Web.batch?

Web.batch?#

Date

Category

2025-09-16 13:50:36

Web API

Batch is a container for submitting, running, monitoring, and downloading multiple simulations on the Tidy3D cloud in one go. It’s similar to a web.Job, but for a whole set of simulations (FDTD, Heat/Charge, EME, Mode solver, etc.) that run in parallel.

Estimating cost#

It is possible to estimate the maximum cost of the whole batch with the Batch.estimate_cost method. For more information on simulation cost, check this article.

Running a batch#

The batch is run with the Batch.run(path_dir="path_dir") method, which saves a batch file that can be loaded later and returns a BatchData object.

Loading a batch#

A batch can be loaded using the Batch.from_file method. If the Batch.run method was previously used, the Batch object will contain information about all executed tasks, and the BatchData object can be returned with the Batch.load method. For more information, refer to this article.

When should I use it?#

  • Parameter sweeps, design-of-experiments, or any workflow with many related runs in parallel.

Minimal example#

python
import tidy3d as td
from tidy3d.web.api.container import Batch

# 1) Build your simulations (FDTD shown as example)
sim_a = td.Simulation(...)  # define as usual
sim_b = td.Simulation(...)
sims = {"run_a": sim_a, "run_b": sim_b}

# 2) Create a Batch
batch = Batch(
    simulations=sims,
    folder_name="my_sweep",
)

# (Optional) Quick cost estimate (max, assuming full run_time)
est_fc = batch.estimate_cost(verbose=True)

# 3) Run (upload+start+monitor+download as needed)
data = batch.run(path_dir="results")

# 4) Iterate over results (lazy-loaded from disk, downloads on demand)
for name, sim_data in data.items():
    print(name, sim_data)
    # ... analyze sim_data ...

# 5) Access final billed cost later (after runs finish)
billed_fc = batch.real_cost(verbose=True)
print("Billed FlexCredits:", billed_fc)
Contents