How to Do a Performance Check on New Rec

Testing the interaction matrix

The first test on interaction matrix is based on a visual check of the matrix with the command: im.displayIM() where "im" is an LgswInteractionMatrix object.

The noise propagation coefficient can calculated as:
np.diag(np.linalg.pinv(np.dot(im.getInteractionMatrix().T,im.getInteractionMatrix())))*(1e9)**2

Interaction Matrix 20180221_234506 on SX side produce these coefficients for modes from 2 to 10 in (nm rms/slopes)^2: [ 63111, 29969 , 61451, 27634, 28181, 23911, 23293, 18151]

The noise propagation coefficients times the noise estimation on the WFS should give the Modal Residual.

Testing the inversion of the interaction matrix to a reconstructor

The inversion of the Lgsw Interaction Matrix produce the Lgsw Reconstructor.

The singular values of the SVD has to be distributed in a factor 10, more or less. To get the singular values after the Pseudo Inversion use the function: im.getSingularValues()

The product of the Reconstructor and a vector of constant slopes has to return mainly Tip and Tilt values and some small numbers on the higher modes. See DayTime20180222 for more details.

Testing the reconstructor in closed loop

Look HowToPerformArgosAoLoopInDayTime to get ready for the test.
Don't forget to load the new combined reconstructor manually!
Before closing the AO loop set a disturbance and enable it.
List of Disturbances are in AoLoopDisturbances.

Close the loop, ramp the gain, wait for truth sensing convergence (in case you are using truth sensing) and then acquire some snapshots.

Once you have some snapshots it is possible to analyze with the argos_SIDE_snapshot_analyzer, for example:

tnnew=terminal.analyzer('20180222_010121')
tnnew.modalPlot()
tns=terminal.set('20180222_002107','20180222_010832')
tns.tWikiLoopLog()
TN AO Gain AO CL TT Res. WF [nm rms] HO Res. WF [nm rms] AO recSorted ascending DIMM ["] JCL PCL Notes
20180222_002107 1.2,0.5,0.4 T 88.3177 43.2277 20180221_235700 -1 T F
20180222_002135 1.2,0.5,0.4 T 94.55 43.2824 20180221_235700 -1 T F
20180222_002351 1.4,0.5,0.4 T 73.8316 56.0694 20180221_235700 -1 T F
20180222_002408 1.4,0.5,0.4 T 72.9462 56.1992 20180221_235700 -1 T F
20180222_004941 1.4,0.5,0.4 T 47.7592 64.1223 20180221_194600 -1 T F
20180222_010832 1.4,0.5,0.4 T 54.8768 49.5231 20180221_194600 -1 T F
TT Res. WF [nm rms] is the WaveFront Error residual measured by the Tip/Tilt WFS (the Pyramid WFS in this case).

HO Res. WF [nm rms] is the WFE residual measured by the LGSW.

It is useful to compare new snapshots with old acquisition from previous runs, or perform new measures with old reconstructors.
Topic revision: r4 - 20 Jan 2019, LorenzoBusoni
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback