Problem with obtaining optimal rho estimate

[VHarisop]

Hello,

I have been trying to obtain the optimal estimate of the rho parameter from the solver but ran into the following issue. The field info.rho_estimate from the result of OSQP.solve!(...) does not match the value reported in the solver log.

I have created this minimal gist, where I solve the AUG3D instance from the Maros-Meszaros dataset. The solver log is as follows:

-----------------------------------------------------------------
           OSQP v0.6.2  -  Operator Splitting QP Solver
              (c) Bartolomeo Stellato,  Goran Banjac
        University of Oxford  -  Stanford University 2021
-----------------------------------------------------------------
problem:  variables n = 3873, constraints m = 1000
          nnz(P) + nnz(A) = 9219
settings: linear system solver = qdldl,
          eps_abs = 1.0e-03, eps_rel = 1.0e-03,
          eps_prim_inf = 1.0e-04, eps_dual_inf = 1.0e-04,
          rho = 1.00e-01 (adaptive),
          sigma = 1.00e-06, alpha = 1.60, max_iter = 4000
          check_termination: on (interval 25),
          scaling: on, scaled_termination: off
          warm start: on, polish: off, time_limit: off

iter   objective    pri res    dua res    rho        time
   1  -4.2768e+02   1.00e+00   1.48e+02   1.00e-01   4.07e-03s
  25   4.3955e+02   3.08e-04   7.00e-04   1.00e-01   6.36e-03s

status:               solved
number of iterations: 25
optimal objective:    439.5500
run time:             6.41e-03s
optimal rho estimate: 1.37e-01

However, when I do println(result.info.rho_estimate), the value is either something really close to zero (or sometimes NaN).

I'm running Julia v1.6.1 on 64-bit Linux. Any help is appreciated. Thanks!