# How to input pi

5,773 views (last 30 days)
Anthony on 20 Sep 2016
Commented: Walter Roberson on 1 Aug 2023
How can i enter pi into an equation on matlab?
##### 2 CommentsShow 1 older commentHide 1 older comment
Walter Roberson on 16 Dec 2022
That is what @Geoff Hayes suggested years before. But it does not enter π into the calculation, only an approximation of π

Geoff Hayes on 20 Sep 2016
Edited: MathWorks Support Team on 28 Nov 2018
Anthony - use pi which returns the floating-point number nearest the value of π. So in your code, you could do something like
sin(pi)
Walter Roberson on 16 Dec 2022
Also see sinpi and cospi

Essam Aljahmi on 31 May 2018
Edited: Walter Roberson on 31 May 2018
28t2e0.3466tcos(0.6πt+π3)ua(t).
John D'Errico on 28 Nov 2018
Edited: John D'Errico on 28 Nov 2018
As I recall, these approximations tend to give a roughly fixed number of digits per term. I'll do it using HPF, but syms would also work.
DefaultNumberOfDigits 500
n = 10;
piterms = zeros(n+1,1,'hpf');
f = sqrt(hpf(2))*2/9801*hpf(factorial(0));
piterms(1) = f*1103;
hpf396 = hpf(396)^4;
for k = 1:n
hpfk = hpf(k);
f = f*(4*hpfk-3)*(4*hpfk-2)*(4*hpfk-1)*4/(hpfk^3)/hpf396;
piterms(k+1) = f*(1103 + 26390*hpfk);
end
piapprox = 1./cumsum(piterms);
pierror = double(hpf('pi') - piapprox))
pierror =
-7.6424e-08
-6.3954e-16
-5.6824e-24
-5.2389e-32
-4.9442e-40
-4.741e-48
-4.5989e-56
-4.5e-64
-4.4333e-72
-4.3915e-80
-4.3696e-88
So roughly 8 digits per term in this series. Resetting the default number of digits to used to 1000, then n=125, so a total of 126 terms in the series, we can pretty quickly get a 1000 digit approximation to pi:
pierror = hpf('pi') - piapprox(end + [-3:0])
pierror =
HPF array of size: 4 1
|1,1| -1.2060069282720814803655e-982
|2,1| -1.25042729756426e-990
|3,1| -1.296534e-998
|4,1| -8.e-1004
So as you see, it generates a very reliable 8 digits per term in the sum.
piapprox(end)
ans =
3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825342117067982148086513282306647093844609550582231725359408128481117450284102701938521105559644622948954930381964428810975665933446128475648233786783165271201909145648566923460348610454326648213393607260249141273724587006606315588174881520920962829254091715364367892590360011330530548820466521384146951941511609433057270365759591953092186117381932611793105118548074462379962749567351885752724891227938183011949129833673362440656643086021394946395224737190702179860943702770539217176293176752384674818467669405132000568127145263560827785771342757789609173637178721468440901224953430146549585371050792279689258923542019956112129021960864034418159813629774771309960518707211349999998372978049951059731732816096318595024459455346908302642522308253344685035261931188171010003137838752886587533208381420617177669147303598253490428755468731159562863882353787593751957781857780532171226806613001927876611195909216420199
hpf('pi')
ans =
3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825342117067982148086513282306647093844609550582231725359408128481117450284102701938521105559644622948954930381964428810975665933446128475648233786783165271201909145648566923460348610454326648213393607260249141273724587006606315588174881520920962829254091715364367892590360011330530548820466521384146951941511609433057270365759591953092186117381932611793105118548074462379962749567351885752724891227938183011949129833673362440656643086021394946395224737190702179860943702770539217176293176752384674818467669405132000568127145263560827785771342757789609173637178721468440901224953430146549585371050792279689258923542019956112129021960864034418159813629774771309960518707211349999998372978049951059731732816096318595024459455346908302642522308253344685035261931188171010003137838752886587533208381420617177669147303598253490428755468731159562863882353787593751957781857780532171226806613001927876611195909216420199
I also ran it for 100000 digits, so 12500 terms. It took a little more time, but was entirely possible to compute. I don't recall which similar approximation I used some time ago, but I once used it to compute 1 million or so digits of pi in HPF. HPF currently stores a half million digits as I recall.
As far as understanding how to derive that series, I would leave that to Ramanujan, and only hope he is listening on on this.

Walter Roberson on 20 Oct 2018
If you are constructing an equation using the symbolic toolbox use sym('pi')
Steven Lord on 22 Oct 2021
That's correct. There are four different conversion techniques the sym function uses to determine how to convert a number into a symbolic expression. The default is the 'r' flag which as the documentation states "converts floating-point numbers obtained by evaluating expressions of the form p/q, p*pi/q, sqrt(p), 2^q, and 10^q (for modest sized integers p and q) to the corresponding symbolic form."
The value returned by the pi function is "close enough" to p*pi/q (with p and q both equal to 1) for that conversion technique to recognize it as π. If you wanted the numeric value of the symbolic π to some number of decimal places use vpa.
p = sym(pi)
p =
π
vpa(p, 30)
ans =
3.14159265358979323846264338328

Dmitry Volkov on 16 Dec 2022
Easy way:
format long
p = pi
Walter Roberson on 16 Dec 2022
That is what @Geoff Hayes suggested years before. But it does not enter π into the calculation, only an approximation of π

AKHIL TONY on 1 Aug 2023
using pi will give an approximate value
Walter Roberson on 1 Aug 2023
Yes, multiple people pointed that out years ago