Adaptive Dynamics

  • Working on simulation based approach to summarize distribution of waiting times across different phases.
  • What to do about zeros in phase two (when first dimorphism survives long enough to be invaded…)
  • Quantify number/distribution of failed attempts at each phase? How to represent these distributions across a parameter space in sigma_c, sigma_k, sigma_mu, and mu?
image
image
image
image
library(BranchingTime)
sim <- branching_time(rep=1000, cpu=8)
save(file = "run1.Rdat")
pdf("distribution.pdf")
m <- max(sim$data)
plot(density(sim$data[1,]), xlim = c(0,m), col="black" )
lines(density(sim$data[2,] ), col = "blue")
lines(density(sim$data[3,] ), col = "green")
lines(density(sim$data[4,] ), col = "red")
legend('topright', c("1", "2", "3", "4"), col=c("black", "blue", "green", "red"), pch = 15)
dev.off()

data output for plot, parameters, etc in dropbox.

Michael Eisen Visit

11pm-noon

  • Excellent talk & discussion with Michael Eisen and the Davis Open Science group today. Video of the discussion to be posted shortly. We’ll be discussing the talk and potential for activism among the open science group at the social on Wednesday.
  • Talk got me thinking about lots of things. Strong examples of the how broken the current system is: not only in the frequency of errors but the difficulty in correcting them – a paper in high impact journal that has been contradicted repeatedly in other literature remains the most cited. Also the credibility assigned to the accuracy of peer review out of proportion with the rigor of the process – most papers are published sooner or later. Better if another standard assured credibility than publication in a peer reviewed journal alone. Goal of review process is to reject papers, makes novel papers harder to publish than more traditional ones. Much of the advice you hear about making successful publications has little to do with the science. (And how about everything is miscellaneous approach to science?)
  • simply pointing out how flawed the system is wasn’t solving anything, even though most tended to agree. So started PLoS. Next year PLoS ONE will be the biggest journal and is financially self-supporting, Other evidence of successful open access models: biomed central sold for (some large amount) to Elsevier.
  • Commenting system on papers does not seem to be working. Interesting discussion of anonymous/signed versions, and filtering/validating comments, and provocative view of being able to know what everyone in your field thought of each thing they read – custom view/lens of a paper, choose your own reviewers style reading.
  • Best place for Open Science Group to take a role: Challenging Intellectual Property policy at UC. Berkeley example: IP department costs more to maintain than IP generates in revenue, and resources are not available to the public – at a time when UC is most in need of public support. Meanwhile private universities such as Harvard are doing much better in open policies. More stats on this would be great.
  • Situation isn’t as bad as you think – data does not support the myth that you need Science/Nature papers to get a job. Getting high profile papers isn’t necessarily the reason for getting the job, correlation doesn’t imply causation. More data on this would be great as well.


Reading ——-