-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #3 from alphal00p/tensors
Tensors
- Loading branch information
Showing
36 changed files
with
9,732 additions
and
1,551 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,158 @@ | ||
| use std::{fmt::Debug, ops::Neg}; | ||
|
|
||
| use _gammaloop::tensor::{ | ||
| ufo::{euclidean_four_vector, gamma}, | ||
| AbstractIndex, ContractionCountStructure, FallibleMul, MixedTensor, Representation, | ||
| SetTensorData, Slot, SparseTensor, TensorNetwork, TensorStructure, | ||
| }; | ||
|
|
||
| use ahash::{AHashMap, HashMap}; | ||
| use criterion::{criterion_group, criterion_main, Criterion}; | ||
|
|
||
| use rand::{distributions::Uniform, Rng, SeedableRng}; | ||
| use rand_xoshiro::Xoroshiro64Star; | ||
| use symbolica::{ | ||
| domains::float::Complex, | ||
| representations::{Atom, AtomView}, | ||
| state::State, | ||
| }; | ||
|
|
||
| fn indices(n: i32, m: i32) -> Vec<i32> { | ||
| let spacings: [i32; 2] = [n, m]; | ||
| let mut start = 1; | ||
| let mut ranges = Vec::new(); | ||
|
|
||
| for &spacing in spacings.iter() { | ||
| ranges.push((start..start + spacing).chain(std::iter::once(-1))); | ||
| start += spacing; | ||
| } | ||
|
|
||
| ranges.into_iter().flatten().collect() | ||
| } | ||
|
|
||
| fn gamma_net_param( | ||
| minkindices: &[i32], | ||
| vbar: [Complex<f64>; 4], | ||
| u: [Complex<f64>; 4], | ||
| ) -> TensorNetwork<MixedTensor<ContractionCountStructure>> { | ||
| let mut i: i32 = 0; | ||
| let mut contracting_index = 0.into(); | ||
| let mut result: Vec<MixedTensor<ContractionCountStructure>> = | ||
| vec![euclidean_four_vector(contracting_index, &vbar).into()]; | ||
| for m in minkindices { | ||
| let ui = contracting_index; | ||
| contracting_index += 1.into(); | ||
| let uj = contracting_index; | ||
| if *m > 0 { | ||
| let p: ContractionCountStructure = vec![Slot::from(( | ||
| usize::try_from(*m).unwrap().into(), | ||
| Representation::Lorentz(4.into()), | ||
| ))] | ||
| .into_iter() | ||
| .collect(); | ||
| i += 1; | ||
| let pid = State::get_or_insert_fn(&format!("p{}", i), None).unwrap(); | ||
|
|
||
| result.push(p.shadow_with(pid).into()); | ||
|
|
||
| result.push(gamma(usize::try_from(*m).unwrap().into(), (ui, uj)).into()); | ||
| } else { | ||
| result.push( | ||
| gamma( | ||
| AbstractIndex::from(usize::try_from(m.neg()).unwrap() + 10000), | ||
| (ui, uj), | ||
| ) | ||
| .into(), | ||
| ); | ||
| } | ||
| } | ||
| result.push(euclidean_four_vector(contracting_index, &u).into()); | ||
| TensorNetwork::from(result) | ||
| } | ||
|
|
||
| fn test_tensor<S>(structure: S) -> SparseTensor<symbolica::domains::float::Complex<f64>, S> | ||
| where | ||
| S: TensorStructure, | ||
| { | ||
| let mut rng: Xoroshiro64Star = Xoroshiro64Star::from_entropy(); | ||
|
|
||
| let mut tensor = SparseTensor::empty(structure); | ||
|
|
||
| let density = tensor.size(); | ||
|
|
||
| let multipliable = Uniform::new(1., 10.); | ||
|
|
||
| for _ in 0..density { | ||
| tensor | ||
| .set_flat( | ||
| rng.gen_range(0..tensor.size()), | ||
| Complex::<f64>::new(rng.sample(multipliable), rng.sample(multipliable)), | ||
| ) | ||
| .unwrap(); | ||
| } | ||
|
|
||
| tensor | ||
| } | ||
|
|
||
| fn const_map_gen<'a, 'b, I>( | ||
| params: &'a [MixedTensor<I>], | ||
| const_map: &mut HashMap<AtomView<'b>, symbolica::domains::float::Complex<f64>>, | ||
| ) where | ||
| 'a: 'b, | ||
| I: TensorStructure + Clone + Debug, | ||
| { | ||
| for (_i, p) in params.iter().enumerate() { | ||
| let pdata = test_tensor(p.structure().clone()).to_dense(); | ||
| p.try_as_symbolic() | ||
| .unwrap() | ||
| .try_as_dense() | ||
| .unwrap() | ||
| .append_const_map(&pdata, const_map); | ||
| } | ||
| } | ||
| fn criterion_benchmark(c: &mut Criterion) { | ||
| let one = Complex::<f64>::new(1.0, 0.0); | ||
|
|
||
| let vbar = [ | ||
| one.mul_fallible(3.0).unwrap(), | ||
| one.mul_fallible(3.1).unwrap(), | ||
| one.mul_fallible(3.2).unwrap(), | ||
| one.mul_fallible(3.3).unwrap(), | ||
| ]; | ||
| let u = [ | ||
| one.mul_fallible(4.0).unwrap(), | ||
| one.mul_fallible(4.1).unwrap(), | ||
| one.mul_fallible(4.2).unwrap(), | ||
| one.mul_fallible(4.3).unwrap(), | ||
| ]; | ||
| let minkindices = indices(20, 24); | ||
|
|
||
| let mut net = gamma_net_param(&minkindices, vbar, u); | ||
| net.generate_params(); | ||
| let params = net.params.clone(); | ||
| println!("{:?}", params.len()); | ||
| net.contract_algo(|tn| tn.edge_to_min_degree_node_with_depth(2)); | ||
| let mut const_map = AHashMap::new(); | ||
|
|
||
| let i = Atom::new_var(State::I); | ||
| const_map.insert(i.as_view(), Complex::<f64>::new(0., 1.)); | ||
|
|
||
| let mut group = c.benchmark_group("evaluate_net"); | ||
|
|
||
| group.bench_function("Evaluate_net", |b| { | ||
| b.iter_batched( | ||
| || net.clone(), | ||
| |mut net| { | ||
| const_map_gen(¶ms, &mut const_map); | ||
|
|
||
| net.evaluate_complex(&const_map); | ||
|
|
||
| net.contract(); | ||
| }, | ||
| criterion::BatchSize::SmallInput, | ||
| ) | ||
| }); | ||
| } | ||
|
|
||
| criterion_group!(benches, criterion_benchmark); | ||
| criterion_main!(benches); |
Oops, something went wrong.