How does matlab improve matrix storage accuracy

Hi,everyone,I encountered a problem. The number in a matrix is too large and exceeds the storage space of 16 bits. I want to expand the storage space of the matrix. It is best if it can be expanded to 200 bits. I know that the vpa function can help us perform high-precision simple operations, but storing high-precision data in a matrix will store double type data by default. Is there any way to expand the number stored in the matrix to 200 digits? The amount of calculation is not a problem.

this is a 10 order martix.

i need calculate 20 order martix.the maximum date of the 20 order martix can reach 10 to the 200th power

Answer 1

[Walter Roberson]

storing high-precision data in a matrix will store double type data by default.

The data type of an array is determined by the first value you store to the array.

If you initialize the array

M = zeros(10,10);

then that creates M as a double precision array, and if you attempt to store a symbolic number into part of M, then the symbolic number will be converted to double precision.

So instead initialize the array as symbolic:

M = zeros(10, 10, 'sym');

After which, for example,

M(1,1) = vpa(exp(sym(pi)),200);

Note that the full 200 digits will not be displayed, but you can tell with calculations that they are stored internally.

[weiting peng]

thank you a lot. This method really works