.007e-14        -115.6       -2.921    1.296e-13            2    Stage 2 Local
    1600     9053   3.007e-14         3.113       -338.2                              Stage 2 Search
    1700     9153   3.007e-14       0.01529       -133.4                              Stage 2 Search
    1800     9253   3.007e-14       0.02712       -54.07                              Stage 2 Search
    1900     9353   3.007e-14         2.076       -17.04                              Stage 2 Search
    2000     9453   3.007e-14        -5.749       -648.8                              Stage 2 Search

which is expected. However, if I use the same objective function but add a line that finds the spectral radius:

 function [fval] = objective_weight2(H,D1,x,op_accorder)

 e1 = D1*x.^(op_accorder+1) - (op_accorder+1).*x.^(op_accorder);
 e2 = D1*x.^(op_accorder+2) - (op_accorder+2).*x.^(op_accorder+1);
 sr = max(abs(eig(D1)));

 fval = (e1+e2)'*abs(H)*(e1+e2);
 end

even though I don't use the spectral radius in calculating fval then I get the following output:

 Num Pts                 Best       Current    Threshold        Local        Local
Analyzed  F-count        f(x)       Penalty      Penalty         f(x)     exitflag        Procedure
 0      109   3.007e-14                               3.007e-14            2    Initial Point

 GlobalSearch stopped because it analyzed all the trial points.

 The local solver ran once and it converged with a positive local solver exit flag.

And it stop at Num Pts Analyzed = 0 and not go up to 2000.