VBA_MFX is slowing down

Hi, I'm using VBA_MFX to estimate choice model parameters. Now I tapped into the following issue: I ran multiple simulations and VBA_MFX fits in a row (~ 360 times to test different models and a range of parameters). I noticed that the VBA_MFX inversion is getting slower and slower (starting from a few seconds up to >60min per run). If I restart MATLAB, the problem is gone. 'clear all' or 'close all' do not solve the problem.

Best, Antonius

Hi Antonius,

Which model are you trying to invert with MFX? It would be nice if you could post a minmal code on https://gist.github.com/ so we can try to reproduce your problem.

Best

Hi, sorry for my late reply. I just came back to this... I'm fitting discounting functions to choice data, nothing fancy. I could not upload files so I attached them to the email. You might have to adjust the path in the beginning of VBA_fit_multi_subject.m

I assume the problem might be related to the diagnostic windows opening... I have not found an option to turn them off. I run Matlab itself with -nodisplay in the terminal, but the problem continues.

Let me know if I can help with anything.

Best, Antonius

On Thu, Sep 28, 2017 at 3:42 PM, Lionel RIGOUX notifications@github.com wrote:

Hi Antonius,

Which model are you trying to invert with MFX? It would be nice if you could post a minmal code on https://gist.github.com/ so we can try to reproduce your problem.

Best

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-332840093, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7sPLOHqf5teqRYYvfn9PyCPt3GDfks5sm6JMgaJpZM4PnOT1 .

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX notifications@github.com wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler notifications@github.com wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX notifications@github.com wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s-aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler notifications@github.com wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX notifications@github.com wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler notifications@github.com wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX notifications@github.com wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s-aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349328190, or mute the thread https://github.com/notifications/unsubscribe-auth/AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

great, thanks for checking! Let me know what the result is. I will try the inversion. how did you manage to supress the windows?

Best, Antonius

On Tue, Dec 5, 2017 at 6:56 PM, Lionel RIGOUX notifications@github.com wrote:

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler notifications@github.com wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX notifications@github.com wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler <notifications@github.com

wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX < notifications@github.com> wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s- aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349328190>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349386756, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7iXnBdrjcm4aoFu1Do1xdhj295Xnks5s9YPKgaJpZM4PnOT1 .

So far I’m just doin a close all /clc after each iteration to avoid crowding the memory. You should see with Jean (and whoever wrote the MFX function) about implementing a proper verbose/display switch.

BTW, could you ask/help Jean to switch his copy of the toolbox to the development branch, he’s started pushing frenetically directly to the master and that’s not good...

On 5. Dec 2017, at 19:00, Antonius Wiehler notifications@github.com wrote:

great, thanks for checking! Let me know what the result is. I will try the inversion. how did you manage to supress the windows?

Best, Antonius

On Tue, Dec 5, 2017 at 6:56 PM, Lionel RIGOUX notifications@github.com wrote:

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler notifications@github.com wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX notifications@github.com wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler <notifications@github.com

wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX < notifications@github.com> wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s- aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349328190>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349386756, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7iXnBdrjcm4aoFu1Do1xdhj295Xnks5s9YPKgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349387875, or mute the thread https://github.com/notifications/unsubscribe-auth/AE5OHZZwkdT9brc2wvs5rVOhDcBRKaHzks5s9YShgaJpZM4PnOT1.

Ouch, I could have done that. facepalm Great immediate fix!

I will speak to Jean tomorrow, he left already.

Best, Antonius

On Tue, Dec 5, 2017 at 7:04 PM, Lionel RIGOUX notifications@github.com wrote:

So far I’m just doin a close all /clc after each iteration to avoid crowding the memory. You should see with Jean (and whoever wrote the MFX function) about implementing a proper verbose/display switch.

BTW, could you ask/help Jean to switch his copy of the toolbox to the development branch, he’s started pushing frenetically directly to the master and that’s not good...

