How to compute a new Double Dummy score?

[Amir-Arsalan]

I want to integrate the simulator in a look where my model chooses the next action and passes it to the simulator. I was assuming that the simulator would calculate a new Double Dummy score for a new set of actions but it turns out that is not the case. How would you suggest one should calculate a new Double Dummy score after the bidding is over? The code below shows that the model has chosen a Double Dummy score in advance even before a bidding which is a big puzzling for me although I know that I could be wrong especially since I'm still learning about how the simulator works.

import jax, pgx
from pgx.bridge_bidding import BridgeBidding
from pgx.bridge_bidding import download_dds_results
from pgx.experimental.utils import act_randomly


download_dds_results()
env = BridgeBidding(dds_results_table_path = "dds_results/dds_results_10M.npy")
# model = env.make_baseline_model('')

init = jax.jit(jax.vmap(env.init))  # vectorize and JIT-compile
step = jax.jit(jax.vmap(env.step))
act_randomly = jax.jit(act_randomly)

batch_size = 1

# prepare PRNGKeys
key = jax.random.PRNGKey(44)
key, subkey = jax.random.split(key)
keys = jax.random.split(subkey, batch_size)

state = init(keys)  # vectorized states

print(state._dds_val)

while not (state.terminated | state.truncated).all():
    action = act_randomly(subkey, state.legal_action_mask)
    state = step(state, action)  # state.reward (2,)

print(state._dds_val)

The code above will print the following:
[[467474 436923 467474 437179]]
[[467474 436923 467474 437179]]

Which seems that the simulator assumes the Double Dummy score is guaranteed to come from a set of 10 million random bidding results. So does this mean that the Bridge Bidding code does not take into account the possibility that one might use a policy network to produce the actions? How should one modify the code so that the Double Dummy score is calculated on fly when using a policy network to drive the simulation?