The Greek word ταχύς (tachys) is an adjective meaning quick or swift.
To reach the system sizes that real materials demand, tachys is optimized to scale across large GPU clusters.
Spin-½ lattice models, benchmarked up to ~0 spins.
Fermionic lattice models, benchmarked up to 0 electrons, a scale already meaningful for unconventional superconductors.
Itinerant fermions and localized spins within a single Hamiltonian, as required by Kondo lattice models of heavy-fermion compounds.
One network trained across an entire family of Hamiltonians, to explore the parameter space of a material in a single optimization.
A ground-state calculation consists of three function calls in an explicit Python loop: configurations are sampled, the energy is estimated on that sample, and the parameters are updated. Each call returns new values rather than modifying its arguments.
A move rule defines a Markov chain over configurations. sample advances every chain and returns the resulting configurations together with their log-amplitudes.
state, log_amps, acceptance = sample(nsweeps, state, action, keys, wf)
Chains are sharded across the available JAX devices, so the call is the same on one CPU and on many GPUs.
A Hamiltonian is a sum of spin or fermionic operator terms. compute_expectation applies it to the sampled configurations and returns the local energies together with ⟨E⟩ and ⟨E²⟩.
E_L, e_mean, e2_mean = compute_expectation(H, wf, state, log_amps)
No matrix representation is constructed, so memory scales with the sample rather than with the dimension of the Hilbert space.
An optimizer maps the local energies to a parameter update. Applying the update returns a new wavefunction rather than modifying the existing one, so a run is a sequence of values that can be checkpointed or compared at any step.
updates, opt_state = optimizer(E_L, opt_state, state, wf) wf = wf.apply_gradients(updates, lr)
Learning-rate schedules, stopping criteria and additional observables are expressed directly in the loop rather than registered through hooks.
Each of these is a pure function over JAX pytrees, compiled with jit. Any Flax module can be used as the ansatz.
tachys is installed with pip install tachys, which also installs JAX.