Hi, Taking a look at your equations, your problem is the fractional power terms (...)^d and especially (...)^c . If the arguments in the parenthesis become negative you can get complex answers. If you don't expect complex answers, then something is wrong with the arguments, which could be the signs of the terms or the initial conditions. Cheers, Kay

Hi David,

That's no trouble, I was just wondering if there was just a quick 'assumptions -> ' method which I had missed out on however that appears not to be the case! As this has somewhat deviated from the original question I may open up a new discussion in the Maths section. It appears to be a common problem so hopefully that may prove useful to several people.

I cannot thank you enough for volunteering your time and knowledge, keep up the good work David!

Best regards, Robin

Dear David,

Thank you very much for the extra file and I hope you have enjoyed your Christmas! Having looked into it, I have found many forum posts from people with a similar result from integration. Apparently this is a property of NDSolve that takes into account the possibility of a complex differential equation as the general solution and therefore, as a general solution, returns complex numbers. As you have been so helpful in sharing your knowledge so far, I was wondering if you would know how to confine the integration to the real domain in the attached file?

Many thanks, Robin

Attachments:

Dear David and David,

I must thank you for such a comprehensive answer, I'm absolutely delighted! I was completely unsure of what the problem was however now you've pointed it out I can go away and double check the equations. I did derive them so I'll have to have a good run through them again.

Another great benefit is that you've given me a really simple way of solving the equations, the syntax I was using before was very clunky (as I'm sure you've noticed!) so I can use this much simpler version from now on.

Kindest regards and thanks for your help, especially on Christmas Eve!

Robin

Yes, a very quick look at the equations does not show me that I'd expect complex results, but I'd have to dig in deeper.

I don't know the details of your differential equation, but it appears that, for some reason, the results are not strictly real, for example:

I

n[87]:= ({CSA[t], CASA[t],
    CASAA[t], CHA[t],
    CH2O[t], CAA[t],
    V[t]} /. Reaction[[1]]) /. t -> 10

Out[87]= {2.14012 + 0.0397231 I, 1.69264 + 0.23278 I, 
 7.36331 + 0.656905 I, 8.99929 - 0.0635311 I, 
 0. + 0. I, -0.0623169 - 0.0743696 I, 141.893 + 2.18947 I}