Premium Components

Premium components are high-fidelity models served from the Hardwave cloud. They integrate into your simulation graph exactly like stdlib components (same port and parameter API, same graph wiring, same engine), but the solver runs on Hardwave servers. Use premium components for complex physics, proprietary lookup tables, or ML-based models without shipping proprietary code in your repository.

Authentication

Authentication uses RSA-SHA256 signatures. The Hardwave dashboard generates a 2048-bit RSA key pair for your organization and gives you the private key (PEM). Your public key is stored server-side. Every API call is signed with your private key. Your private key is never sent over the wire and is never stored on Hardwave servers.

Setup

Get your credentials

Open the Hardwave dashboard and copy your organization ID and private key PEM.

Configure and sync

Use like stdlib

Add premium components to any SimulationGraph and run normally.

Configure in code

import hardwave
import hardwave.stdlib
import hardwave.premium

# Load the PEM from a file; never commit it to source control
private_key = open("hardwave_private.pem").read()

hardwave.premium.configure(
    organization_id="<your-org-uuid>",
    secret_key=private_key,
)
count = hardwave.premium.sync()
print(f"Registered {count} premium components")

Configure with environment variables

export HARDWAVE_ORGANIZATION_ID="<your-org-uuid>"
export HARDWAVE_SECRET_KEY="$(cat hardwave_private.pem)"

hardwave.premium.configure()  # reads from env automatically
hardwave.premium.sync()

Optional: set HARDWAVE_SERVER_URL to override the default server URL (https://hardwave.dev).

Using premium components

Once sync() has run, premium components are in ComponentRegistry and behave identically to stdlib components:

from hardwave.components import ComponentRegistry
from hardwave.simulation import SimulationGraph, SimulationEngine

ThermalModel = ComponentRegistry.instance().get("AdvancedThermalModel")

graph = SimulationGraph()
graph.add_component(ThermalModel("mosfet_thermal", param_values={
    "thermal_resistance_jc": 3.2,
    "thermal_resistance_ca": 15.0,
}))
engine = SimulationEngine(graph)
result = engine.run(inputs={
    "mosfet_thermal": {"power_dissipation": 4.5, "ambient_temp": 35.0},
})
print(result.get_output("mosfet_thermal", "junction_temp"))

When the engine calls solve() on a premium component, it transparently dispatches to the Hardwave server. Your private key signs each request and simulation results come back exactly as if the solver ran locally.

Hosted ML models

You can also create org-owned ML models, append training records, and train versioned solvers in the cloud. See Hosted ML models.

Available premium components

Advanced Thermal Model

Multi-mode junction-to-ambient heat transfer for power semiconductors

Error handling

Exception	Cause
`PremiumNotConfiguredError`	`configure()` not called before `sync()`
`PremiumAuthError`	Server rejected your credentials
`PremiumSyncError`	Component manifest could not be fetched
`PremiumSolveError`	Remote solve request failed

See Error Handling for the full exception hierarchy.

Never commit your private key PEM to version control. Load it from a file or environment variable.

​Authentication

​Setup

​Configure in code

​Configure with environment variables

​Using premium components

​Hosted ML models

​Available premium components

Advanced Thermal Model

​Error handling

Authentication

Setup

Configure in code

Configure with environment variables

Using premium components

Hosted ML models

Available premium components

Error handling