quantum.harness

Use when the user wants a numerical result (ED / DMRG / QMC / VMC / any method) checked against an exact solution, or names a model from the solvable-models catalog in a verification context. Trigger phrases: "verify against the exact solution", "check my DMRG/ED/QMC result", "benchmark <method> on <model>", "is this energy right", "does this match the exact answer", "sanity-check this against Bethe ansatz / Onsager / free-fermion solution". Also fires when the user names any of the 63 catalog models below in a verification context — not only when they use the word "verify": - T1 quadratic/free-particle: tfim-chain (TFIM), xy-chain, kitaev-chain, ssh-chain, kitaev-honeycomb, hofstadter-harper, haldane-chern, anderson-1d, harmonic-chain - T2 2D classical/transfer matrix: ising-2d-onsager (Onsager / 2D Ising), ising-triangular, dimer-kasteleyn, six-vertex, eight-vertex, hard-hexagons - T3 Bethe ansatz/Yang–Baxter: xxz-chain, heisenberg-xxx, hubbard-1d-lieb-wu (Lieb–Wu / 1D Hubbard), lieb-liniger, yang

Use when an exact diagonalization track, ED reproduction, full spectrum, symmetry sector, scar, ETH, level statistics, quench dynamics, finite-temperature ED (FTLM/TPQ), spectral function, interior-spectrum / MBL eigenstates, or finite-cluster oracle needs method-level route and tool selection.

Use when the user names or describes a harness-tracked quantum lattice model. Match user prose to one of: - transverse-field-ising (TFIM): quantum-critical Ising chain / 2D Wilson-Fisher - heisenberg: SU(2) magnet, AFM or FM by sign of J - xxz-chain: spin-1/2 XXZ, Bethe-ansatz integrable, Δ tunes FM / Luttinger / Néel - j1-j2: frustrated Heisenberg, J2/J1≈0.5 spin-liquid candidate - shastry-sutherland: orthogonal-dimer AFM, exact dimer phase, magnetization plateaus - spin-1-xxz: Haldane phase, single-ion anisotropy - aklt: spin-1 bilinear-biquadratic, exact VBS ground state - kitaev-honeycomb: bond-dependent exchange, exactly solvable Z2 spin liquid, anyons - spin-ice-pyrochlore: 2-in-2-out ice rule, Coulomb phase, magnetic monopoles - mbl-disordered-heisenberg: random-field XXZ chain, ETH-to-MBL transition - rydberg-pxp: blockaded Rydberg chain, quantum many-body scars, |Z2⟩ revivals - dissipative-spin-lindblad: open spin lattice, Liouvillian spectrum, steady states - potts-clock: q-state, first-order / cont

Use when the user wants to reproduce a paper's figure or main result. Triggers include "reproduce paper X", "redo the figures of Y", "reproduce arXiv <id>", "put this paper through the harness as a calibration target", "walk me through reproducing this paper", "beginner reproduction", "I don't know what size to choose", "explain while reproducing", or right after `/download-ref` lands a new paper.

Use when choosing or running QuSpin as the Python fallback for exact diagonalization, constrained-basis ED, spin-chain ED examples, or QuSpin setup failures.

Use when choosing or running XDiag.jl for exact diagonalization, symmetry-resolved sectors, Lanczos/Krylov calculations, or XDiag setup failures.

Generate a challenge completion report from a finished run in tracks/. Checks submission cleanliness first, then walks through each block interactively.

Use to render a run's HTML report — a reproduction proposal before compute, results after, or any structured run summary. Triggers include "render report", "build the report", "publish results", "share results", or a skill like `/reproduce-paper` composing its report page.

Use when the user brings a concrete quantum many-body research problem — phrases like "ground state of Heisenberg N = 20", "is kagome a spin liquid", "Hubbard at U/t = 8", "compute the gap for TFIM at h = 1", "what does VMC give for J1-J2 at 0.5". Also fires when `/onboard` routes here or when the user pivots to a new problem mid-session.

Use when a new harness user (assume undergraduate level) starts the beginner training and needs to choose which track to run — asks which of the five tracks (setup check, reproduce a paper, literature survey, develop code like an expert, go beyond) to take, then walks that track as a guided, confirm-gated tutorial that explains every step and waits for the student to confirm understanding before running anything. Triggers on "start the training", "beginner training", "/beginner-training", "which track should I take", "I'm new here, teach me the harness".

Use when a VMC, variational Monte Carlo, neural quantum state, NQS, NetKet, ansatz, sampler, or optimizer reproduction needs method-level route and tool selection.

