sympy fails to solve a linear system of first order ODEs

[maeshtro]

I tried to solve the following initial value problem

y1'   =    y1 + 2 y2,
y2'   = -2 y1 +   y2 + 2 exp(t),
y1(0) =  y2(0) = 1

with sympy using the code

import sympy as sy
t=sy.symbols('t')
y1=sy.Function('y1')   
y2=sy.Function('y2')
eqs=(sy.Eq(y1(t).diff(t),y1(t)+2*y2(t)), sy.Eq(y2(t).diff(t),-2*y1(t)+y2(t)+2*sy.exp(t))) 
s=sy.dsolve(eqs)        # General solution
y1g=s[0].args[1]
y2g=s[1].args[1]
# Find C1 and C2 so that the initial condition is satisfied
sol=sy.solve([y1g.subs(t,0)-1,y2g.subs(t,0)-1])
y1=y1g.subs(sol)
y2=y2g.subs(sol)
print([y1,y2])

The solution I get is

[2*(sin(2*t)/2 + cos(2*t)/2)*exp(t), (-sin(2*t) + cos(2*t))*exp(t)]

which is wrong. The correct solution is

[exp(t)*(1+sin(2*t)), exp(t)*cos(2*t)]

I am running python 3.3.2 with sympy 0.7.3 under fedora 19 x64.

[asmeurer]

Your version of SymPy is ancient. Please upgrade to 0.7.6.

Even so, I get the same answer in 0.7.6.

[oscarbenjamin]

On current master I get NotImplementedError for this system:

NotImplementedError                       Traceback (most recent call last)
~/current/sympy/sympy/tmp/dsolvesys.py in <module>
      4 y2=sy.Function('y2')
      5 eqs=(sy.Eq(y1(t).diff(t),y1(t)+2*y2(t)), sy.Eq(y2(t).diff(t),-2*y1(t)+y2(t)+2*sy.exp(t)))
----> 6 s=sy.dsolve(eqs)        # General solution
      7 y1g=s[0].args[1]
      8 y2g=s[1].args[1]

~/current/sympy/sympy/sympy/solvers/ode.py in dsolve(eq, func, hint, simplify, ics, xi, eta, x0, n, **kwargs)
    618             "number of functions being equal to number of equations")
    619         if match['type_of_equation'] is None:
--> 620             raise NotImplementedError

This is what the equations look like:

In [4]: eqs                                                                                                                                    
Out[4]: 
⎛d                            d                                 t⎞
⎜──(y₁(t)) = y₁(t) + 2⋅y₂(t), ──(y₂(t)) = -2⋅y₁(t) + y₂(t) + 2⋅ℯ ⎟
⎝dt                           dt                                 ⎠