thanks for adding the anderson acceleration, it is great.
it works on regular problem, but i encountered this issue on a trivial one:
x = cvx.Variable()
cvx.Problem(cvx.Minimize(x), [x>=0]).solve(solver=cvx.SCS, verbose=True)
which gives
----------------------------------------------------------------------------
SCS v2.0.0 - Splitting Conic Solver
(c) Brendan O'Donoghue, Stanford University, 2012-2017
----------------------------------------------------------------------------
Lin-sys: sparse-indirect, nnz in A = 1, CG tol ~ 1/iter^(2.00)
eps = 1.00e-06, alpha = 1.50, max_iters = 5000, normalize = 1, scale = 1.00
acceleration_lookback = 20, rho_x = 1.00e-03
Variables n = 1, constraints m = 1
Cones: linear vars: 1
Setup time: 1.31e-04s
----------------------------------------------------------------------------
Iter | pri res | dua res | rel gap | pri obj | dua obj | kap/tau | time (s)
----------------------------------------------------------------------------
0| 1.33e+00 1.00e+03 5.71e-01 -1.33e+00 -0.00e+00 1.71e-16 1.03e-04
posv ran into numerical issues, info 4, switching to gels
20| 8.03e-20 0.00e+00 4.62e-20 4.62e-20 -0.00e+00 1.17e-19 3.56e-02
python(37729,0x7fffba43a3c0) malloc: *** error for object 0x7f7f7c2cf928: incorrect checksum for freed object - object was probably modified after being freed.
*** set a breakpoint in malloc_error_break to debug
Abort trap: 6
ghost
commented
6 years ago
Brendan,
thanks for adding the anderson acceleration, it is great.
it works on regular problem, but i encountered this issue on a trivial one:
which gives