a basic question about subgradient method

greetings, hi... here is what i'm dealing with...

1. i need to test a program for subgradient method with several value of alpha.
2. i have the program but i need to modify it a little bit..
3. this is my template

function [x,hist] = sgm_pwl_nonsum_dimin(A,b,x1,a,MAX_ITERS)  
% subgradient method for linear piecewise minimization
% uses nonsummable diminishing step size rule, **alpha_k = a/sqrt(k)**

f = [+Inf]; fbest = [+Inf];  
k = 0;
x = x0;

while k < MAX_ITERS  
  % subgradient calculation
  [fval,ind] = max(A*x+b);
  g = A(ind,:)';

    % step size selection
    alpha = a/sqrt(k);

    % store objective values
    f(end+1) = fval;
    fbest(end+1) = min( fval, fbest(end) );

    % subgradient update
    x = x - alpha*g;  k = k + 1;
end

% collect history information
hist{1} = f; hist{2} = fbest;

My questions are

1. I need to update the value of x (x_new=x_old-alpha*g)
2. I need to find the difference between x_new and x_old
3. Since the difference is in matrix and have a few elements, I need to find the min of x_new-x_old
4. Same goes to subgradient, after finding one, I need to find the difference and find the max value .

How can I do that? thx