Optimal control module - Revision history

Kuprov: sync with Spinach main 3975f139: control.fid_type, control.traj_pen, ens_block.m, mas_drifts.m, renamed basis field

2026-09-18T10:00:34Z

sync with Spinach main 3975f139: control.fid_type, control.traj_pen, ens_block.m, mas_drifts.m, renamed basis field

2026-08-30T08:14:02Z

sync with Spinach main f053e432: add non_orth.m to the distortion and penalty function list

2026-08-30T08:04:35Z

new page for Spinach main f053e432: add ens_catalog.m to the auxiliaries list

2026-08-30T07:51:04Z

sync with Spinach main f053e432: index bloch_siegert.m and bss_ops.m

2026-06-06T18:39:27Z

Add missing Spinach function to function index

2026-04-25T12:15:25Z

Retarget obsolete linesearch links

2026-04-25T11:38:53Z

Replace obsolete function links

2026-04-25T11:27:43Z

Normalise sequential blank lines

2026-04-25T11:24:15Z

‎Optimal control function list: optimisation

2026-04-25T11:24:05Z

‎Stage 8: run the optimisation

@@ Line 25: / Line 25: @@
       [[bas]].formalism='sphten-liouv';
       [[bas]].approximation='IK-0';
-      [[bas]].level=4;
+      [[bas]].inter_level=4;
       % Spinach housekeeping
@@ Line 82: / Line 82: @@
 |O
 |A character string specifying the type of fidelity functional. Specify 'real' for the real part of the overlap between the final state and the target state, 'imag' for the imaginary part, and 'square' for the absolute square of the overlap. The default is 'real'.
 |-
 |control.operators
@@ Line 331: / Line 339: @@
 [[drifts.m]] - drift Hamiltonian extraction from ensemble wrappers
 [[ens_catalog.m]] - ensemble case catalog for the control sequence fidelity evaluation
@@ Line 337: / Line 347: @@
 [[hess_reorder.m]] - Hessian reordering for multi-channel optimisations
 [[objeval.m]] - objective function evaluation wrapper

← Older revision		Revision as of 08:14, 30 August 2026
Line 383:		Line 383:

	[[no_dist.m]] - unit distortion that does nothing		[[no_dist.m]] - unit distortion that does nothing
		+
		+	[[non_orth.m]] - non-orthogonal control channel distortion

	[[penalty.m]] - penalty functionals and their derivatives		[[penalty.m]] - penalty functionals and their derivatives

← Older revision		Revision as of 08:04, 30 August 2026
Line 331:		Line 331:

	[[drifts.m]] - drift Hamiltonian extraction from ensemble wrappers		[[drifts.m]] - drift Hamiltonian extraction from ensemble wrappers
		+
		+	[[ens_catalog.m]] - ensemble case catalog for the control sequence fidelity evaluation

	[[ensemble.m]] - ensemble preprocessing and parallelisation		[[ensemble.m]] - ensemble preprocessing and parallelisation

← Older revision		Revision as of 07:51, 30 August 2026
Line 323:		Line 323:

	[[aux_mat.m]] - auxiliary matrix building for directional derivatives		[[aux_mat.m]] - auxiliary matrix building for directional derivatives
		+
		+	[[bloch_siegert.m]] - Bloch-Siegert augmentation of Cartesian control pulses
		+
		+	[[bss_ops.m]] - Bloch-Siegert response operators for the control channels

	[[dirdiff.m]] - directional derivatives of matrix exponentials		[[dirdiff.m]] - directional derivatives of matrix exponentials

@@ Line 337: / Line 337: @@
 [[trapdiff.m]] - multi-point propagator derivatives
 ==Optimal control function list: optimisation==
@@ Line 381: / Line 387: @@
 [[transfermat.m]] - transfer matrix calculator
 ''Version 2.11, authors: [[Ilya Kuprov]], [[David Goodwin]], [[Uluk Rasulov]]''

@@ Line 348: / Line 348: @@
 [[lbfgs.m]] - LBFGS pseudo-Hessian update
-[[linesearch.m]] - cubic line search
+[[bracketing.m]], [[sectioning.m]] - cubic line search
 ==Optimal control function list: input and output==

@@ Line 289: / Line 289: @@
 In principle, the derivatives these functions return can be supplied to any optimisation function, including those supplied with ''Matlab''. ''Spinach'' does, however, include optimisation tools that are designed specifically for optimal control settings:
-;[[fminnewton.m]]
+;[[Fmaxnewton.m]]
 :Gradient descent, quasi-Newton and Newton-Raphson methods.
@@ Line 295: / Line 295: @@
       % Run the optimization
-      [[fminnewton.m|fminnewton]](spin_system,@[[grape_xy.m|grape_xy]],guess);
+      [[Fmaxnewton.m|Fmaxnewton.m]](spin_system,@[[grape_xy.m|grape_xy]],guess);
 To run phase optimisation, do not forget to specify the amplitude profile above (same dimensions as your phase guess) and call [[grape_phase.m]] instead:
       % Run the optimization
-      [[fminnewton.m|fminnewton]](spin_system,@[[grape_phase.m|grape_phase]],guess);
+      [[Fmaxnewton.m|Fmaxnewton.m]](spin_system,@[[grape_phase.m|grape_phase]],guess);
 To use the Newton-Raphson method, specify 'newton' instead of 'lbfgs' in the control.method setting above. This is generally a good idea with fewer than 100 points in the waveform.
@@ Line 340: / Line 340: @@
 ==Optimal control function list: optimisation==
-[[fminnewton.m]] - Newton-Raphson and LBFGS optimisers
+[[Fmaxnewton.m]] - Newton-Raphson and LBFGS optimisers
 [[bfgs_upd.m]] - one-step BFGS Hessian update for maximisation