Repository navigation
mthsurf #4
Description
Activity
Hi Nik,
I agree but sometimes we need higher resolution for perturbation analysis.
For example, mtheta=128 is not enough for m=64, as it violates Nyquist
criterion and causes aliasing (as you know). I believe we have input
parameter somewhere controlling mthvac (or mthsurf). I would leave the
choice..Jong-Kyu
On Tue, Jun 9, 2015 at 5:02 PM, Nikolas Logan notifications@github.com
wrote:Why do we use mthvac for mthsurf and why is this so much larger than any m
limit would suggest is necessary? This is responsible for large output
ascii files and slow fun_flag operation, but I suspect it is not necessary
for any fundamental calculations.Suggest: change to mthsurf = mtheta in idcon.f.
What do you think @parkjk https://github.com/parkjk ?
—
Reply to this email directly or view it on GitHub
#4.Why is the typical inherited mthvac 960? This seems very high.
I see two courses: 1) Reduce the default mthvac to more reasonable number like 256 or 2) set mthsurf as something like 2*nyquist using max(abs(mlow),mhigh).
Hi Nik,
We can make the default as mthvac=4_max(abs(mlow),mhigh)) (=2_Nyquist
limit) as you suggest and also give an option to choose it.Jong-Kyu
On Tue, Jun 9, 2015 at 5:39 PM, Nikolas Logan notifications@github.com
wrote:Why is the typical inherited mthvac 960? This seems very high.
I see two courses: 1) Reduce the default mthvac to more reasonable number
like 256 or 2) set mthsurf as something like 2*nyquist using
max(abs(mlow),mhigh).—
Reply to this email directly or view it on GitHub
#4 (comment)
.Added mthsurf to IPEC_INPUT and enforced that it be at least 4*max(abs(mlow),mhigh)) in idcon.
This was done in the output_size feature branch.
Release 0.1.0 is giving errors when dcon mthsurf is < 4*Nyquist. Bug!
A mthsurf bug was fixed in b61e608, which showed perfect reproduction of the equilibrium in 10% the time (see commit for times, and plots).
I am not closing this issue because mthsurf requires ahg2msc.out to be in the run directory, which often causes an error (I run IPEC multiple places for one DCON run). Perhaps @parkjk can weight in on our options:
- Include mscfile in IPEC_INPUT.
- Edit vacuum code mscvac to except this file as input, pass it to dskmd1, and then to readahg.
- Pros: Fixes mthsurf "bug". Consistent with standard IPEC IO. Backwards compatible.
- Cons: Lots of edits to vacuum. IPEC list of file inputs is unwieldy (idcon,ieq,ivacuum,rdcon,ahg,...)
- Include dcon_dir in IPEC_INPUT (my preference)
- Edit vglabal_mod to include this, and readahg to read dcon_dir//'ahg2msc.out'
- Pros: Minimal changes to vacuum. Consolidates all the IPEC file inputs into one parameter. Can be made backwards compatible (idconfile overrides dcon_dir//'euler.bin').
- Cons: Gradual change of IPEC IO standard.
- Leave as is.
- Despite its gigantic savings, using mthsurf will be error prone and probably not used.
Why do we use mthvac for mthsurf and why is this so much larger than any m limit would suggest is necessary? This is responsible for large output ascii files and slow fun_flag operation, but I suspect it is not necessary for any fundamental calculations.
Suggest: change to mthsurf = mtheta in idcon.f.
What do you think @parkjk ?