Need help fixing errors on code written in old MATLAB syntax

So this code was written to accompany a textbook I bought but the code was written back in the late 90's so it has old syntax that doesn't run well in the newer versions of Matlab. Particularly, I would like to know how to fix the first error saying "Too many output arguments" on the first if loop. The code also uses the statement clf which is old matlab syntax, how can I replace this?

RankDeficient();

Error using type
Too many output arguments.

Error in solution>RankDeficient (line 41)
if (type == 1)

function RankDeficient
% For "Applied Numerical Linear Algebra", Figures 3.4 and 3.5
% Written by James Demmel, Feb 4, 1996
%                Modified, Jun 2, 1997
%
% Illustrate solution of Rank Deficient Least Squares Problem using
% Truncated SVD
%
% Given any m-by-n input matrix A and use supplied tolerance tol,
% we produce a matrix Al with a range of singular values from
% small to large. Then we compute a rank-deficient Ard by setting 
% the singular values of Al less than tol to zero. 
% Then we generate a rank-deficient least
% squares problem whose solution we know exactly as follows:
%
%  1) Compute the SVD Ard = U*S*V', where U and V have rnk columns,
%     where rnk is the rank of Ard.
%  2) Pick a random xrd of dimension rnk, let x=V*xrd, and 
%     let b = Ard*x (= U*S*xrd). Then x is the exactly, minimum norm
%     solution of the rank-deficient least squares problems 
%     min(norm(A*y-b,2)).
%  3) Add random noise dA of varying norm to A, and solve the perturbed
%     least squares problem using the truncated SVD, getting perturbed 
%     solution y.
%  4) Plot norm(y) and norm(y-x) versus norm(dA).
%
% Inputs:
%   dimensions m (number of rows) and n (number of columns), m >= n
%   type = 1 to computed matrix with floor(n/2) zero singular values;
%            for figure 3.4
%        = 2 to computed matrix with singular values roughtly geometrically
%            distributed between 1e-15 and 1; for figure 3.5
%   tol = tolerance, below which to set small singular value to 0
%
%
% Here are two ways to compute Al. One should be commented out.
m=20;n=10;
A = randn(m,n);
if (type == 1)
%    Compute Al by zeroing out the last half of the columns of A:
    Al = randn(m,m) * ...
         [diag([ones(floor(n/2),1);zeros(ceil(n/2),1)]);zeros(m-n,n)] * ...
         randn(n,n);
end
if (type == 2)
%    Compute Al by multiplying columns by geometric sequence from 1e-16 to 1.
%    As a result Al will have singular values which form an approximate
%    geometric sequence in the same range.
     Al = A * diag(exp(log(10)*(0:n-1)*(-16/n)));
end
%
[Uorig,Sorig,Vorig]=svd(Al,0);
diag(Sorig);
rnk = length(find(diag(Sorig) >= tol));
disp(['rank(A) = ',int2str(rnk)])
U=Uorig(:,1:rnk);
S=Sorig(1:rnk,1:rnk);
V=Vorig(:,1:rnk);
Ard = U*S*V';
xrd = randn(rnk,1);
x = V * xrd;
b = Ard*x;
xnorm = norm(xrd);
nrm = sort([(10*ones(1,17)).^(-16:0),tol*(.1:.1:.9),tol*(.9:.01:1),...
             tol*(1:.1:1.9),tol*(2:10)]);
rnkdsav=[];
ynorm=[];
xmynorm=[];
for nn=nrm
  dA = randn(m,n);
  dA = dA/norm(dA);
  dA = nn*dA;
  ArdpdA = Ard + dA;
  [Ud,Sd,Vd] = svd(ArdpdA);
  rnkd = length(find(diag(Sd) >= tol));
  rnkdsav=[rnkdsav;rnkd];
  Udd=Ud(:,1:rnkd);
  Sdd=Sd(1:rnkd,1:rnkd);
  Vdd=Vd(:,1:rnkd);
  y = Vdd*(Sdd\(Udd'*b));
  ynorm = [ynorm;norm(y)/xnorm];
  xmynorm = [xmynorm;norm(x-y)/xnorm];
end
figure(1),
hold off
clf
%
subplot(2,1,1)
handl=semilogx(nrm',rnkdsav,'b'); set(handl,'LineWidth',2)
hold on
dS = min(max([diag(S)],1e-16),1);
handl=semilogx(dS,ones(size(dS)),'rx');set(handl,'MarkerSize',10)
axis([1e-16 1 0 min(m,n)]), grid
title(['Rank of perturbed A, original singular values are x''s, tol=',num2str(tol)])
xlabel('Norm of perturbation')
%
subplot(2,1,2)
handl=loglog(nrm',xmynorm,'b');set(handl,'LineWidth',2)
hold on
handl=semilogx(dS,ones(size(dS))*1e-15,'rx');set(handl,'MarkerSize',10)
axis([1e-16 1 1e-16 1]), grid
title('Norm(solution-perturbed solution)/norm(solution)')
xlabel('Norm of perturbation')
end

Answer 1

[Image Analyst]

It's expecting inputs that were not listed in the function line. So change it to

function RankDeficient(type, tol)

and make sure you pass in type and tol when you call the function.

[Christopher Arreola]

Thank you!

Answer 2

[Christine Tobler]

I'd also recommend replacing the calls svd(A) and svd(A, 0) with calling svd(A, 'econ'), since the code here is only using the first rnk (rnkd) singular values and the rest are discarded.