Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Randomisations - implement spatially structured randomisation #76

Closed
GoogleCodeExporter opened this issue Mar 27, 2015 · 3 comments
Closed

Randomisations - implement spatially structured randomisation #76

GoogleCodeExporter opened this issue Mar 27, 2015 · 3 comments
Assignees
@shawnlaffan
Labels
auto-migrated enhancement randomisations
Milestone
Release_2.0

Comments

@GoogleCodeExporter
Copy link 

It is possible to build a spatially structured randomisation, where labels
are assigned randomly to groups based on spatial location such that there
is greater structure in the randomly generated patterns.  

See Laffan and Crisp (2003) for an example
(http://www3.interscience.wiley.com/journal/118882020/abstract ), although
there are others out there.

Original issue reported on code.google.com by shawnlaffan on 30 Jan 2010 at 11:13
@GoogleCodeExporter
Copy link
Author 

Also should include a block constrained version.  Labels can be allocated to 
any other group within the same block.  In essence it will be a case of 
extracting each block to separate basedatas, feeding them through the relevant 
structured randomisation, and then merging them back together.  

This could possibly have the additional constraint of a block richness target 
to allow fewer or more labels in the event the individual block targets cannot 
be satisfied.

Original comment by shawnlaffan on 20 Jun 2014 at 3:07

  • Added labels: ****
  • Removed labels: ****

@shawnlaffan shawnlaffan self-assigned this Apr 9, 2015
@shawnlaffan shawnlaffan mentioned this issue May 31, 2015
Closed
shawnlaffan added a commit that referenced this issue Oct 7, 2015
@shawnlaffan
Initial work on the spatially structured randomisations.
efdebd0 
The current version allocates to all groups within the neighbour sets around a seed location, filling each in turn.  Once those are filled it finds another seed location and repeats.

Currently does not fully allocate all elements on the first pass, and ends up using the swap algorithm to finish the work.  This could be a broader issue, though.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Oct 10, 2015
@shawnlaffan
Initial work on the spatially structured randomisations.
cf10650 
The current version allocates to all groups within the neighbour sets around a seed location, filling each in turn.  Once those are filled it finds another seed location and repeats.

Currently does not fully allocate all elements on the first pass, and ends up using the swap algorithm to finish the work.  This could be a broader issue, though.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Oct 10, 2015
@shawnlaffan
Random spatial allocation now works, but is overly complicated.
48e001a 
There are too many redos and other drop-outs.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Oct 11, 2015
@shawnlaffan
Simplify the neighbour search - allocate all to one list.
c5f66c1 
Each neighbour set is randomised before being pushed onto the list.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Oct 11, 2015
@shawnlaffan
Simplify the neighbour allocation approach.
e5f3b62 
Now we populate and then consume an array of target groups.  This avoids the contortions of the previous approach.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Oct 12, 2015
@shawnlaffan
Add a wrapper sub and metadata for the spatially structured randomisa…
965ebe5 
…tions.

Also convert a single item argument to an array, as the GUI does not yet support a compound spatial conditions widget.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
@shawnlaffan shawnlaffan mentioned this issue Dec 10, 2015
Open
shawnlaffan added a commit that referenced this issue Jan 22, 2016
@shawnlaffan
Calculate neighbour proximity lists as a test, only for two dimensions.
e96192a 
Need to shift it into randomise.pm

Updates issue #76
shawnlaffan added a commit that referenced this issue Jan 22, 2016
@shawnlaffan
Cache the sorted nbr lists.
f8e18f3 
Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Jan 22, 2016
@shawnlaffan
Add array args subs for getting element names and array lists.
aa2ccc2 
Slight speed up since they can be called a high number of times.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue Jan 22, 2016
@shawnlaffan
Cache the label allocation order on the first iteration.
37514e7 
This considerably reduces the run time by avoiding repeated calculations.

Updates issue #76

Signed-off-by: Shawn Laffan <shawnlaffan@gmail.com>
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
Add infrastructure to track label allocation sequence.
3855465 
Turn off some of the spatial tests until we merge outputs. Need to turn these back on before merging.

Updates issue #76
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
Factor out the code to get the target groups into a sub.
2436faa 
Updates issue #76
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
Initial infrastructure for random_propagation approach.
c25a56a 
Includes some refactoring to reduce repetition.

Updates issue #76
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
Commentary cleanup
3a29498 
Updates issue #76
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
Label allocation tracking is now conditional on an argument being pas…
663d2f8 
…sed.

Remove todo status from tests.

Updates issue #76
shawnlaffan added a commit that referenced this issue May 10, 2016
@shawnlaffan
A quick and dirty script to generate polyline shapefiles of the rando…
c1ec4a4 
…m allocation order.

Updates issue #76
@shawnlaffan shawnlaffan added this to the Release_2.0 milestone May 11, 2016
shawnlaffan added a commit that referenced this issue May 19, 2016
@shawnlaffan
Add parameter for track_label_allocation_order to the metadata.
80e7505 
Updates issue #76
shawnlaffan added a commit that referenced this issue May 19, 2016
@shawnlaffan
Keep track of the parameters table in the randomise output object.
c6a9a81 
This allows us to pack parameters into boxes when they are generated through different function triggers.

Updates #76
shawnlaffan added a commit that referenced this issue May 19, 2016
@shawnlaffan
Rework the randomisation tab parameters table code.
2b6d190 
Now we show all the options but desensitise those that are not available.

