QEncode Benchmark
Back to leaderboard

NH3

/jordan wigner/uccsd tapered
Certified
Beats CCSD(T)
cc-pvdz

NH3_ccpvdz_JW_UCCSD_v4_tapered__sha256_bff85ae0c8311f46

VQE Gap

3.2445e-5 Ha

CASCI ref

-56.197889 Ha

Qubits (tapered)

5

CCSD(T) corr.

-2.085e-1 Ha

Molecule / Problem

NameNH3
Basis setcc-pvdz
GeometryN 0.0 0.0 0.1173; H 0.0 0.9289 -0.2737; H 0.8045 -0.4645 -0.2737; H -0.8045 -0.4645 -0.2737
Charge0
Spin0
Active electrons4
Active orbitals4
Orbital optimizationHF

Encoding / Circuit

Qubit mappingjordan-wigner
Ansatz typeuccsd_tapered
Ansatz reps (HEA)1
TaperingZ2 symmetry — 3 symmetries removed
Qubits (original)8
Qubits (tapered)5
Num parameters64
Circuit depth120
2Q gates24
Non-Clifford gates125
T gates (FT estimate)1,500 (ε = 0.001)

T-gate count is the resource-relevant cost of a fault-tolerant implementation. It is an estimate from the non-Clifford rotations in the pre-transpilation circuit, assuming standard rotation synthesis to precision ε — not a compiled count for a specific device.

Note on UCCSD circuit metrics

UCCSD uses exponential Pauli operators (exp(iθH)) that are symbolic before compilation. Circuit depth and 2-qubit gate counts are not reported here — they depend on the target hardware and transpiler.

Energy Results

HF energy-56.1957314939 Ha
MP2 energy-56.3845132569 Ha
CCSD energy-56.4004649816 Ha
CCSD(T) energy-56.4042812742 Ha
CASCI ground state-56.1978885887 Ha
VQE best energy-56.1978561435 Ha
|VQE − CASCI| gap3.244515e-5 Ha
CCSD(T) correlation-2.085498e-1 Ha

Optimizer / Run Config

AlgorithmCOBYLA
Multistart runs1
Max iterations500
Total func evals500
Num parameters64
Backenddefault.qubit

Provenance / Reproducibility

Entry IDNH3_ccpvdz_JW_UCCSD_v4_tapered__sha256_bff85ae0c8311f46
Schema version4.0.0
Created (UTC)2026-07-15T13:22:23.704023+00:00
SHA-256 hashbff85ae0c8311f460149c33ceccd60b01fca644f3aea689401b6c3991bf463da
Python3.11.15
PySCF2.6.2
PennyLane0.45.0
OpenFermion1.6.1
NumPy2.2.6
SciPy1.13.1

Reproduce this result

# Clone repo and install pinned environment

git clone https://github.com/qencode-benchmark/qencode-benchmark

cd qencode-benchmark

pip install -r requirements-v4.txt

# Run this entry

python scripts/generate_entry_v4.py \

--molecule NH3 \

--mapping jordan_wigner \

--ansatz-type uccsd \

--multistart 1 \

--max-iter 500 \

--out-dir releases/v4/db