pub struct SVEOptions { /* private fields */ }Expand description
Options of compute_sve_with
SVEOptions::new gives the defaults: the best accuracy of the working
precision, which is chosen from the accuracy (TworkType::Auto), every
singular value above 2 * machine epsilon of it, and the even/odd
expansion of a centrosymmetric kernel. Each method sets one option:
use sparse_ir_sve::{SVEOptions, TworkType};
let options = SVEOptions::new()
.epsilon(1e-10)
.max_num_svals(50)
.twork(TworkType::Float64X2);Implementations§
Source§impl SVEOptions
impl SVEOptions
Sourcepub fn epsilon(self, epsilon: f64) -> Self
pub fn epsilon(self, epsilon: f64) -> Self
Required accuracy, in (0, 1); by default the best accuracy of the working precision (about 1.6e-16 in Float64X2)
Sourcepub fn cutoff(self, cutoff: f64) -> Self
pub fn cutoff(self, cutoff: f64) -> Self
Relative cutoff, in [0, 1]: the singular values below cutoff times
the largest one are discarded
By default 2 * machine epsilon of the working precision, about
4.44e-16 in Float64 and 4.93e-32 in Float64X2 (the libsparseir
default). The SVD of each block already discards the singular values
below 2 * machine epsilon times its largest one, so a smaller
cutoff has little effect. Replaces Self::cutoff_from_epsilon.
Sourcepub fn cutoff_from_epsilon(self) -> Self
pub fn cutoff_from_epsilon(self) -> Self
Keep only the singular values that an expansion accurate to
epsilon uses
Truncates the spectrum at max(2 * machine epsilon, f * epsilon)
times the largest singular value, with f =
ACCURACY_CUTOFF_FACTOR. In Float64X2 (epsilon below about 1e-8)
the block SVDs stop there, which is about 1.5 to 2.3 times faster
than the default, with about 20% fewer singular values. In Float64
the result is that of the default truncated at the cutoff, which for
epsilon ≤ 4.4e-6 is the same result.
The singular values ≥ epsilon times the largest one, plus one more,
agree with those of the default to about 1e-10 relative, not to the
working precision. Replaces Self::cutoff.
Sourcepub fn max_num_svals(self, max_num_svals: usize) -> Self
pub fn max_num_svals(self, max_num_svals: usize) -> Self
Maximum number of singular values to keep, positive
Sourcepub fn twork(self, twork: TworkType) -> Self
pub fn twork(self, twork: TworkType) -> Self
Working precision; by default chosen from the accuracy
(TworkType::Auto)
Sourcepub fn full_domain(self, full_domain: bool) -> Self
pub fn full_domain(self, full_domain: bool) -> Self
Whether to expand a centrosymmetric kernel over its full domain, without exploiting its symmetry; false by default
The singular values agree with the even/odd expansion up to rounding, but the SVD is taken of one matrix with twice as many rows and columns as each even/odd block (about four times the work), the singular functions have no parity, and those of (nearly) degenerate singular values may mix the even and odd sectors. Other kernels are always expanded over their full domain.
Trait Implementations§
Source§impl Clone for SVEOptions
impl Clone for SVEOptions
impl Copy for SVEOptions
Source§impl Debug for SVEOptions
impl Debug for SVEOptions
Source§impl Default for SVEOptions
impl Default for SVEOptions
Source§impl PartialEq for SVEOptions
impl PartialEq for SVEOptions
impl StructuralPartialEq for SVEOptions
Auto Trait Implementations§
impl Freeze for SVEOptions
impl RefUnwindSafe for SVEOptions
impl Send for SVEOptions
impl Sync for SVEOptions
impl Unpin for SVEOptions
impl UnsafeUnpin for SVEOptions
impl UnwindSafe for SVEOptions
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Source§impl<T> CloneToUninit for Twhere
T: Clone,
impl<T> CloneToUninit for Twhere
T: Clone,
impl<T> Scalar for T
§impl<SS, SP> SupersetOf<SS> for SPwhere
SS: SubsetOf<SP>,
impl<SS, SP> SupersetOf<SS> for SPwhere
SS: SubsetOf<SP>,
§fn to_subset(&self) -> Option<SS>
fn to_subset(&self) -> Option<SS>
self from the equivalent element of its
superset. Read more§fn is_in_subset(&self) -> bool
fn is_in_subset(&self) -> bool
self is actually part of its subset T (and can be converted to it).§fn to_subset_unchecked(&self) -> SS
fn to_subset_unchecked(&self) -> SS
self.to_subset but without any property checks. Always succeeds.§fn from_subset(element: &SS) -> SP
fn from_subset(element: &SS) -> SP
self to the equivalent element of its superset.