There is SymPy.jl. The matrix bit could use some ironing out. By default, 
it uses an Array of symbolic objects, rather than a symbolic array. This 
means taking an inverse is bit cumbersome:

using SymPy

r = sympy.Rational 

m = [r(1,2) r(2,3); r(3,4) r(4,5)] ## Array of Sym objects

inv(m) ## converts to float, so no good

m = convert(SymMatrix, m)     ## SymMatrix object
m[:inv]()

Might be better to just interface directly with sympy through PyCall for 
this.

On Thursday, January 9, 2014 1:48:12 AM UTC-5, Alasdair McAndrew wrote:
>
> Since Julia plays so nicely with Python, why not just - at least in the 
> short term - create a package that interfaces with SymPy (
> http://sympy.org/en/index.html)?  This will certainly give you what you 
> want, and a great deal more besides.
>
> On Thursday, January 9, 2014 6:50:01 AM UTC+11, David Zhang wrote:
>>
>> Does Julia have any support for linear algebra on rational-valued 
>> matrices? Currently, most linear algebra functionality seems to be provided 
>> by BLAS/LAPACK, which only support operations on floating-point matrices.
>>
>> I am interested in this because I would like to use Julia to derive 
>> high-order numerical integration methods, whose coefficients are the 
>> solutions of rational linear system.
>>
>> (Continuation from 
>> http://stackoverflow.com/questions/20985783/rational-matrix-division-in-julia
>> )
>>
>>

Reply via email to