how to change the precision of symbolic variables.

s^3*246655351680430349161861489360895748665691918000757954295365632i - s^2*(36212255157820456428489768408649977023491870935380775514669056 + 578091697775954585022159881976519846765626019509350787237019648i) - s*(130062300639770074374222879395026417216202027377365642475732992 + 861105664727459103565145461543508318180271162190507702581460992i) - (223478472834610436324352002622703850555766990429849990494534772 + 585856833391770884939511626951491940256735302346867433027453683i)

I need it to be 6 digit precised.

Answer 1

[Star Strider]

Use the vpa function:

syms s 
p = s^3*246655351680430349161861489360895748665691918000757954295365632i - s^2*(36212255157820456428489768408649977023491870935380775514669056 + 578091697775954585022159881976519846765626019509350787237019648i) - s*(130062300639770074374222879395026417216202027377365642475732992 + 861105664727459103565145461543508318180271162190507702581460992i) - (223478472834610436324352002622703850555766990429849990494534772 + 585856833391770884939511626951491940256735302346867433027453683i);
p_vpa = vpa(p, 6)

producing:

p_vpa =
s^3*2.46655e+62i - s^2*(3.62123e+61 + 5.78092e+62i) - s*(1.30062e+62 + 8.61106e+62i) - (2.23478e+62 + 5.85857e+62i)

The expression retains full internal precision, so nothing is lost.

[Sumit dash]

actually i need it to show me only

s^3*2.46655 - s^2*(0.362123 + 5.78092i) - s*(1.30062 + 8.61106i) - (2.23478 + 5.85857i)

Can it be posssible ?

[Star Strider]

It can.

Try this:

p_vpa = vpa(p*1E-62, 6)

producing:

p_vpa =
s^3*2.46655i - s^2*(0.362123 + 5.78092i) - s*(1.30062 + 8.61106i) - (2.23478 + 5.85857i)

Remember the 1E-62 scaling factor if you intend to use these results in other computations.

[Sumit dash]

when we will be using such kind of polynomials for multiple no of times, it is not possible to write everytime this code.

So is there any method where we can declare at the beginning of a program to take only 6 significant digits or 5 significant digits, but not more than that.

[Star Strider]

Use the digits function.

If you want to automatically determine what I call the ‘scaling factor’, use this:

scaling_factor = vpa(10.^-fix((log(coeffs(p))/log(10))))

where ‘p’ is your polynomial of interest.

Experiment to get the result you want.