Physics-Informed Compilation

SynapseX LLM

The translation layer bridging natural language optimization specs directly to hardware-level quadratic cost formulations.

Physical Hamiltonian Compilation

Select a mathematical model below to view the optimization energy function mapped onto QPU qubits.

H(σ) = - ∑ J_ij σ_i σ_j - ∑ h_i σ_i

The Ising model operates on binary spins σ_i ∈ {-1, +1}. It represents physical magnetic interactions, where:

  • J_ij represents the coupling strength between spin i and spin j. Positive values represent ferromagnetic alignment (spins want to align), while negative values represent antiferromagnetic frustration.
  • h_i represents the external magnetic field biasing the individual spin i.

The Translation Pipeline

1. Natural Prompt Input

The researcher inputs complex scientific problems or constraint parameters in human readable language: e.g. \"Optimize distribution centers to minimize transport latency under maximum cost limit.\"

2. Semantic Variable Mapping

SynapseX LLM parses the input, identifies optimization variables, establishes penalty matrices for hard constraints, and outputs structural matrix parameters.

3. Hardware Compilation

The compiled coupling factors (J_ij, h_i) are structured into QCOS SDK payloads and dispatched natively as magnetic couplings to physical trapped-ion qubits on-device.

Example QUBO Compilation Payload

SCHEMA v2.4
{
  "problem_type": "QUBO",
  "num_variables": 6,
  "linear_terms": {
    "x0": -1.5,
    "x1": 0.5,
    "x2": 2.0,
    "x3": -0.8,
    "x4": 1.2,
    "x5": -2.5
  },
  "quadratic_terms": {
    "x0_x1": 2.5,
    "x2_x3": -1.2,
    "x4_x5": 3.0,
    "x1_x4": -0.5,
    "x0_x5": 1.8
  },
  "qpu_constraints": {
    "max_interaction_strength": 5.0,
    "precision_bits": 16
  }
}

Ready to submit optimizations directly to private QPUs?

Download Desktop Console