Returning to wrong result with multiple return values

General Usage
1

I used Newton interpolation for function q and Horner’s algorithm for the evaluation, here is the code which works correctly;

function Q(q,t)
  d = q;   
# create divided difference
    for j = 1:length(q)-1
    for i = length(q):-1:j+1    
        d[i] = (d[i]-d[i-1])/(η[i]-η[i-j]); 
    end
    end
# the evaluation    
    p = d[end] ; dp = 0;
  for i = length(q)-1:-1:1 
          dp = p + (t-η[i]) * dp
          p = d[i] + p*(t-η[i])
          
    end
    return p, dp;  #p returns to Q(t) and dp returns to Q'(t)
end

As an example if we take η = [1.0, 1.3, 1.6, 1.9, 2.2]; q = [0.7651977, 0.6200860 , 0.4554022, 0.2818186, 0.1103623], then Q(q , 1.5) it gives me the correct result (0.5118199942386832, -0.5578831893004117).

My problem is, I would like to create two functions return to Q(q , t)[1] and Q(q , t)[2] like q̃ = (q,t) -> Q(q,t)[1] ; q̃′ = (q,t) -> Q(q,t)[2] but the result is (0.5118199942386832, 1.6121262983073357) which is wrong for Q(q , t)[2].

2

The problem is that Q modifies its argument q. d = q does not create a new vector; it just defines a new name for the same vector. The solution is to use d = copy(q).

3

Yeah, it works very well, thank you @mbaz

1 Like