Batched chain greeks

SIMD-batched Black-Scholes price and greeks across a whole option chain in one call, matching the scalar pricing functions to 1e-9.

option_greeks_batch prices an entire option chain — hundreds to thousands of strikes and expiries — in a single call, vectorizing the Black-Scholes math across hardware SIMD lanes. It replaces per-contract scalar pricing loops where call overhead and scalar transcendentals dominate.

The modules

Module	What it does
option_greeks_batch	SIMD-batched Black-Scholes price, delta, gamma, vega and theta for an entire option chain in one call
option_pricing	the scalar Black-Scholes, binomial-tree, greeks and implied-volatility functions that option_greeks_batch is verified to match to 1e-9

Functions

Function	What it does
chain_greeks(spots, strikes, times, rates, sigmas, is_calls)	Batched BS price, delta, gamma, vega and theta for mixed call/put chains (is_calls 1.0/0.0). Returns ChainGreeksResult with five parallel List[Float64] outputs.

Semantics and guarantees

• Results match the scalar option_pricing functions to 1e-9 on the valid domain (verified across a 144-cell hostile grid of moneyness, expiry and volatility, calls and puts).
• Put greeks derive from parity — delta_put = delta_call - 1, theta_put = theta_call + r·K·e^(-rT); gamma and vega are shared. Theta is per year, the same convention as theta_call.
• Degenerate lanes (T <= 0 or sigma <= 0) return discounted intrinsic value with zero greeks instead of NaN.
• The SIMD main loop and scalar tail produce identical results at every batch size (verified for n = 1..11 against single-element calls).

Throughput

Measured on a 2-lane NEON float64 machine (wider AVX lanes scale further), with dead-code elimination defeated by checksum accumulation:

1,000-contract chain: batch 28.2 us | scalar loop 88.1 us | 3.1x
= 28 ns per contract for price + all four greeks

Next: feed the greeks into statistical model comparison experiments, or browse the engine catalog.