# Series summation issue - conversion to double error

6 views (last 30 days)
Jon on 20 Feb 2013
Hi, I'm trying to write a series summation code and I'm getting the "conversion to double" error when I plot.
Here's the part of the code with trouble:
syms m
rhs = (1./(1+(2.*(t- m.*p))./tc)).* exp((-2.*r^2)./(r^2.*(1+ (2.*(t-m.*p)./tc))));
summation1 = symsum(rhs,0,1.89E5);
Where tc, r, and p are constants and t is a timespace with 36 spaces.
Is the problem that I'm asking it to sum from 0 to 1.89e5 with those 36 time spaces?
Thanks
EDIT: http://i.imgur.com/mz2KV5h.jpg Link is of the sum

bym on 21 Feb 2013
At some point, Matlab gives up & returns the indefinite summation. You can try this:
clc;clear
syms m
tc = 3;
p = 3;
t = 1:10;
r =2;
rhs = (1./(1+(2.*(t- m.*p))./tc)).* exp((-2.*r^2)./(r^2.*(1+ (2.*(t-m.*p)./tc))));
summation1 = symsum(rhs);
f = matlabFunction(summation1);
f(189000)
f(10)
ans =
1.0e-005 *
Columns 1 through 8
-0.2646 -0.2646 -0.2646 -0.2646 -0.2646 -0.2646 -0.2646 -0.2646
Columns 9 through 10
-0.2646 -0.2646
ans =
Columns 1 through 8
-0.0608 -0.0634 -0.0662 -0.0692 -0.0725 -0.0762 -0.0802 -0.0847
Columns 9 through 10
-0.0897 -0.0954
Jon on 23 Feb 2013
Thank you, this will work for now while I have a dozen or so points, but it gets quite complex when I want to extend my time.
Many thanks!

Walter Roberson on 20 Feb 2013
You could try
double(summation1)
If that gives you the same error then your summation1 must still contain some symbolic variable. Try
symvar(summation1)
##### 2 CommentsShowHide 1 older comment
Jon on 21 Feb 2013
Here's the sum for completeness