Hi
I want to create a table
column 1 will be the value of n from 1 to 20 n= 1, 2, 3, ... , 20
column 2 will be the nth (respectively) taylor approximation of a function [ so n terms of a taylor approximation of ]
column 3 will be the error of this taylor approximation compared to the real value of erf(2) so [erf(2)-nth term taylor approx]
column 4 will be a different series approximation of the same function erf(2) [asymptotic series ] where the double !! means only odd numbers, i.e. .
column 5 will be the error of this second approximation, i.e. [erf(2) minus my n-term asymptotic approximation]
Below is my try. my tayterm doesn't work, neither does my asymterm. The values given are incorrect.
:( thank you.
i would also like gridlines on the table...not sure how to do that.
format long
syms k x t X
%tayt =[taylor(erf(x),x,'Order',2),taylor(erf(x),x,'Order',3),taylor(erf(x),x,'Order',4),taylor(erf(x),x,'Order',5),taylor(erf(x),x,'Order',6),taylor(erf(x),x,'Order',7),taylor(erf(x),x,'Order',8),taylor(erf(x),x,'Order',9),taylor(erf(x),x,'Order',10),taylor(erf(x),x,'Order',11),taylor(erf(x),x,'Order',12),taylor(erf(x),x,'Order',13),taylor(erf(x),x,'Order',14),taylor(erf(x),x,'Order',15),taylor(erf(x),x,'Order',16),taylor(erf(x),x,'Order',17),taylor(erf(x),x,'Order',18),taylor(erf(x),x,'Order',19)];
for n=1:18
N(n)=n;
tayterm(n)=taylor(erf(x),x,'Order',n);
ercol2(n) = erf(2)-tayterm(n);
asymterm(n)=1-2/sqrt(pi)*exp(-2^2)*(symsum((-1)^k*x^(2*k+1)/((2*k+1)*factorial(k)),k,0,n-1));
ercol4(n) = erf(2)-asymterm(n);
end
format long
T = table(transpose(N),transpose(double(tayterm)),transpose(ercol2), transpose(asymterm),transpose(ercol4));
T.Properties.VariableNames = {'n' 'n-term Taylor Approx for erf(2)' 'Error in col 2' 'n-term asymptotic approx of erf(2)' 'Error in col 4'}

7 Kommentare

Sindar
Sindar am 21 Jan. 2020
In what way does it not work?
I haven't used symbolic math much, but my first thought is that it doesn't want to create an array. Try cell arrays instead:
tayterm{n}=taylor(2/sqrt(pi)*erf(x),t,0,'Order',n);
Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 21 Jan. 2020
it gives me incorrect values for the tayterm and asymterm
the curly brackets give me an error
Error using sym/subsasgn (line 942)
Invalid indexing assignment.
Error in testing (line 7)
tayterm{n}=taylor(2/sqrt(pi)*erf(x),t,0,'Order',n);
Sindar
Sindar am 21 Jan. 2020
Could be that your function ( erf(x) ) and variable (t) are different, try:
tayterm(n)=taylor(erf(t),t,0,'Order',n);
Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 21 Jan. 2020
but the input x has to do with the bounds of the error function which is an integral of t.
Sindar
Sindar am 22 Jan. 2020
You are mixing up "x, the input to the function" and "x, the value where you evaluate the function". There are a few ways you could fix your code, but
taylor(erf(x),t)
doesn't make sense. It's (almost exactly) like asking for d[f(x)]/dy instead of d[f(x)]/dx.
syms k x t
x0=2;
for n=1:20
N(n)=n;
% Taylor series of erf(x) with respect to x, centered at 0, order n
tayterm(n)=taylor(erf(x),x,0,'Order',n);
% error between taylor apprx and value, at chosen x0
ercol2(n) = double(erf(x0)-tayterm(n));
% issue with asymterm was replacing the first of the x's with 2
asymterm(n)=1-2/sqrt(pi)*exp(-x^2)*(symsum((-1)^k*x^(2*k+1)/((2*k+1)*factorial(k)),k,0,n-1));
ercol4(n) = erf(x0)-asymterm(n);
end
Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 22 Jan. 2020
Okay, this is giving me the following error:
Error using symengine
Unable to convert expression into double array.
Error in sym/double (line 700)
Xstr = mupadmex('symobj::double', S.s, 0);
Error in testing (line 19)
T = table(transpose(N),transpose(double(tayterm)),transpose(ercol2), transpose(asymterm),transpose(ercol4));
Sindar
Sindar am 22 Jan. 2020
You are trying to turn a symbolic expression into a number. That doesn't work. You can either leave the symbolic expression alone or evaluate it for a particular input (using subs), depending on what you want in the table

Melden Sie sich an, um zu kommentieren.

 Akzeptierte Antwort

Walter Roberson
Walter Roberson am 21 Jan. 2020

0 Stimmen

syms X
tayterm(n) = taylor(erf(X),X,x,'Order',n)
You always need to taylor() a function, not a constant.

5 Kommentare

Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 21 Jan. 2020
Bearbeitet: Ursula Trigos-Raczkowski am 22 Jan. 2020
thanks for your comment, again, this is the wrong function, i want a taylor function with respect to t not X and it should be around 0 not around x
so the taylor approximation of erf(2) with 2 terms is something like
a taylor approximation of erf(2) in general is something like
the asymptotic approximtion of erf(2) with 2 terms is something like
Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 22 Jan. 2020
how can i set a string with the taylor values and then later evaluate it at x=2?
Sindar
Sindar am 22 Jan. 2020
subs(tayterm(n),2)
Ursula Trigos-Raczkowski
Ursula Trigos-Raczkowski am 22 Jan. 2020
where? in the table? inside transpose? in the loop?
syms t
tayterm(n) = subs( taylor(erf(t),t,0,'Order',n), t, x );
That would taylor around 0 but evaluate at 2

Melden Sie sich an, um zu kommentieren.

Weitere Antworten (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by