How to completely reset Matlab's workspace?

74 次查看（过去 30 天）

Zohar 2017-9-19

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/357286-how-to-completely-reset-matlab-s-workspace

评论： Walter Roberson 2022-9-2

How can I reset Matlab's workspace (environment) as if I restarted it without actually restarting it (to avoid overhead)? Perhaps specifically to trigger garbage collection?

This is in order to resolve the problem of Matlab degrading and become slower over time, probably due to poor memory management:

https://www.mathworks.com/matlabcentral/answers/331930-can-anyone-explain-why-matlab-gets-slower-and-slower-until-restart-if-large-cell-or-struct-arrays-ar

https://groups.google.com/forum/#!topic/yalmip/VNoBeBKc9vM

3 个评论
显示 1更早的评论隐藏 1更早的评论

Adam 2017-9-19

在 MATLAB Online 中打开

clear all

will also remove functions and classes from the cache which may further help to trigger garbage collection (for class objects). It may not though.

Zohar 2017-9-20

在 MATLAB Online 中打开

As I mention in the second link, running the following every iteration doesn't help:

clear classes; clear functions; dbclear all; clear all; pack;

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Jan 2017-9-19

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/357286-how-to-completely-reset-matlab-s-workspace#answer_282238

编辑：Jan 2017-9-19

在 MATLAB Online 中打开

system('matlab &')
exit

See https://www.mathworks.com/matlabcentral/answers/1093-how-do-i-reset-matlab-to-its-launched-state

8 个评论
显示 6更早的评论隐藏 6更早的评论

Louis-Philippe Guinard 2022-9-2

I'm coming back for more.

I'm using parallel processing to analyze many images at once. If I run the code once, everything is fine, but as soon as I run it a second time, I get an error saying 'Out of memory'. This is annoying because technically I have to exit and restart MatLab every time. Moreover, the updated solution Mr. Robertson has provided doesn't seem to work for me. Please also note that I boosted my laptop's RAM to 32Go, 3200MHz (up from 8Go, 2667MHz) expressively for the purpose of running parallel processing.

This means I need to reconnect to multiple workers (one for each physical core I want to use) every time I need to restart the code. Granted, it's still WAAAAAAAAAY faster than analyzing sequentially (which, for me, took more than an hour for five measly images, compared to about 7 minutes for 5 cores) but still it feels unpolished. Any solution to that would be greatly appreciated!

Thanks in advance! :)