How big is the polynomial that sym2poly can hold?

How big is the polynomial that sym2poly can hold?

For example, I have this y=sym2poly(sym('x^1000'))

and it's ok. But if I increase the number to x^10000000000, Matlab crashes. How can I avoid this?

Answer 1

[Walter Roberson]

Use 64 bit MATLAB on a system that has at least 80 gigabytes of memory.

Suppose that the sym2poly worked, then the vector returned would have to have 10000000001 elements, each a double precision number. That would be 80000000008 bytes, which is about 75 gigabytes.

The limitation you are encountering isn't sym2poly itself: the limitation is the amount of memory you have.

Answer 2

[Sean de Wolski]

2.8823e+17

[sadel]

:)

[Walter Roberson]

Hmmm... not 3.5184e+13 or 2.8147e+14? There is evidence for 48 total in (e.g.) the documentation for sparse(), but I don't recall seeing any evidence for 58 or 61 ?

[Sean de Wolski]

I was thinking (2^64)/64?

[Walter Roberson]

"64 bit" systems could theoretically address up to 64 bytes; each double occupies 8 bytes. In practice, http://en.wikipedia.org/wiki/X86-64 says that only the lower 48 bits are used in current implementations. That would imply a limit of (2^48)/8 = 2^45 double precision values in a single array, if there was nothing else in the address space.

[Sean de Wolski]

Per usual, you win.

[Walter Roberson]

Did I say "up to 64 bytes" ? I meant "up to 2^64 bytes".

Maybe some day we'll get to edit comments...