Changeset b9c432ac in sasview

Timestamp:

Mar 30, 2019 12:31:38 AM (6 years ago)

Author:

wojciech

Branches:

master

Children:

4ecf91a

Parents:

104613f (diff), 332c10d (diff)
Note: this is a merge changeset, the changes displayed below correspond to the merge itself.
Use the (diff) links above to see all the changes relative to each parent.

Message:

Merge branch 'master' of ​https://github.com/SasView/sasview into ticket-1267

Files:

• INSTALL.txt (modified) (1 diff)
• docs/sphinx-docs/source/user/RELEASE.rst (modified) (3 diffs)
• src/sas/sasgui/perspectives/fitting/media/fitting.rst (modified) (2 diffs)
• src/sas/sasgui/perspectives/fitting/media/fitting_help.rst (modified) (1 diff)
• src/sas/sasgui/perspectives/fitting/media/fitting_sq.rst (added)
• src/sas/sasview/media/getting_started_with_sasview.pdf (modified) (previous)
• src/sas/sasgui/perspectives/fitting/gpu_options.py (modified) (2 diffs)

INSTALL.txt

@@ -47 +47 @@
 More information can be found here:
 
-http://www.sasview.org/help.html
-http://www.sasview.org/faq.html
+http://www.sasview.org/help/
+http://www.sasview.org/faq/
 http://www.sasview.org/trac/wiki/DevNotes

docs/sphinx-docs/source/user/RELEASE.rst

@@ -156 +156 @@
 * Fixes # 784: Add 3D integral to Correlation Function analysis
 * Fixes # 786: core_shell_parallelepiped 1-D model is incorrect
-* Fixes # 818: “report button” followed by “save” makes an empty pdf file???
+* Fixes # 818: "report button" followed by "save" makes an empty pdf file???
 * Fixes # 830: Check compliance of loader against NXcanSAS-1.0 release
 * Fixes # 838: Fix model download from marketplace
@@ -770 +770 @@
           that any previous versions of SasView are uninstalled prior to 
           installing the new version UNLESS you are installing SasView to 
-          versioned folders.
+          versioned folders. **We recommend you avoid using folder paths 
+          that contain spaces or characters that cannot be found on a 
+          standard keyboard (eg, characters with accents, umlauts, etc).**
 
 .. note:: The easiest approach to setting up the proper environment to
@@ -965 +967 @@
 Frequently Asked Questions
 ==========================
-http://www.sasview.org/faq.html
+http://www.sasview.org/faq/
 
 

src/sas/sasgui/perspectives/fitting/media/fitting.rst

@@ -17 +17 @@
    Smearing Functions <resolution>
 
+   Fitting Models with Structure Factors <fitting_sq>
+
+   Writing a Plugin Model <plugin>
+
    Polarisation/Magnetic Scattering <magnetism/magnetism>
-   
+
    Oriented Particles <orientation/orientation>
 
@@ -27 +31 @@
    Fitting SESANS Data <sesans/sesans_fitting>
 
-   Writing a Plugin Model <plugin>
-
    Computations with a GPU <gpu_setup>
 

src/sas/sasgui/perspectives/fitting/media/fitting_help.rst

@@ -338 +338 @@
 These optimisers form the *Bumps* package written by P Kienzle. For more information
 on each optimiser, see the :ref:`Fitting_Documentation`.
+
+.. ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ
+
+Fitting Integer Parameters
+--------------------------
+
+Most of the parameters in SasView models will naturally take floating point (decimal) 
+values, but there are some parameters which can only have integer values. Examples 
+include, but are not limited to, the number of shells in a multilayer vesicle, the 
+number of beads in a pearl necklace, the number of arms of a star polymer, and so on.
+Wherever possible/recognised, the integer nature of a parameter is specified in the 
+respective model documentation and/or parameter table, so read the documentation 
+carefully!
+
+Integer parameters must be fitted with care.
+
+Start with your best possible guess for the value of the parameter. And using 
+*a priori* knowledge, fix as many of the other parameters as possible.
+ 
+The SasView optimizers treat integer parameters internally as floating point 
+numbers, but the values presented to the user are truncated or rounded, as 
+appropriate.
+
+In most instances integer parameters will probably be greater than zero. A good 
+policy in such cases is to use a constraint to enforce this.
+
+Because an integer parameter should, by definition, only move in integer steps, 
+problems may be encountered if the optimizer step size is too small. Similarly, 
+be **very careful** about applying polydispersity to integer parameters.
+
+The Levenberg-Marquardt and Quasi-Newton BFGS (and other derivative-based) 
+optimizers are probably best avoided for fitting models with integer parameters.
 
 .. ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ

src/sas/sasgui/perspectives/fitting/gpu_options.py

@@ -257 +257 @@
         #The same block of code as for OK but it is needed if we want to have
         #active response to Test button
+        import numpy as np
         no_opencl_msg = False
         if self.sas_opencl:
@@ -269 +270 @@
         try:
             env = sasmodels.kernelcl.environment()
-            clinfo = [(ctx.devices[0].platform.vendor,
-                       ctx.devices[0].platform.version,
-                       ctx.devices[0].vendor,
-                       ctx.devices[0].name,
-                       ctx.devices[0].version)
-                      for ctx in env.context]
+            dtype64 = np.dtype('float64')
+            dtype32 = np.dtype('float32')
+            if dtype64 in env.context:
+                ctx = env.context[dtype64].devices[0]
+            else:
+                ctx = env.context[dtype32].devices[0]
+            clinfo = (ctx.platform.vendor,
+                        ctx.platform.version,
+                        ctx.vendor,
+                        ctx.name,
+                        ctx.version)
         except Exception:
             clinfo = None