Integration in Octave between float point

Question

I am trying to calculate the integral between two float points. The programm can do it, but it sends this next messagge

warning: passing floating-point values to sym is dangerous, see "help sym"

This is my code

i=2.5;

syms x y

f=((2*x-1)/10)*((2*y-1)/10)/(x+y);

variable=double(int(int(f,x,i,1),y,i,1));

is there anyway to do it whithout that inconvenient?

Answer 1

I assume you're casting to double at the end because of the answer you got to your other question: Could someone explain this error on fprintf() in Octave?, where casting your answer to double was suggested so that you could pass it to fprintf.

Octave is giving you a warning about combining floating type and symbolic numbers. This is a potential concern because symbolic numbers are stored 'exactly', while floating points may not be, depending on their value. But it is only a warning. Not recommended, but you can disable warnings. See https://docs.octave.org/latest/Enabling-and-Disabling-Warnings.html

The problem is occurring because you are using a float value 2.5 in your integration limits. To avoid the issue, define it as a symbolic value:

pkg load symbolic;

syms x, y;

i = sym("2.5")
i = (sym) 5/2

variable = int(int(f,x,i,1),y,i,1)
variable = (sym)

    11*log(5)   log(2)    33   28*log(7/2)
  - --------- - ------ + --- + -----------
        30       150     200        75

variable = double (variable)
variable = 0.037950

As suggested in a comment above, if you have no need for a symbolic answer to your question, and you're casting it immediately to a floating point number, you might find it easier to do numeric integration instead. That could be accomplished for your problem as follows:

# define equation as a function. Note use of .* and ./ 'elementwise' 
# operators to avoid unwanted matrix multiplication/division.
f = @(x,y) ((2*x-1)/10).*((2*y-1)/10)./(x+y); 

# Solve the 2D integral, returning a numeric result
variable = integral2 (f, 2.5, 1, 2.5, 1)
variable = 0.037950