-
Notifications
You must be signed in to change notification settings - Fork 161
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Make a safer/easier TRIM_PERM #3576
Comments
One option for an improvement would be, whenever you trim a perm, to also trim its stored inverse (if it has one) to the same length, rather than just throwing it away, which is what #3575 implements. |
There is another issue with |
Hmm, off the top of my head I don't see a better solution for HPC-GAP than the one you suggested. The problem is that permutations are public, so we can't special case the situation where only the current thread is using one. |
Swapping master pointers doesn't help, as once a thread has looked at a permutation and found it's a perm4, it will continue treating it like one, something like:
|
TRIM_PERM/TrimPerm is used when a permutation's internal representation contains values bigger than it's LargestMovedPoint -- alll such values must map to themselves. In this case we can save memory by shrinking the permutation by removing all such values.
TRIM_PERM requires as an input a size to trim the permutation to. This has two problems:
A simple method would be to just calculate the LargestMovedPoint, and trim to that. This would make a method which was easier to use, and couldn't be used to corrupt memory (except in HPC-GAP, which is a different problem).
I suggest we introduce such a method (I can't think of a good name.. SquashPerm? ShrinkPerm?), and port all uses of TrimPerm over to it.
We could in principle make a general method (ShrinkAllocation?) and implement it for all types (Perm, Trans, Plists, records,...) which would try to free any unused memory. This method already exists for plists and strings, and is called ShrinkAllocationPlist and ShrinkAllocationString.
The text was updated successfully, but these errors were encountered: