work on the megadrought runs!

[lizzieinvancouver]

Excellent news! I have started this. As a reminder, we varied s and phi (in a correlated fashion) and we also did several versions of tau_i:

• megadrought runs with low tauI; corr(s,phi)
• megadrought runs with mod tauI; corr(s,phi)
• megadrought runs with large tauI; corr(s,phi)

@donahuem I need help figuring out the structure of the runs (what happens in each 400 years and how there can be a wet and dry ID for one run in one period... I assume it is always wet then dry perhaps? So period 1 goes with the wet ID and period 2 goes with the dryID? I think this is what happens in getEnvt.R) and how to see the species differences in s and phi ...

I have the info on tau_i for each species, but how to get either the corr(s,phi) or s or phi?

Also, I could use a reminder of how the 400 plus 400 years go ... are they a set of 400 wet years, followed by 400 dry? Or something else?

[donahuem]

R0 is always 400 yrs wet then 400 yrs dry The SummaryOut should now have a line for each 400 years - it should output after wet and then output after dry For megaD, we write out: wetID dryID [each ID is a uniform random between 1 and 9999 indicating which of Ben's R0 timeseries] rho - correlation between s and phi R0_mean - mean(R0) for the period (either wet period or dry period) R0_median - median (R0) for the period R0_autocor - autocor(R0) for the period Updated so (s,phi) pairs drawn for each species (this was wrong before)

output should problem should be fixed

[donahuem]

Next attempt at MegaD runs are these: 855723 855748 855753

[lizzieinvancouver]

Update! Not so many single species exist (10% of the 40K runs) and not so many coexist (40 out of the 40K runs) so we need to work on some next steps. If we get through these next steps and don't get very far, we give up on the megadroughts project.

1. Do some runs with more restricted s (seems like the surviving species are 0.65-0.95 s).
2. Think on trade-off of phi and s, make sure we believe it is an actual trade-off.
3. Do runs without competition (just one species) and check that these have much more survival, if they don't we need to constrain region of persistence (that is, if we are not getting much survival even with no competition we probably have the wrong parameter set for this environment).

[donahuem]

Given low persistence (even through wet period), new set of megaD runs with restricted range for s = 0.65-0.95 1008362 1008404 1008425

[donahuem]

All prior megaD runs had the same varI (regardless of the values of tauI in the input file)-- tauI was randomly selected for all species -- this is not what was intended. The code now runs correctly, which means that tauI is fixed and takes a different level (lo, med,hi) between each set of runs

New batch, which 3 levels of tauI (and still with restricted range of s) 1009682 1009731 1009761

[donahuem]

Disregard 1009682 1009731 1009761 problem with range of s

New batch with correct s values and 3 levels of tauI 1009960 1009991 1000022

[lizzieinvancouver]

@donahuem Oooh, moderate tauI in the middle for the win ...

[1] "the current folder ID is 1009960 the total rows are: 4921 and the total coexisting rows are: 0"
[1] "the current folder ID is 1009991 the total rows are: 9317 and the total coexisting rows are: 345"
[1] "the current folder ID is 1010022 the total rows are: 4931 and the total coexisting rows are: 0"

[donahuem]

Updated megaD runs. These have a tradeoff between s and tauI, where s varies from .8-.98 (s=0.8 in our "normal"runs) and tauI ranges from 0.45 to 0.55 (varies from .1-.9 in our normal runs). s and tauI are correlated by rho (=-0.7 in these runs). The parm xDrought is mulitplied by the megaD year values from Ben (xDrought= 0.5 in these runs). The run numbers are: 9014727 9014765 9014815 [Lizzie has downloaded these.]

[donahuem]

I have added two sets of megaD runs. The first set has corr(s,alpha), long (800y) stationary period. alpha =U(0,1) The second set has corr(s,tauI), long (800 y) stationary period" tauI=U(0.4-0.6) Both sets have s=0.5-0.98 Tracking can now be a 0/1 flag or give a range, so it can have two columns in runparms (min to max)

Submitted batch job 9270489 Submitted batch job 9270555

[Lizzie has downloaded these.]