On 5. Dec 2017, at 19:00, Antonius Wiehler notifications@github.com wrote:

great, thanks for checking! Let me know what the result is. I will try the inversion. how did you manage to supress the windows?

Best, Antonius

On Tue, Dec 5, 2017 at 6:56 PM, Lionel RIGOUX notifications@github.com wrote:

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler <notifications@github.com

wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX < notifications@github.com> wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler < notifications@github.com

wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX < notifications@github.com> wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s- aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349328190>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349386756, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7iXnBdrjcm4aoFu1Do1xdhj295Xnks5s9YPKgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349387875>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZZwkdT9brc2wvs5rVOhDcBRKaHzks5s9YShgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349389339, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7u_ngEyBDS7A-bxT9mEebThdYchSks5s9YXLgaJpZM4PnOT1 .

Hey Antonius,

Your script ran overnight and is till running. So far I’ve seen no consistent slowing of the inversion across the ~250 blocs than were inverted, one bloc taking between 150 and 350s. I guess the accumulation of windows was indeed the issue (it can be quite memory greedy). I suggest that:

you insert a clc and a close all between each block of MFX inversion
ask Jean for a more elegant solution using a noDisplay and a verbose flag in the MFX (and please on the development branch)
see if you still have the problem. I am sure you could also make use of the parallel toolbox to speed the things a bit up.

Let me know if this solves your problem Cheers L

On 5. Dec 2017, at 19:09, Antonius Wiehler notifications@github.com wrote:

Ouch, I could have done that. facepalm Great immediate fix!

I will speak to Jean tomorrow, he left already.

Best, Antonius

On Tue, Dec 5, 2017 at 7:04 PM, Lionel RIGOUX notifications@github.com wrote:

So far I’m just doin a close all /clc after each iteration to avoid crowding the memory. You should see with Jean (and whoever wrote the MFX function) about implementing a proper verbose/display switch.

BTW, could you ask/help Jean to switch his copy of the toolbox to the development branch, he’s started pushing frenetically directly to the master and that’s not good...

On 5. Dec 2017, at 19:00, Antonius Wiehler notifications@github.com wrote:

great, thanks for checking! Let me know what the result is. I will try the inversion. how did you manage to supress the windows?

Best, Antonius

On Tue, Dec 5, 2017 at 6:56 PM, Lionel RIGOUX notifications@github.com wrote:

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler <notifications@github.com

wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX < notifications@github.com> wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler < notifications@github.com

wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX < notifications@github.com> wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349311531>, or mute the thread https://github.com/notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s- aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349328190>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349386756, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7iXnBdrjcm4aoFu1Do1xdhj295Xnks5s9YPKgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349387875>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZZwkdT9brc2wvs5rVOhDcBRKaHzks5s9YShgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349389339, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7u_ngEyBDS7A-bxT9mEebThdYchSks5s9YXLgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349390644, or mute the thread https://github.com/notifications/unsubscribe-auth/AE5OHVIbXOHr5GB6AWQcFFajo-f-ebiHks5s9YbZgaJpZM4PnOT1.

Hi, so I double checked. You had the right idea, closing the figures in every loop iteration did the trick. The problem disappeared. I still not fully understand it, I started Matlab with no graphical display and there was no more than 1/3 of my RAM used, but anyway...

I could help with setting up a no display option for VBA_MFX if needed.

Thanks again for checking!

Best, Antonius

On Wed, Dec 6, 2017 at 10:18 AM, Lionel RIGOUX notifications@github.com wrote:

Hey Antonius,

Your script ran overnight and is till running. So far I’ve seen no consistent slowing of the inversion across the ~250 blocs than were inverted, one bloc taking between 150 and 350s. I guess the accumulation of windows was indeed the issue (it can be quite memory greedy). I suggest that:

you insert a clc and a close all between each block of MFX inversion

ask Jean for a more elegant solution using a noDisplay and a verbose flag in the MFX (and please on the development branch)

see if you still have the problem. I am sure you could also make use of the parallel toolbox to speed the things a bit up.

