Recent changes to ShapeGeneration

ShapeGeneration modified by A. Smith

A. Smith — Thu, 08 Sep 2016 14:41:40 -0000

--- v3
+++ v4
@@ -1,6 +1,6 @@
 ### Shape Generation from Acquisition and Processing Information
 INFOS improves fit quality by generating lineshapes based on acquisition and processing information. Shapes are generated by processing either Gaussian and/or Lorentzian signal decay, according to the acquisition time, spectrum width, processed resolution, and the apodization function. The shapes are then generated for a range of linewidths, which are applied in the spectrum fitting. This calculation is done during setup, and then is stored for later use in the program, contributing significantly to the fitting speed.

-Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction (details are found [here](/p/infos/wiki/dn_struct). One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, [‘shapes0’](/p/infos/wiki/shapes0), which contains structures for each dimension that contain the lineshapes.
+Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction (details are found [here](/p/infos/wiki/dn_struct)). One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, [‘shapes0’](/p/infos/wiki/shapes0), which contains structures for each dimension that contain the lineshapes.

 The major limitation to this method of lineshape generation is that only one optimization variable may describe the lineshape. This is helped by the ability to add some Lorentzian character to Gaussian lineshapes and vice versa. A second limitation is that lineshapes may not be correlated between dimensions. For example, if a lineshape is the result of a poor shim leaving a foot, then that foot should extend diagonally between two dimensions. However, even if an arbitrary lineshape is specified by the user to include the foot, it will not appear diagonally across the dimensions since this would require correlation of lineshapes between the dimensions.

ShapeGeneration modified by A. Smith

A. Smith — Thu, 08 Sep 2016 14:41:20 -0000

--- v2
+++ v3
@@ -1,6 +1,6 @@
 ### Shape Generation from Acquisition and Processing Information
 INFOS improves fit quality by generating lineshapes based on acquisition and processing information. Shapes are generated by processing either Gaussian and/or Lorentzian signal decay, according to the acquisition time, spectrum width, processed resolution, and the apodization function. The shapes are then generated for a range of linewidths, which are applied in the spectrum fitting. This calculation is done during setup, and then is stored for later use in the program, contributing significantly to the fitting speed.

-Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction. Specification of acquisition and processing parameters, and signal decay is described in section 2.3. One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, ‘shapes0’, which contains structures for each dimension that contain the lineshapes. This cell is described in section 2.6.
+Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction (details are found [here](/p/infos/wiki/dn_struct). One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, [‘shapes0’](/p/infos/wiki/shapes0), which contains structures for each dimension that contain the lineshapes.

 The major limitation to this method of lineshape generation is that only one optimization variable may describe the lineshape. This is helped by the ability to add some Lorentzian character to Gaussian lineshapes and vice versa. A second limitation is that lineshapes may not be correlated between dimensions. For example, if a lineshape is the result of a poor shim leaving a foot, then that foot should extend diagonally between two dimensions. However, even if an arbitrary lineshape is specified by the user to include the foot, it will not appear diagonally across the dimensions since this would require correlation of lineshapes between the dimensions.

ShapeGeneration modified by A. Smith

A. Smith — Thu, 08 Sep 2016 09:50:06 -0000

--- v1
+++ v2
@@ -1,4 +1,6 @@
 ### Shape Generation from Acquisition and Processing Information
 INFOS improves fit quality by generating lineshapes based on acquisition and processing information. Shapes are generated by processing either Gaussian and/or Lorentzian signal decay, according to the acquisition time, spectrum width, processed resolution, and the apodization function. The shapes are then generated for a range of linewidths, which are applied in the spectrum fitting. This calculation is done during setup, and then is stored for later use in the program, contributing significantly to the fitting speed. 
-   Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction. Specification of acquisition and processing parameters, and signal decay is described in section 2.3. One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, ‘shapes0’, which contains structures for each dimension that contain the lineshapes. This cell is described in section 2.6.
-   The major limitation to this method of lineshape generation is that only one optimization variable may describe the lineshape. This is helped by the ability to add some Lorentzian character to Gaussian lineshapes and vice versa. A second limitation is that lineshapes may not be correlated between dimensions. For example, if a lineshape is the result of a poor shim leaving a foot, then that foot should extend diagonally between two dimensions. However, even if an arbitrary lineshape is specified by the user to include the foot, it will not appear diagonally across the dimensions since this would require correlation of lineshapes between the dimensions.
+
+Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction. Specification of acquisition and processing parameters, and signal decay is described in section 2.3. One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, ‘shapes0’, which contains structures for each dimension that contain the lineshapes. This cell is described in section 2.6.
+
+The major limitation to this method of lineshape generation is that only one optimization variable may describe the lineshape. This is helped by the ability to add some Lorentzian character to Gaussian lineshapes and vice versa. A second limitation is that lineshapes may not be correlated between dimensions. For example, if a lineshape is the result of a poor shim leaving a foot, then that foot should extend diagonally between two dimensions. However, even if an arbitrary lineshape is specified by the user to include the foot, it will not appear diagonally across the dimensions since this would require correlation of lineshapes between the dimensions.

ShapeGeneration modified by A. Smith

A. Smith — Thu, 08 Sep 2016 09:49:51 -0000

Shape Generation from Acquisition and Processing Information

INFOS improves fit quality by generating lineshapes based on acquisition and processing information. Shapes are generated by processing either Gaussian and/or Lorentzian signal decay, according to the acquisition time, spectrum width, processed resolution, and the apodization function. The shapes are then generated for a range of linewidths, which are applied in the spectrum fitting. This calculation is done during setup, and then is stored for later use in the program, contributing significantly to the fitting speed.
Because lineshape calculations are done during setup and stored, the FitSpec function is limited to only one variable describing the linewidth. Therefore, optimizing mixes of Gaussian and Lorentzian lineshapes separately for each peak is not possible. However, one may specify a fixed amount of Gaussian or Lorentzian line-broadening to be applied to all peaks, and optimize the other type of broadening. Additionaly, one may specify a particular fraction of Gaussian and Lorentzian broadening, so that both widths vary proportionally according to the given fraction. Specification of acquisition and processing parameters, and signal decay is described in section 2.3. One may go further and specify arbitrary lineshapes. This is considerably more complicated to setup, but is executed by giving a cell, ‘shapes0’, which contains structures for each dimension that contain the lineshapes. This cell is described in section 2.6.
The major limitation to this method of lineshape generation is that only one optimization variable may describe the lineshape. This is helped by the ability to add some Lorentzian character to Gaussian lineshapes and vice versa. A second limitation is that lineshapes may not be correlated between dimensions. For example, if a lineshape is the result of a poor shim leaving a foot, then that foot should extend diagonally between two dimensions. However, even if an arbitrary lineshape is specified by the user to include the foot, it will not appear diagonally across the dimensions since this would require correlation of lineshapes between the dimensions.