Talk:Particle swarm optimization: Difference between revisions
m
Thundergnat moved page Talk:Particle Swarm Optimization to Talk:Particle swarm optimization: Follow normal task title capitalization policy
No edit summary |
Thundergnat (talk | contribs) m (Thundergnat moved page Talk:Particle Swarm Optimization to Talk:Particle swarm optimization: Follow normal task title capitalization policy) |
||
| (5 intermediate revisions by 2 users not shown) | |||
Line 15:
: So... ok, maybe any function from wikipedia's [[wp:Test_functions_for_optimization|Test functions for optimization]] should be considered valid, here. But how do we then judge an implementation for correctness?
: Judging a particle swarm implementation for correctness seems like a serious problem. Basically you start with a set of random numbers and then have them perform a not-quite-random walk within the space defined by some arbitrary function. And, on top of that, "
: In other words, currently the task suggests that each implementation should be different. And, since we're working with random numbers, we should expect different results from two runs of the same program. But that won't give us comparable implementations.
Line 22:
: With those nailed down, I think that implementations and results should be comparable. --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 19:02, 1 August 2015 (UTC)
The Wikipedia article describes the basic algorithm fairly specifically and also mentions the existence of variants (which the task description fails to mention). Note that the current examples now both conform to the basic algorithm, and also exercise it with the same objective function and search bounds, and have comparable results. Should the task description suggest, or require, an algorithm version, objective function, and parameters? It appears that all other issues have been addressed. --[[User:JimTheriot|JimTheriot]] ([[User talk:JimTheriot|talk]]) 23:30, 3 August 2015 (UTC)
: I think it should specify those thing. Ideally, two different cases. --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 14:19, 4 August 2015 (UTC)
This is a draft task "for reasons that should be found in its talk page" - for clarity, can someone summarize the remaining obstacles? --[[User:JimTheriot|JimTheriot]] ([[User talk:JimTheriot|talk]]) 18:57, 13 August 2015 (UTC)
: For the generic draft issues, there's the [[Rosetta_Code:Add_a_Task#Draft_vs_non-draft|draft vs non-draft]] part of the "Add a Task" page. Also, I think it has been typical practice to leave a task in draft status until it has been implemented in four different languages. (Though there have been exceptions to that approach also - especially when people have been unaware of the draft issues.) --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 19:33, 13 August 2015 (UTC)
| |||