Struct VecEnv 

pub struct VecEnv { /* private fields */ }

Vectorized market-making environment.

Manages N independent engine instances. On each step():

1. Read actions (N, 2) as (bid_distance, ask_distance)
2. For each env: set orders, advance engine, compute reward
3. Return (states, rewards, dones) as contiguous f32 arrays
4. Auto-reset finished environments

Implementations

impl VecEnv

pub fn new(config: VecEnvConfig, num_envs: usize) -> Self

pub fn num_envs(&self) -> usize

Number of environments.

pub fn reset(&mut self) -> &[f32]

Reset all environments and return initial states.

pub fn step(&mut self, actions: &[f32]) -> (&[f32], &[f32], &[f32])

Step all environments with the given actions.

actions must have length num_envs * 2, laid out as [bid_dist_0, ask_dist_0, bid_dist_1, ask_dist_1, ...].

Returns (&states, &rewards, &dones) where:

• states: [f32; N*6] flattened row-major
• rewards: [f32; N]
• dones: [f32; N] (1.0 if episode ended, 0.0 otherwise)

pub fn get_states(&self) -> &[f32]

Current states buffer (read-only).

pub fn get_equities(&self) -> &[f32]

Per-env equity after the last step (captured before auto-reset).

pub fn get_positions(&self) -> &[f32]

Per-env inventory after the last step (captured before auto-reset).

pub fn get_step_fills(&self) -> &[f32]

Per-env fill count from the last step (captured before auto-reset).