Resuming from previous run

[mattpitkin]

Hi @eggplantbren @dfm, I was wondering if it is possible (in particular using the Python interface) to resume sampling from where a previous run left off (in case, for example, a run gets aborted for some reason unrelated to a problem with the code)? From looking at the code it appears that this might not be possible as the backend gets reset and samples are initialised each time the code is run, but I may be missing something.

[eggplantbren]

I don't think so. This will be possible in DNest5 (yeah, I'm writing that as we speak :))

On Thu, Aug 6, 2020 at 10:15 PM Matt Pitkin notifications@github.com wrote:

Hi @eggplantbren https://github.com/eggplantbren @dfm https://github.com/dfm, I was wondering if it is possible (in particular using the Python interface) to resume sampling from where a previous run left off (in case, for example, a run gets aborted for some reason unrelated to a problem with the code)? From looking at the code it appears that this might not be possible as the backend gets reset and samples are initialised each time the code is run, but I may be missing something.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/eggplantbren/DNest4/issues/35, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAMBKOVZPYM7MNQ2BMAZXB3R7J7EVANCNFSM4PWNWOHQ .

-- Dr Brendon J. Brewer Phone: +64 27 5001336 Web: https://www.brendonbrewer.com LBRY: lbry://@BrendonBrewer:3 https://lbry.tv/@BrendonBrewer:3 Keybase: https://keybase.io/brendonbrewer

[mattpitkin]

Thanks @eggplantbren. I await DNest5 with baited breath!

[eggplantbren]

Here you go: https://github.com/eggplantbren/DNest5 There's no RJObject or Python model specification yet. But any feedback would be appreciated.

On Thu, Aug 6, 2020 at 10:34 PM Matt Pitkin notifications@github.com wrote:

Thanks @eggplantbren https://github.com/eggplantbren. I await DNest5 with baited breath!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/eggplantbren/DNest4/issues/35#issuecomment-669850251, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAMBKOTBXCC4UUXQVP4FWYTR7KBLPANCNFSM4PWNWOHQ .

-- Dr Brendon J. Brewer Phone: +64 27 5001336 Web: https://www.brendonbrewer.com LBRY: lbry://@BrendonBrewer:3 https://lbry.tv/@BrendonBrewer:3 Keybase: https://keybase.io/brendonbrewer

[mattpitkin]

Here you go: https://github.com/eggplantbren/DNest5 There's no RJObject or Python model specification yet. But any feedback would be appreciated.

Excellent! Could you say briefly what the rational behind DNest5 is? Does it enhance the functionality, change the interface, or just generally speed things up?

[eggplantbren]

I was experimenting with different ideas to try to make a better algorithm, failed on that goal, but succeeded in making much better software. A bit more detail here: https://brendonbrewer.com/blog.html#dnest5-happening

On Wed, Aug 12, 2020 at 9:49 PM Matt Pitkin notifications@github.com wrote:

Here you go: https://github.com/eggplantbren/DNest5 There's no RJObject or Python model specification yet. But any feedback would be appreciated.

Excellent! Could you say briefly what the rational behind DNest5 is? Does it enhance the functionality, change the interface, or just generally speed things up?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/eggplantbren/DNest4/issues/35#issuecomment-672772789, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAMBKOUWDHFFAOBZPHIQYELSAJQSZANCNFSM4PWNWOHQ .

-- Dr Brendon J. Brewer Phone: +64 27 5001336 Web: https://www.brendonbrewer.com LBRY: lbry://@BrendonBrewer:3 https://lbry.tv/@BrendonBrewer:3 Keybase: https://keybase.io/brendonbrewer

[mattpitkin]

Thanks!