MAOS
Multithreaded Adaptive Optics Simulator
Loading...
Searching...
No Matches
recon_mvst.c File Reference

Functions

void setup_recon_mvst (recon_t *recon, const parms_t *parms)

Detailed Description

Setup Minimum variance split tomography. This function is deprecated.

Function Documentation

◆ setup_recon_mvst()

void setup_recon_mvst ( recon_t * recon,
const parms_t * parms )

compute the MVST split tomography NGS mode reconstructor.

2010-03-16: New Implementation.

Definition:

  • \(G_{lgs}\) is LGS gradient operator from xloc, contained in recon->GXtomo as dspcell
  • \(G_{ngs}\) is NGS gradient operator from xloc, contained in recon->GXL as dense matrix
  • \(C_{lgs}, C_{ngs}\) is the LGS, and NGS measurement noise covariance matrix.
  • \(\hat{x}_{lgs}\) is LGS tomography output
  • \(\hat{x}_{ngs}\) is NGS minimum variance split tomography output
  • \(a_{ngs}\) is NGS minimum variance DM output.
  • \(F\) is the fitting operator
  • \(H_A\) is the ray tracing operator from aloc to ploc, contained in recon->HA.
  • \(MVModes\) is the MVST NGS modes
  • \(MVRngs\) is the MVST NGS reconstructor

We have

\begin{eqnarray*}\hat{x}_{lgs}&=&A^{-1}G^{T}_{lgs}C_{lgs}^{-1}s_{lgs}\\ Uw&=&A^{-1}G_{ngs}^TC_{ngs}^{-1}\\ \hat{x}_{ngs}&=&Uw(1+G_{ngs}Uw)^{-1}(s_{ngs}-G_{ngs}\hat{x}_{lgs});\\ a_{NGS}&=&F\hat{x}_{ngs}\\ MVModes&=&F\cdot Uw\\ MVRngs&=&(1+G_{ngs}Uw)^{-1} \end{eqnarray*}

If we want to orthnormalize the NGS modes. Propagate the NGS modes \(F\cdot Uw\) to fitting directions and compute the cross-coupling matrix with weighting on \(W_0, W_1\), we have

\begin{eqnarray*}Q&=&H_A\cdot F \cdot Uw\\ M_{CC}&=&Q^T(W_0-W_1 W_1^T)Q\\ \end{eqnarray*}

Do SVD on \(MCC\) we have \(M_{CC}=u\Sigma v^T \) where \(u{\equiv}v\) because \(MCC\) is symmetric. Redefine the NGS modes and reconstructor as

\begin{eqnarray*}MVModes&=&F\cdot Uw\cdot u \Sigma^{-1/2}\\ MVRngs&=&\Sigma^{1/2}u^T (1+G_{ngs}A^{-1}G_{ngs}^T C_{ngs}^{-1})^{-1} \end{eqnarray*}