Use when a DFPT electron-phonon coupling and superconductivity calculation needs method-level route and tool selection — native PHonon/DFPT workflow for alpha2F(omega), lambda, omega_log, McMillan/Allen-Dynes Tc, q/k convergence. This skill owns the method understanding, method-vs-method routing (DFPT vs EPW), and the knob JUDGMENT. This is for QE PHonon workflows, not EPW.

Use when a quantum Monte Carlo (QMC) reproduction needs method-level route and tool selection — stochastic series expansion (SSE) for sign-free spin/boson finite-temperature targets, constrained-path / phaseless auxiliary-field QMC (CPMC/AFQMC) for fermionic Hubbard ground states, or fixed-node diffusion Monte Carlo (DMC) for plane-wave solids using Quantum ESPRESSO orbitals. Triggers include "QMC", "SSE", "stochastic series expansion", "finite-temperature susceptibility", "sign problem", "AFQMC", "CPMC", "constrained path", "Hubbard ground state by Monte Carlo", "DMC", "diffusion Monte Carlo", "QMCPACK", "Slater-Jastrow", "twist averaging", "fixed-node".

Use when using QMCPACK to prepare, run, monitor, debug, or interpret plane-wave solid Slater-Jastrow VMC and DMC workflows, including Slater determinant setup, Jastrow optimization, twist averaging, fixed-node DMC, timestep checks, scalar analysis, and MPI or twist-parallel launch tuning.

Use when using Quantum ESPRESSO's native PHonon/DFPT package workflow to prepare, run, monitor, debug, or interpret electron-phonon coupling and superconductivity calculations involving pw.x, ph.x, q2r.x, matdyn.x, alpha2f.x, lambda_tetra, q/k convergence, or PHonon scratch failures. This skill owns the QE package run mechanics, literal input keywords, and templates. Use when using Quantum ESPRESSO to generate plane-wave orbitals for a Slater-Jastrow VMC and DMC workflow, including QE SCF/NSCF and pw2qmcpack.x ESHDF conversion, and MPI launch tuning.

Use when a student has finished reproducing their track's reference paper and wants to take on a challenge — go beyond the reference and ship a PR. Drives challenge ideation (literature survey → brainstorm → write-up), a help-desk go/no-go, the attempt, then hands off to /challenge-report. Triggers include "I reproduced the paper, now what", "take on a challenge", "go beyond the reproduction", "what should I try next on this track", "/challenge", or right after /reproduce-paper completes a run.

Use when working with harness-relevant scientific/physics software APIs, local KB or reference snippets, examples, package capabilities, setup, migrations, or hard-to-find package documentation before writing code.

Use when first-touch harness setup is needed: user is new, asks where or how to start, invokes `/onboard`, opens with an empty or unclear prompt, or has no configured harness environment.

Use when a student needs a reproducible software environment for cluster jobs and must build + validate an Apptainer/Singularity container (.sif) for the harness stack before submitting — covers writing a pinned definition file, building with --fakeroot, validating Julia/Python/GPU inside the image, and proving it on a compute node. Triggers on "build a container", "make a Singularity/Apptainer image", "containerize the stack", "set up a reproducible env for the cluster".

Use when a computation needs to run on a remote Slurm cluster (CPU- or GPU-heavy beyond laptop budget), when a multi-cell array job is ready to submit, when a submitted job needs status / monitoring / cancel / log-tail, or when only a few cells failed and the user wants to resume — phrases like "send this to the cluster", "sbatch this", "submit on HPC2", "check job status", "resubmit failed cells".

Use when a finite-temperature Linearized Tensor Renormalization Group (LTRG) reproduction needs method-level route and tool selection — Trotterized classical tensor network from a quantum lattice model, layer-by-layer boundary contraction with SVD truncation, thermodynamic observables (free energy, internal energy, specific heat, susceptibility).

Use when a Monte Carlo Renormalization Group reproduction or calculation needs method-level route and tool selection — real-space block-spin RG on lattice spin models, extracting renormalized coupling constants and critical exponents from the eigenvalues of the RG Jacobian. Covers Swendsen's classic MCRG and the variational (bias-potential) MCRG that removes critical slowing down.

Use when a mean-field reproduction needs method-level routing — Hartree-Fock / unrestricted Hartree-Fock for lattice fermions, Weiss mean-field decoupling of spin models, self-consistent-field order parameters and phase diagrams.

Use when an MPS / matrix-product-state reproduction or calculation needs method-level route and tool selection — ground states, order parameters, correlations, gaps, dynamics, or finite-T of 1D and quasi-1D (narrow-cylinder) Hamiltonians, in either the finite-chain regime (DMRG, TEBD) or the infinite / uniform regime (VUMPS, IDMRG, iTEBD, TDVP). Covers picking the algorithm and routing to the tool.

