Replies: 5 comments 2 replies
-
Thanks to @magv I now know that the poles can be found inside the However, I'm now struggling with some integrals where even that part doesn't really run through. For example,
Here I never even get to the stage of having the integral library source files ready. And the memory consumption easily exceeds 1TB (!) without much happening. Is there perhaps a better way to deal with this integral using pySecDec? |
Beta Was this translation helpful? Give feedback.
-
Hi, so to document: for your example the time is spend calculating the contour deformation jacobian, which you don't need if only the poles are of interest -- disabling contour deformation should speed it up. A better solution is to use one of the sector decomposition functions directly to figure out the poles. We'll try to add a simple example for how to do this at some point. |
Beta Was this translation helpful? Give feedback.
-
Hi Vitaly, many thanks for your help. Upon deactivating contour deformation, the above example went through really fast while the memory consumption remained pretty low. However, there seem to be some other cases where this trick doesn't really help. For example,
Here everything goes well until the point where the log outputs I'm wondering if there is perhaps some trick, which could help here as well? |
Beta Was this translation helpful? Give feedback.
-
Not quite sure what's going on (maybe complicated calculation of subtraction terms), but to skip writing integrands you'll need to use sector decomposition functions directly. |
Beta Was this translation helpful? Give feedback.
-
In release 1.6.5 we have now introduced the function loop_integral.leading_order(). |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi,
I have a calculation where I only need to extract the highest pole of the amplitude. This means that
a lot of master integrals get thrown away since their leading pole starts say at 1/ep or 1/ep^3 instead
of 1/ep^6.
To weed out the irrelevant integrals I'm currently using pySecDec as a mean to check which poles does
the given integral has. At this point the only information I need is that there is a pole at 1/ep^x. It's numerical
value is irrelevant for me.
However, I noticed that some integrals are so complicated, that even this kind of simple calculation takes
days and generates multiple GBs of C++-libraries. For example
and
Since pySecDec actually needs to resolve all the poles when preparing the sector decomposition, I'm wondering if there's a better way to acheive what I want rather than just using
requested_orders = [-1]
. What I need is just something likefor each integral, not more than that.
Any help would be much appreciated!
Cheers,
Vlad
Beta Was this translation helpful? Give feedback.
All reactions