b130 dev1: Extremely subtle differences in next generations even though mutation is set to zero

When I set mutation to zero I would expect no mutations to occur. Yet there are very subtle differences between some new strategies and last generation strategies  implying either mutation is still occurring OR only a tiny part of one parent makes it in to the child. If breeding is occurring properly without mutation we should see more characteristics in child made of each parent than one single variable.

In the screenshots are one example of a previous generation strat vs present generation strat. The highlight is the only difference in the pseudocode. Mutation is still set to zero so no mutation should be happening. This is generation number 2.




Attachments
Screenshot 66.png
(97.68 KiB)
Screenshot 63.png
(118.48 KiB)
Screenshot 67.png
(71.98 KiB)
  • Votes +6
  • Project StrategyQuant X
  • Type Bug
  • Status Refused
  • Priority Normal

History

b
#1

bentra

16.11.2020 20:37

Task created

b
#2

bentra

16.11.2020 20:49
Voted for this task.
k
#3

Karish

16.11.2020 21:45
Voted for this task.
o
#4

Enric

17.11.2020 08:51
Voted for this task.
h
#5

hankeys

17.11.2020 11:29
Voted for this task.
JK
#6

Insanity82007

17.11.2020 12:45
Voted for this task.
MF
#7

Mark Fric

18.11.2020 14:38

Status changed from New to Refused

you have crossover turned on, it is not strange that there are changes in values. Crossover doesn't replace only full rules, it is random and it can replace any part of the tree - whole rule, a condition or only a parameter.


So I expect the difference in paraeters (2.7 vs 1.3) is a reult of crossover breeding - exchange between two different strategies.


Votes: +6

Drop files to upload

or

choose files

Max size: 5MB

Not allowed: exe, msi, application, reg, php, js, htaccess, htpasswd, gitignore

...
Wait please