Use when a PEPS, iPEPS, CTMRG, 2D tensor-network, environment dimension, or classical partition-function reproduction needs method-level route and tool selection.

Use when a noncommutative polynomial optimization reproduction needs method-level route and tool selection — certified lower bounds on ground-state energy via the moment-SOS / SOHS (NPA-style) hierarchy solved as a semidefinite program, Bell-inequality maximum quantum violation, or state-polynomial / tracial optimization. Triggers include polynomial optimization, SOS / SOHS relaxation, moment-SOS hierarchy, NPA hierarchy, certified energy lower bound, Bell inequality, semidefinite relaxation, NCTSSoS.

Use when a quantum circuit simulation, VQE, TensorCircuit-NG, JAX backend, contraction path, statevector, MPS circuit, gradient, or circuit-performance reproduction needs method-level route and tool selection.

Use when installing, smoke-testing, locating, invoking, or choosing run parameters for the official CPMC-Lab Matlab package; covers MATLAB path setup, package-root discovery, batch execution, download/license handling, package output extraction, and the documented CPMC_Lab parameter set (meaning, constraints, setup strategy).

Use when choosing or running ITensors.jl or ITensorMPS.jl for DMRG, TEBD, MPS calculations, tensor-network checks, or ITensors setup failures.

Use when choosing or running MPSKit.jl for infinite / uniform matrix product states — VUMPS, IDMRG / IDMRG2, finite DMRG, TDVP — including unit-cell construction, the tangent-space gradient-norm convergence probe (calc_galerkin), symmetry setup via TensorKit, or MPSKit setup failures. MPSKit has no TEBD; route iTEBD to /using-tenpy.

Use when choosing or running NCTSSoS.jl (with a Clarabel or Mosek SDP backend) for general noncommutative polynomial optimization — the moment-SOS / SOHS hierarchy giving certified bounds, correlative/term sparsity, symmetry reduction (Wedderburn block-diagonalization, Clifford detection), state-polynomial / tracial optimization, GNS reconstruction — or for NCTSSoS setup failures. One of the two step-2 handoff targets from /method-polyopt (the general-purpose engine).

Use when choosing or running NetKet for VMC, neural quantum states, sampler or optimizer choices, JAX CPU/GPU setup, or NetKet setup failures.

Use when choosing or running PEPSKit.jl or TensorKit.jl for PEPS, CTMRG, 2D classical/quantum tensor-network calculations, or PEPSKit setup failures.

Use when choosing or running QMBCertify.jl — the dedicated structured-NPA certifier for ground-state properties of 1D/2D (J1-J2) Heisenberg spin models, producing numeric certified lower/upper bounds on energy, correlations, structure factors, and partition functions via a Mosek SDP, with Gram-matrix export feeding an exact rational post-certification step (1D chains) — or for QMBCertify setup failures. One of the two step-2 handoff targets from /method-polyopt (the structure-exploiting one).

Use when choosing or running StochasticSeriesExpansion.jl or Carlo.jl for sign-free QMC/SSE workflows, MPI setup, or SSE setup failures.

Use when choosing or running TeNPy (Python) for MPS calculations — especially iTEBD (the algorithm MPSKit lacks), plus iDMRG, VUMPS, finite DMRG/TEBD, and finite-T purification — including model setup, the tangent-space gradient-norm probe (tangent_projector_test), symmetry via conserve, threading, or TeNPy/numpy-ABI setup failures.

Use when choosing or running TensorCircuit-NG for quantum-circuit simulation, differentiable circuits, VQE, JAX backend setup, or TensorCircuit-NG setup failures.

Use when a student wants to run their own job on the cluster and have it kept safe — "submit my job", "send this script to the cluster", "check my job", "is my job done", "cancel my job", "download my results", "get my data back". Guards against runaway compute, credential leaks, and destructive file ops via preview-and-confirm. For harness array sweeps the agent drives, use `/using-slurm` instead.

Use when a usable cluster profile is needed and none exists — a fresh account, no `skills/using-slurm/profiles/active.toml`, "set up my cluster", "configure HPC", or when `/cluster-jobs` / `/onboard` / `/using-slurm` find no profile. Builds the unified TOML profile (ssh + scheduler + partitions), probes live resources, and seeds the student safety `[limits]`.

Use when Julia is missing, wrong-version, uninstantiated, or mirror-misconfigured — symptoms like `julia: command not found`, missing `julia-env/Manifest.toml`, package downloads timing out, fresh cluster account, fresh laptop, package-mirror change, or Julia-version bump.