Closed molpopgen closed 6 years ago
looks good to me. but, how about I will (tomorow) write a complementary abstract, about computing general statistics on tree sequences (and using them for inference); and we tie them together? I will leave this open to do that.
Sure.
On Mon, Jan 8, 2018 at 10:18 PM Peter Ralph notifications@github.com wrote:
looks good to me. but, how about I will (tomorow) write a complementary abstract, about computing general statistics on tree sequences (and using them for inference); and we tie them together? I will leave this open to do that.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356190907, or mute the thread https://github.com/notifications/unsubscribe-auth/AGHnH1GxnGfxmLVcUt7vjAvpTq2W9XWiks5tIwQzgaJpZM4RWxwx .
--
Kevin Thornton
Associate Professor
Ecology and Evolutionary Biology
UC Irvine
I made some edits, trying to make it sound more interesting:
Nice edits 👏
On Tue, Jan 9, 2018 at 21:30 Peter Ralph notifications@github.com wrote:
I made some edits, trying to make it sound more interesting:
- some concrete numbers (except I left these as XX's), so it's clear this isn't an abstract method applicable to hundreds of individuals, but a smoking-fast one
- emphasize that not only it's faster but also stores way more information
- mention the cool thing about initialization.
— You are receiving this because your review was requested.
Reply to this email directly, view it on GitHub https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356503972, or mute the thread https://github.com/notifications/unsubscribe-auth/AAfLOHF6dB5cOsNScY9C2Y9DPDFqYDknks5tJErkgaJpZM4RWxwx .
-- -Jaime
some concrete numbers (except I left these as XX's), so it's clear this isn't an abstract method applicable to hundreds of individuals, but a smoking-fast one
I avoided this, because X hours on my server is Y on yours, which largely depends on budget. IMO, it is the 50x improvement that is important.
I wanted to put something in that says how big feasible simulations are with this change. "Feasible" is relative, I know. But it seems useful to somehow make it clear the scale of things we're talking about, especially since it seems unbelievable we can deal with the whole-population history for tens of thousands of individuals.
The 50x number would be good to drop in there, also.
On Tue, Jan 9, 2018, 10:56 PM Peter Ralph notifications@github.com wrote:
I wanted to put something in that says how big feasible simulations are with this change. "Feasible" is relative, I know. But it seems useful to somehow make it clear the scale of things we're talking about, especially since it seems unbelievable we can deal with the whole-population history for tens of thousands of individuals.
Really? Msprime already does it, implying that you can do it for any forward sim if you simplify every generation.
Regarding feasibility, we have not addressed that with the work we've done. We capped things at 72 hours. Asking what is feasible with the specific implementation I've done would require a different study design, and ultimately boils down to how patient one is, and what ones needs are. For example, much bigger regions will be possible if you are willing to rely on parameters scaling. But, if you think, even wrongly, that you "need" to simulate N=1e9, then the picture is very different.
The 50x number would be good to drop in there, also.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356516212, or mute the thread https://github.com/notifications/unsubscribe-auth/AGHnH2dhXa6JQ24JIVFHuHJkbGmUrEPuks5tJF7bgaJpZM4RWxwx .
--
Kevin Thornton
Associate Professor
Ecology and Evolutionary Biology
UC Irvine
I mean order-of-magnitude. Two purposes: (a) give people a rough idea of what they can do (b) cite impressively big numbers.
re: "unbelievable" - I mean to people who haven't been following things like msprime.
Also: I have decided to talk about my project with Josh, since that project is actually done: https://www.biorxiv.org/content/early/2017/12/11/231209 as opposed to my usual MO of stressing myself out by saying I'll talk about something I haven't properly done yet. So: no additional abstract to tie in. Huh: unless I submit two abstracts; one for a poster...
Confused--do you want to do a poster on this paper?
On Wed, Jan 10, 2018 at 10:31 AM Peter Ralph notifications@github.com wrote:
Also: I have decided to talk about my project with Josh, since that project is actually done: https://www.biorxiv.org/content/early/2017/12/11/231209 as opposed to my usual MO of stressing myself out by saying I'll talk about something I haven't properly done yet. So: no additional abstract to tie in. Huh: unless I submit two abstracts; one for a poster...
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356691764, or mute the thread https://github.com/notifications/unsubscribe-auth/AGHnH62nboWGmThos-BzPESPrGML-V_3ks5tJQCAgaJpZM4RWxwx .
--
Kevin Thornton
Associate Professor
Ecology and Evolutionary Biology
UC Irvine
I think Peter was joking. But in any case the poster would be complementary and on "computing general statistics on tree sequences (and using them for inference)".
your last sentence was cut off?
On Wed, Jan 10, 2018 at 11:32 AM Jaime Ashander notifications@github.com wrote:
I think Peter was joking. But in any case the poster would be complementary and on
computing general statistics on tree sequences (and using them for inference)
On Wed, Jan 10, 2018 at 11:18 Kevin R. Thornton notifications@github.com wrote:
Confused--do you want to do a poster on this paper?
On Wed, Jan 10, 2018 at 10:31 AM Peter Ralph notifications@github.com wrote:
Also: I have decided to talk about my project with Josh, since that project is actually done: https://www.biorxiv.org/content/early/2017/12/11/231209 as opposed to my usual MO of stressing myself out by saying I'll talk about something I haven't properly done yet. So: no additional abstract to tie in. Huh: unless I submit two abstracts; one for a poster...
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356691764 , or mute the thread <
.
--
Kevin Thornton
Associate Professor
Ecology and Evolutionary Biology
UC Irvine
—
You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356707217 , or mute the thread < https://github.com/notifications/unsubscribe-auth/AAfLOORFApq7DPzyh2NP4q4MOrtrnLn8ks5tJQzkgaJpZM4RWxwx
.
-- -Jaime
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/petrelharp/ftprime_ms/pull/50#issuecomment-356710837, or mute the thread https://github.com/notifications/unsubscribe-auth/AGHnH956nkH_1nCDPr2CySXb6kQM0_3nks5tJQ_VgaJpZM4RWxwx .
--
Kevin Thornton
Associate Professor
Ecology and Evolutionary Biology
UC Irvine
Fixed on github. Damn GH-email bridge can't handle manual >
to quote things. I said
I think Peter was joking. But in any case the poster would be complementary and on "computing general statistics on tree sequences (and using them for inference)".
I meant that I wanted to try to talk about that paper; but I liked the idea of synergizing tree sequence stuff, but don't want to give a talk; so maybe a poster about using tree sequences to compute statistics quickly.
Added new .tex file w/draft of PEQ abstract.