# Insufficient RAM for computing newforms

In a Mac OS X, with a 2.5Ghz processor and 4Gb RAM I ran the following lines in Sage:

D = DirichletGroup(20)
g = D[7].extend(1600) # order 4 character
N = Newforms(g,2,names='a')


In two hours the 4Gb were full and it started writing to swap. Is it normal that 4Gb RAM is not enough to perform the above computation? I'm new to Sage (and to the forum) but since Sage tells me that the space

S = ModularSymbols(g,2,sign=1).cuspidal_subspace().new_submodule()


has dim 34 I was expecting it to be within the powers of my computer. Thanks, Nuno

edit retag close merge delete

I looked into this by typing "S=ModularSymbols(g,2,sign=1).cuspidal_subspace().new_submodule()" into sage-4.7.2 and was surprised that it completely crashes with a major error in the new_submodule computation. I'm opening a blocker ticket about this. Sorry I didn't notice your post by the way, but I hadn't been checking ask.sagemath enough lately. If S could be constructed, I would suggest perhaps directly computing q-expansions by decomposing S itself -- there is some overhead in the Newforms command that you can avoid. http://trac.sagemath.org/sage_trac/ticket/12020

( 2011-11-13 14:31:31 +0100 )edit

Sort by » oldest newest most voted

This only took a few minutes in 6.4.beta3:

sage: version()
'Sage Version 6.4.beta3, Release Date: 2014-09-10'
sage: D = DirichletGroup(20)
sage: g = D[7].extend(1600)
sage: S = ModularSymbols(g,2,sign=1).cuspidal_subspace().new_submodule()
sage: S
Modular Symbols subspace of dimension 34 of Modular Symbols space of dimension 240 and level 1600, weight 2,   character [-1, 1, -zeta4], sign 1, over Cyclotomic Field of order 4 and degree 2

more

In Sage 6.1 on MacOS X in a Fedora 20 VM on VirtualBox. Model: iMac Processor: Intel Core i7 (QuadCore) RAM: 16 GB.

The VM needs a lot of CPU-Time and the use of RAM increases constantly.

Start: 1.9 GB RAM After: 30 Min. 2.3 GB RAM After: 60 Min. 2.6 GB RAM

I cancelled the calculation after 60 Minutes.

more