PatD

20 Reputation

6 Badges

11 years, 109 days

MaplePrimes Activity


These are replies submitted by PatD

@Markiyan Hirnyk 

Here is equation C2 in a .mw for you to look at:

C2.mw

 

Pat

@Markiyan Hirnyk 

No I haven't used the DirectSearch package, sounds like a promising avenue. Thanks. I will try to see if I can save my equations to a .m file for you. Hopefully this would allow you to play around with them.

Pat

@Markiyan Hirnyk 

I wish I could, however there is some intellectual property involved. It would defenitly make it easier. I will try to clarify:

 

The C equations are 9th order polynomials (which complicates things a lot) and they look like this portion which I have extracted from one of the equations:

The equations themselves usually exceed a limit of 1000000 and are not displayed. The fsolve command looks specifically like this:

 

I know I have significant digit issues and I have been playing around a lot with the number of Digits that maple uses to solve. I have also tried using fulldigits in fsolve. The only time maple solves is when I use only 6 digits or less and I don't use fulldigits in fsolve, however the solution maple gives in this case do not even satisfy the original equations. Which I find weird. In solving the known problem and knowing the error that is involved I would normally need to keep about 12 digits. But maple craps out before then anyways. Also In my problem, I would like to be acurate to only about 4 digits since data is only collected to be about that precise. Not sure how to even do that. I know you guys are all really smart and mathematically I am at a stand still. The defect of being an engineer I suppose.

 

So let us pretend that the actually solution requires c1=1.72463215623. I would be completly happy if maple found 1.724 and then whatever followed could be numerically different. But maple never even gets there, it returns unevaluated instead in everything that I have tried. I would like to point out that I am using a computer that has 16gigs of ram so memory should not be an issue.

 

Thanks again for all your help,

 

Pat

Thank you Preben, this is pretty much what I expected also having read the fsolve,details. However I suppose I am getting frustrated because I am testing my code on a system that has known solutions and fsolve still returns unevaluated. I also have a very powerful computer. It seems to me that fsolve should conitune trying longer before it returns unevaluated. I feel like it just quits trying and I was hoping there was something I could tweek about fsolve too make it try longer.

 

Pat

Thank you Preben, this is pretty much what I expected also having read the fsolve,details. However I suppose I am getting frustrated because I am testing my code on a system that has known solutions and fsolve still returns unevaluated. I also have a very powerful computer. It seems to me that fsolve should conitune trying longer before it returns unevaluated. I feel like it just quits trying and I was hoping there was something I could tweek about fsolve too make it try longer.

 

Pat

1 2 Page 2 of 2