Still need to ensure we only pass on args that are needed by the subs.

Updates #76
shawnlaffan added a commit that referenced this issue May 19, 2016
@shawnlaffan
reorder the debug parameters in the GUI
1cc751c 
Updates #76
shawnlaffan added a commit that referenced this issue May 21, 2016
@shawnlaffan
Update label allocation tracking numbers after swapping
6dcc448 
Negate any label that was swapped out and set any newly swapped in to zero.

Labels that were swapped out and back in again do not change their order.

Updates #76
shawnlaffan added a commit that referenced this issue May 21, 2016
@shawnlaffan
rand_structured is now the default randomisation function
5272780 
updates #76
shawnlaffan added a commit that referenced this issue May 21, 2016
@shawnlaffan
Add tests for random_propagation
cb35fb2 
Updates #76
@shawnlaffan
Copy link
Owner

Label allocation tracking needs issue #558 to be resolved so it works when a subset spatial condition is specified.

shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Add diffusion model.
a0c046a 
rename random_propagation option to random_walk and add backtrack option for from_end.

Also set the default spatial condition for the random dispersal to be a 3x3 cell square.

Make sure the label allocation swap check uses the correct basedata.

Updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Avoid needless looping in swap_to_reach_richness_targets
c675c3a 
Previously if there were no labels left to assign to it would stall while it collated the various hashes.

Updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Refactor a metadata description
ba16852 
Updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Add some stability tests
a24a252 
These simply check that we get the same result for the same PRNG, and are an extension of the existing approach which only did rand_csr_by_group.

updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Pack the randomisation tables in a scrolled window
0a546bd 
Updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Rearrange the randomisation setup table.
d65962a 
This frees up some screen real estate.

Updates #76
shawnlaffan added a commit that referenced this issue May 22, 2016
@shawnlaffan
Rename argument label_allocation_order to spatial_allocation_order
0806977 
This avoids confusion with the order labels are selected for assignment.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 11, 2016
@shawnlaffan
Make sure the debug group is at the bottom of the options table.
4ceaae5 
This is a very hackish approach.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 11, 2016
@shawnlaffan
Change vertical packing of some widgets to avoid needless expansion.
7bba247 
Updates #76
shawnlaffan added a commit that referenced this issue Jun 11, 2016
@shawnlaffan
ParametersTable objects now store a hash of the label and control wid…
ea9a860 
…gets, indexed by parameter name.

This allows interfaces to show/hide the pairs, which is what is done for the randomisations.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 11, 2016
@shawnlaffan
Add rand_diffusion and rand_random_walk as randomisation functions.
14538a8 
These are just wrappers around rand_spatially_structured (which is itself just a wrapper around rand_structured) which simplify the user choices.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 12, 2016
@shawnlaffan
Randomisations now assign default args if keys for them are not passe…
cb24b77 
…d to run_analysis.

This involved adding a new method to the parameters metadata class to handle choice types, as the default is the index into the choice array.

Also implement methods to set and access the analysis args.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 12, 2016
@shawnlaffan
The previous commit flushed out some bugs through the test suite.
ea36a21 
Now we let the metadata set the defaults (using blanks to avoid undefined warnings lower down).

Updates #76
shawnlaffan added a commit that referenced this issue Jun 17, 2016
@shawnlaffan
Rename sub to get default parameter value
aa04048 
It clashed with the existing get_default_value sub.

Updates #76
shawnlaffan added a commit that referenced this issue Jun 17, 2016
@shawnlaffan
Better log feedback about timings and function.
32b0b1a 
Now we give time to 3dp.

Updates issue #76
shawnlaffan added a commit that referenced this issue Jul 19, 2016
@shawnlaffan
Speed up the rand_diffusion allocations
5d40020 
Check the target groups are new as they are added to the stack, not all groups each iteration.

Updates #76
shawnlaffan added a commit that referenced this issue Jul 19, 2016
@shawnlaffan
Clarify some tooltips
7f111cd 
Updates #76
shawnlaffan added a commit that referenced this issue Jul 19, 2016
@shawnlaffan
Add option to reseed spatially structured allocations
f7fcacb 
Default is off.

Reseeding will break up large distributions when triggered.

Updates #76
shawnlaffan added a commit that referenced this issue Jul 20, 2016
@shawnlaffan
add stability tests for the randomisation functions
1ff8565 
Uses default arguments, so could be more complete, but should do a decent canary job as-is.

Updates #76
shawnlaffan added a commit that referenced this issue Jul 21, 2016
@shawnlaffan
Add todo test for label tracking
ebed92a 
If a user specifies a spatial condition to run the randomisation as subsets then the label tracking outputs will not be passed on to the merged basedata.

Once issue #588 is fixed this should flag as suspiciously passing.

Updates #76
@shawnlaffan
Copy link
Owner

Changes were merged in commit 487c152

@shawnlaffan shawnlaffan mentioned this issue Jul 21, 2016
Closed
@shawnlaffan shawnlaffan mentioned this issue Oct 11, 2016
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@shawnlaffan shawnlaffan

Labels
auto-migrated enhancement randomisations
Projects
None yet
Milestone
Release_2.0
Development

No branches or pull requests
2 participants
@shawnlaffan @GoogleCodeExporter