Theory and Exactness Scope
This chapter defines what SchroSIM means by "exact" and where that claim stops.
1) Mathematical Conventions
SchroSIM’s CV Gaussian model uses:
- quadrature ordering
R = (q1, p1, q2, p2, ...), - units
hbar = 1, - Gaussian state parameterization by mean
dand covarianceV, - symplectic form
Omega = direct_sum_k [[0,1],[-1,0]].
Formal derivations are in 11-mathematical-proofs.md.
2) Exactness Taxonomy in SchroSIM
2.1 Model-Exact (Gaussian Core)
Within the Gaussian model assumptions, updates are exact identities:
- affine symplectic evolution:
d' = S d + uV' = S V S^T- Gaussian channels (loss, thermal-loss) in closed form,
- linear-Gaussian conditioning for homodyne and heterodyne updates.
This is the exactness regime referenced in the README for tractable workflows.
Terminology used across docs:
- model-exact: exact with respect to the implemented Gaussian model equations,
- controlled approximation: bounded/explicit approximation regimes (cutoff, truncation, routed non-Gaussian paths).
2.2 Deterministic Runtime Reproducibility
Given the same:
- circuit payload,
- seed,
- backend selection and runtime configuration,
- foundry profile/policy inputs,
SchroSIM targets deterministic replay at the result/trace level.
2.3 Approximation Regimes
Model-exactness does not imply exact infinite-dimensional quantum dynamics in every mode of operation. Controlled approximation enters when:
- Fock simulations use finite cutoff dimension,
- matrix exponential operators are series-truncated numerically,
- effective/derived non-Gaussian handling paths are used for practical workflows,
- backend routing selects an alternate compatible execution path.
3) Physicality and Validity Conditions
Core validity conditions include:
- uncertainty compatibility:
V + i Omega/2 >= 0,- channel parameter domains:
eta in [0,1],n_th >= 0,- IR parameter finiteness and mode-index validity,
- foundry policy constraints (
max_modes, squeezing limits, measurement/non-Gaussian allowances).
4) Claim Boundary
SchroSIM claims:
- exact implementation of the stated Gaussian model equations,
- explicit policy-aware execution and diagnostics,
- deterministic, reproducible runtime behavior under fixed inputs/configuration.
SchroSIM does not claim:
- exact simulation of arbitrary infinite-dimensional non-Gaussian dynamics without truncation,
- hardware-faithful behavior without calibrated model assumptions,
- approximation-free Fock evolution at finite cutoff.
5) Practical Interpretation for Users
Use this decision rule:
- For Gaussian and tractable circuits: treat SchroSIM results as model-exact.
- For Fock/non-Gaussian-heavy or very large circuits: treat results as controlled approximations and report cutoff/backend settings with outputs.
- For pre-hardware studies: always bind runs to foundry profile/policy identifiers and seeds.
6) Cross-References
- Proofs: 11-mathematical-proofs.md
- Backends: 06-execution-backends.md
- Compiler/foundry constraints: 05-compiler-and-foundry.md