CODEWITHSUNDEEP
matlabmatrixjulia
aberdysh
aberdysh

Reputation: 1664

Julia's equivalent of MATLAB's `sym`?

My research involves casting problems from scalar formulations to matrix formulations and vice versa. Sometimes finding corresponding matrix patterns and necessary operations to perform on them can get non-obvious and hard to visualize (especially when resulting matrix patterns are large and sparse). To validate my derivations I usually implement both formulations with MATLAB's sym variables (which allow performing all mathematical operations on them) and check if they are equal.

A trivial example of what I mean:

vec = sym('x',[2,1])
a = (3:4)'
vectorResult = a'*vec

scalar1 = sym('x1')
scalar2 = sym('x2')
scalarResult = a(1)*scalar1 + a(2)*scalar2

isequaln(vectorResult,scalarResult)

ans =

     1

So my question is there equivalent alternative for doing this in Julia?

At the moment this is the only thing (aside from the abscence of the MATLAB like IDE) that is preventing me from fully migrating to Julia.

Upvotes: 3

Views: 266

Answers (2)

Reza Afzalan
Reza Afzalan

Reputation: 5746

Here is a Julia equivalent form of the above example, with the help of SymPy package:

using SymPy # load SymPy package, you must Pkg.add("SymPy") before
n=10; # vector length 
vec=Sym[Sym(symbol(:x,i)) for i in 1:n]; # create the vector of Sym
a=rand(Int,n); # random vector of factors 
vectorResult= transpose(a)*vec; # matrix product
scalarResult=[sum([vec[i]*a[i] for i in 1:n])]; # scaler operation
scalarResult==vectorResult # => true

Upvotes: 3

StefanKarpinski
StefanKarpinski

Reputation: 33259

Base Julia does not ship with support for symbolic computation. For that functionality you can use something like SymPy.jl, which is a package for symbolic computation in Julia (via Python) or Nemo, which is a full computer algebra system based on Julia.

Upvotes: 4

Related Questions