Module qmc::sse

QMC algorithms and traits. A module with various QMC algorithms and traits.

Simple functionality can be accessed through the qmc_ising and qmc_runner modules, the later being a generalization of the former. Other modules are useful for building custom backends for sse (see fast_ops for an example).

Basic Quantum Ising Example

use qmc::sse::*;
use rand::prelude::*;

// H = J_ij s_i s_j
let edges = vec![
  ((0, 1), -1.0), // ((i, j), J)
  ((1, 2), 1.0),
  ((2, 3), 1.0),
  ((3, 0), 1.0)
];
let transverse = 1.0;
let longitudinal = 0.0;
let beta = 1.0;

// Make an ising model using default system prng.
let rng = rand::thread_rng();
let initial_state = Some(vec![false, false, false, false]);
let initial_cutoff = 4;
let mut g = DefaultQMCIsingGraph::<ThreadRng>::new_with_rng(edges, transverse, longitudinal, initial_cutoff, rng, initial_state);

// Take timesteps
g.timesteps(1000, beta);

// Take timesteps and sample states (as Vec<Vec<bool>>).
let (state, average_energy) = g.timesteps_sample(1000, beta, None);

Arbitrary QMC

use qmc::sse::*;
use rand::prelude::*;

let rng = rand::thread_rng();
let nvars = 3;
let mut g = DefaultQMC::new(nvars, rng, false);
// sz + sx + 1
let mat = vec![2.0, 1.0,
               1.0, 0.0];
for i in 0 .. nvars {
  g.make_interaction(mat.clone(), vec![i]).unwrap()
}
// Some energy offsets for the macrostates.
let diag = (0 .. 2_u32.pow(nvars as u32))
                .map(|i| i as f64)
                .collect::<Vec<f64>>();
g.make_diagonal_interaction(diag, vec![0, 1, 2]).unwrap();
g.timesteps(1000, 1.0);

Re-exports

pub use qmc_debug::*;

pub use qmc_ising::*;

pub use qmc_runner::*;

pub use qmc_traits::*;

Modules

fast_ops	Clever operator management
qmc_debug	Debugging stuff.
qmc_ising	A QMC graph for easy TFIM.
qmc_runner	A generic QMC framework.
qmc_traits	Traits which, when implemented, run SSE.
qmc_types	QMC utility classes.