Let me know if this solves your problem Cheers L

On 5. Dec 2017, at 19:09, Antonius Wiehler notifications@github.com wrote:

Ouch, I could have done that. facepalm Great immediate fix!

I will speak to Jean tomorrow, he left already.

Best, Antonius

On Tue, Dec 5, 2017 at 7:04 PM, Lionel RIGOUX notifications@github.com wrote:

So far I’m just doin a close all /clc after each iteration to avoid crowding the memory. You should see with Jean (and whoever wrote the MFX function) about implementing a proper verbose/display switch.

BTW, could you ask/help Jean to switch his copy of the toolbox to the development branch, he’s started pushing frenetically directly to the master and that’s not good...

On 5. Dec 2017, at 19:00, Antonius Wiehler <notifications@github.com

wrote:

great, thanks for checking! Let me know what the result is. I will try the inversion. how did you manage to supress the windows?

Best, Antonius

On Tue, Dec 5, 2017 at 6:56 PM, Lionel RIGOUX < notifications@github.com> wrote:

I got the script running with no windows. For me, even the first files take way more than “few seconds” to run, as I am closer to 3min so far. I will let run overnight while logging the duration this if I can reproduce you bug. Are you sure the lengthening is due to the repetition and not the specific parametrization you’re testing in later loops? Meaning, if you reverse the order of model inversion, do you still see an increase of duration or does it reverse?

On 5. Dec 2017, at 15:55, Antonius Wiehler < notifications@github.com

wrote:

Hi, sorry not for beeing clear enough. Unzip everything. "Die-hard-discounting" is the main folder, "main_exp" includes (simulated) log files, "_output" priors for modeling. The main script to run is VBA_fit_multi_subject.m. Folder locations in line 21-24 need to be adjusted, depending on the location of the folders.

hope this helps.

Best, Antonius

On Tue, Dec 5, 2017 at 3:01 PM, Lionel RIGOUX < notifications@github.com> wrote:

How should I organize all the folders and what should I change to run it?

On 5. Dec 2017, at 14:58, Antonius Wiehler < notifications@github.com

wrote:

done. (sent to your gmail).

On Tue, Dec 5, 2017 at 2:55 PM, Lionel RIGOUX < notifications@github.com> wrote:

Hey Antonius

This is a Github thread and answering to this email is logged into github. Because of this (I guess) I could not see the files you attached. Please send me privately the code, or start a Gist. Please ensure to enclose all dependencies. I’ll then try to reproduce your bug.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349310793, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7qbOoHYPO6INK6WUXvLKQo4P08axks5s9UthgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349311531>, or mute the thread https://github.com/ notifications/unsubscribe- auth/AE5OHfU-6PFc5PPw_YYpzjzlWA3rJlNuks5s9Uv4gaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349312450, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7i-s- aL26qoHhJtdeXOnmLv0btk5ks5s9Uy2gaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349328190>, or mute the thread https://github.com/notifications/unsubscribe- auth/ AE5OHZa1YSYJZTflzi1sFgwoaPnlLh95ks5s9VlqgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349386756, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_ c7iXnBdrjcm4aoFu1Do1xdhj295Xnks5s9YPKgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349387875>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHZZwkdT9brc2wvs5rVOhDcBRKaHzks5s9YShgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46# issuecomment-349389339, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7u_ngEyBDS7A- bxT9mEebThdYchSks5s9YXLgaJpZM4PnOT1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub < https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349390644>, or mute the thread https://github.com/notifications/unsubscribe-auth/ AE5OHVIbXOHr5GB6AWQcFFajo-f-ebiHks5s9YbZgaJpZM4PnOT1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MBB-team/VBA-toolbox/issues/46#issuecomment-349580538, or mute the thread https://github.com/notifications/unsubscribe-auth/AV_c7sDydZ23ev84-R-DfL1shUkW1W4Kks5s9lvWgaJpZM4PnOT1 .

MBB-team / VBA-toolbox

VBA_MFX is slowing down #46