Trait DlrFromIr 

pub trait DlrFromIr<S>: Sized
where
    S: StatisticsType,
{
    // Required methods
    fn from_ir_with_poles<B>(basis: &B, poles: Vec<f64>) -> Result<Self, Error>
       where B: IrBasis<S>;
    fn from_ir<B>(basis: &B) -> Result<Self, Error>
       where B: IrBasis<S>;
}

Constructors of a DiscreteLehmannRepresentation from an IR basis

The returned DLR carries the IrDlrTransform of its source basis, so DiscreteLehmannRepresentation::from_ir_nd and DiscreteLehmannRepresentation::to_ir_nd work on it.

Required Methods

fn from_ir_with_poles<B>(basis: &B, poles: Vec<f64>) -> Result<Self, Error>

where B: IrBasis<S>,

Create a DLR from an IR basis with custom poles

The tau-domain pole basis is built from the logistic representation, while kernel-specific regularizers are preserved for compatible kernels.

Arguments

• basis - The IR basis to construct DLR from
• poles - Pole positions on the real-frequency axis, in [-ωmax, ωmax] of basis

Errors

• Error::KernelStatisticsMismatch if the kernel does not support the requested statistics (e.g. RegularizedBoseKernel with fermionic statistics)
• Error::EmptyInput if poles is empty
• Error::NonFiniteInput if a pole is NaN or infinite, and Error::OutOfDomain if a pole is outside [-ωmax, ωmax] of basis, both named poles, for the first such pole
• Error::NotSupported for a bosonic pole at 0 if the kernel has a ypower other than 0 or 1 (the DLR knows the limit at 0 for those only)

Duplicate poles are accepted. They make DiscreteLehmannRepresentation::from_ir_nd ill-conditioned: the coefficients of equal poles are not unique, although the round trip through DiscreteLehmannRepresentation::to_ir_nd still recovers the IR coefficients.

fn from_ir<B>(basis: &B) -> Result<Self, Error>

where B: IrBasis<S>,

Create a DLR from an IR basis with its default pole locations

Uses the default omega sampling points from the basis.

Arguments

• basis - The IR basis to construct DLR from

Errors

• Error::InsufficientDefaultPoles if the basis has fewer default poles than functions. This can happen with certain kernel types (e.g. RegularizedBoseKernel) due to numerical precision limitations in root finding.
• Error::KernelStatisticsMismatch as in Self::from_ir_with_poles
• Error::NotSupported as in Self::from_ir_with_poles: for a default pole at 0 of a bosonic basis whose kernel has a ypower other than 0 or 1
• The errors of Basis::default_omega_sampling_points (NotSupported for an SVE with too few singular functions)

Dyn Compatibility

This trait is not dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors

impl<S> DlrFromIr<S> for DiscreteLehmannRepresentation<S>

where S: StatisticsType + 'static,