Convergence Diagnostics: Geweke vs. Heidelberger & Welch

[mliebano]

Hello, I have an issue with convergence diagnostics. I understand that convergence with the Geweke diagnostic is achieved when 𝑝 > 0.05. Could you please clarify this for me? Additionally, my model converges according to the Geweke diagnostic but not for the Heidelberger & Welch diagnostic. Moreover, numerically it shows convergence, but the plot does not seem to indicate convergence. Could you help me with this? Thank you.

`GEWEKE CONVERGENCE DIAGNOSTIC:

Fraction in first window = 0.1
Fraction in last window = 0.5

Chain: postgibbs

            i     var_a      var_e     var_p        ha

Z-Score -27.55983 1.0480840 -1.0664712 0.7819335 1.0089113
p-value 0.00000 0.2945999 0.2862107 0.4342537 0.3130172
`

`HEIDLEBERGER AND WELCH STATIONARITY AND INTERVAL HALFWIDTH TESTS:

Halfwidth test accuracy = 0.1

Chain: postgibbs

  Stationarity Test Keep Discard   C-von-M Halfwidth Test         Mean    Halfwidth

i failed 180 270 3.6128414 passed 3.380000e+05 2.117949e+04
var_a passed 450 0 0.1908887 failed 5.623096e-02 7.184450e-03
var_e passed 450 0 0.2084764 passed 1.833262e-01 6.924330e-03
var_p passed 450 0 0.1693914 passed 2.395572e-01 1.031661e-03
ha passed 450 0 0.1846487 failed 2.345089e-01 2.979838e-02
`