@scallopedpancake So I thought it was magnitudes faster: turns out it's actually slower (I hadn't set up the nodes properly so it wasn't actually doing anything). The calcuations on the nodes seem to be fast, but collating the result takes a very long time.

@ecterrab Thanks for letting me know: sorry I completely blanked on the format! Hopefull thius should work SmallExample.mw This is just an example of Euclidean gravity. The momentum of the h[4,4] field should vanish, but it doesn't vanish for the right reasons in the code. In particular, Eq. 10 is clearly wrong: symbol is a bug that I don't understand; Eq. 11-12 would be fine but Maple doesn't seem to be differentiating indies correctly, since there should be some delta[mu,4]s there but there aren't

@Hullzie16 Hi, here's a short example of 4D Euclidean linearised gravity, where it fails. SmallExample.maple 

@janhardo I appreciate the sentiment, but that code is wrong. By writing coordindates=[t,x1,x2,x3], it has defined four coordinate systems.

@acer Thanks for that, I thought it would just recognise ++ and +=. I'll do some more reading on the compiler: I don't think I considered it being a strict compiler since I've not had any issues in the past.

I've added my MWE in an edit. Alternatively, here (MWE.mw) it is as a worksheet.

From https://www.mapleprimes.com/posts/209047-Minimize-The-Number-Of-Tensor-Components, it appears Maple cannot use identities when reducing the number of components. Maybe this is the issue?

@tomleslie As mentioned by @acer it's a subpackage.
 

Download MWE.mw

This is also the case in Maple 2020 as I have just verified.