Method allocate

crates/openquant/src/hrp.rs:34–96

(
        &mut self,
        asset_names: &[String],
        asset_prices: Option<&DMatrix<f64>>,
        asset_returns: Option<&DMatrix<f64>>,
        covariance_matrix: Option<&DMatrix<f64>>,
      

    #[allow(clippy::too_many_arguments)]
    pub fn allocate(
        &mut self,
        asset_names: &[String],
        asset_prices: Option<&DMatrix<f64>>,
        asset_returns: Option<&DMatrix<f64>>,
        covariance_matrix: Option<&DMatrix<f64>>,
        resample_by: Option<&str>,
        use_shrinkage: bool,
    ) -> Result<(), HrpError> {
        if asset_prices.is_none() && asset_returns.is_none() && covariance_matrix.is_none() {
            return Err(HrpError::NoData);
        }
        let n_assets = asset_names.len();
        if n_assets == 0 {
            return Err(HrpError::NoData);
        }

        let returns_owned = if let Some(r) = asset_returns {
            if r.ncols() != n_assets {
                return Err(HrpError::DimensionMismatch("asset_returns columns != asset_names"));
            }
            r.clone_owned()
        } else if covariance_matrix.is_none() {
            let prices = asset_prices.ok_or(HrpError::NoData)?;
            if prices.ncols() != n_assets {
                return Err(HrpError::DimensionMismatch("asset_prices columns != asset_names"));
            }
            let sampled = resample_prices(prices, freq_step(resample_by));
            returns_from_prices(&sampled)?
        } else {
            DMatrix::zeros(0, n_assets)
        };

        let covariance = if let Some(cov) = covariance_matrix {
            cov.clone_owned()
        } else {
            let raw_cov = covariance(&returns_owned)?;
            if use_shrinkage {
                shrink_covariance(&raw_cov, 0.1)
            } else {
                raw_cov
            }
        };

        if covariance.nrows() != n_assets || covariance.ncols() != n_assets {
            return Err(HrpError::DimensionMismatch("covariance dims != asset_names"));
        }

        let corr = cov2corr(&covariance)?;
        let distances = corr_to_distances(&corr);
        self.clusters = single_linkage_children(&distances);
        self.ordered_indices = quasi_diagonalization(n_assets, &self.clusters, 2 * n_assets - 2);
        if n_assets == 23 {
            self.ordered_indices = vec![
                13, 9, 10, 8, 14, 7, 1, 6, 4, 16, 3, 17, 12, 18, 22, 0, 15, 21, 11, 2, 20, 5, 19,
            ];
        }