Access certain group..

41 views
Skip to first unread message

Elsaid Younes

unread,
Aug 16, 2021, 5:36:14 AMAug 16
to hoomd...@googlegroups.com
Dear all,

I wonder How can I access particles with type'B'

GSD file:


data1 = np.linspace([-451.0,-451.0,-451.0], [451.0,451.0,451.0], 2940)
np.random.shuffle(data1[:,0])
np.random.shuffle(data1[:,1])

data2 = np.linspace([-376.0,-376.0,-376.0], [376.0,376.0,376.0], 21)
np.random.shuffle(data2[:,0])
np.random.shuffle(data2[:,1])

result = [*data1,*data2]

s = gsd.hoomd.Snapshot()
s.particles.N = 2961
s.particles.typeid = [0]*2940 + [1]*21
s.particles.types = ['A']*2940 + ['B']*21
s.particles.diameter = [4.0]*2940 + [150.0]*21
s.configuration.box = [906.0,906.0,906.0,0,0,0]
s.particles.charge = [+1]*2940 + [-140]*21
s.particles.position = result

with gsd.hoomd.open(name='file.gsd', mode='wb') as f:
    f.append(s)

Best regards,
Elsaid Mohamed

Eric Irrgang

unread,
Aug 16, 2021, 8:16:42 AMAug 16
to hoomd...@googlegroups.com
Hi Elsaid.

1. There is a logic error in the script you share. Per https://gsd.readthedocs.io/en/stable/hoomd-examples.html#define-a-snapshot, the `types` field is supposed to be a list of types, such that the indices of the list correspond to the integers used in the `typeid` field.

2. You would find particles of a given type by looking for particles with a given typeid. Python has several ways to do this for lists, but since GSD provides a NumPy interface, you could do something like

>>> s.particles.typeid = [0]*2940 + [1]*21
>>> s.particles.types = ['A', 'B']
>>> with gsd.hoomd.open(name='file.gsd', mode='wb') as f: f.append(s)
...
>>> with gsd.hoomd.open(name='file.gsd', mode='rb') as t: frame = t.read_frame(0)
...
>>> frame.particles.position[:].shape
(2961, 3)
>>> frame.particles.typeid[:] == 0
array([ True, True, True, ..., False, False, False])
>>> frame.particles.position[frame.particles.typeid[:] == 0].shape
(2940, 3)
>>> for pos in frame.particles.position[frame.particles.typeid == frame.particles.types.index('B')]: # Do something...

Note that that last line only works if you fix the `types` as noted above.

See https://numpy.org/doc/stable/reference/arrays.indexing.html#advanced-indexing (or just https://www.w3schools.com/python/python_lists_comprehension.asp)

Best,
Eric
> --
> You received this message because you are subscribed to the Google Groups "hoomd-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hoomd-users...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/CAO9cOeB6SRTPbEE9FdKvFV-f_WGT4jXLmFOiPtAwTsJr%2BaUyUA%40mail.gmail.com.

Elsaid Younes

unread,
Aug 16, 2021, 4:00:02 PMAug 16
to hoomd...@googlegroups.com
Dear all,

GSD  file:

s = gsd.hoomd.Snapshot()
s.particles.N = 2961
s.particles.typeid = [0]*2940 + [1]*21
s.particles.types = ['A','B']

s.particles.diameter = [4.0]*2940 + [150.0]*21
s.configuration.box = [906.0,906.0,906.0,0,0,0]
s.particles.charge = [+1]*2940 + [-140]*21
s.particles.position = result

with gsd.hoomd.open(name='file.gsd', mode='wb') as f:
    f.append(s)
    

Simulation script:

import hoomd
import freud
import gsd.hoomd
import numpy as np
import matplotlib.pyplot as plt
import hoomd
from hoomd import md

hoomd.context.initialize('--mode=cpu')

system = hoomd.init.read_gsd('file.gsd')

# groupB = hoomd.group.type(name="b-particles", type='B')


with gsd.hoomd.open(name='file.gsd', mode='rb') as t: frame = t.read_frame(0)
frame.particles.position[:].shape
frame.particles.typeid[:] == 0

frame.particles.position[frame.particles.typeid[:] == 0].shape
groupB = []
for pos in frame.particles.position[frame.particles.typeid == frame.particles.types.index('B')]: groupB.append(frame.particles.types.index('B'))
print(groupB)

nl = hoomd.md.nlist.tree()

# NVT integration

hoomd.md.integrate.mode_standard (dt = 0.005)
nvt = hoomd.md.integrate.nvt (group = groupB, kT = 1.2, tau = 1.0)
nvt.randomize_velocities (seed = 1)


# fslj = pair.force_shifted_lj(r_cut=453, nlist=nl)
# fslj.pair_coeff.set('B', 'B', epsilon=150.0, sigma=1.0)



pppm = hoomd.md.charge.pppm(groupB, nl)
pppm.set_params(Nx=150, Ny=150, Nz=250, order=6, rcut=450.0, alpha=0.007320808910601237)

# run the simulation
hoomd.run (100)

rdf = freud.density.RDF(bins=150, r_max=450)
def calc_rdf(timestep):
    hoomd.util.quiet_status()
    snap = system.take_snapshot()
    hoomd.util.unquiet_status()
    rdf.compute(system=snap, reset=False)

# Equilibrate the system a bit before accumulating the RDF.
hoomd.run(100)
hoomd.analyze.callback(calc_rdf, period=10)
hoomd.run(100)
# Store the computed RDF in a file
np.savetxt('rdf_273_100s_bins=150_period=10_N(x,y,z)=256_order=2.csv', np.vstack((rdf.bin_centers, rdf.rdf)).T,
delimiter=',', header='r, g(r)')
rdf_data = np.genfromtxt('rdf_273_100s_bins=150_period=10_N(x,y,z)=256_order=2.csv', delimiter=',')
plt.plot(rdf_data[:, 0], rdf_data[:, 1])
plt.title('Radial Distribution Function')
plt.xlabel('$r$')
plt.ylabel('$g(r)$')
plt.show()

Error message:

----
Traceback (most recent call last):

  File "/home/elsaid/untitled4.py", line 28, in <module>
    nvt = hoomd.md.integrate.nvt (group = groupB, kT = 1.2, tau = 1.0)

  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/md/integrate.py", line 221, in __init__
    thermo = hoomd.compute._get_unique_thermo(group=group);

  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 282, in _get_unique_thermo
    res = thermo(group);

  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 209, in __init__
    if group.name != 'all':

AttributeError: 'list' object has no attribute 'name'

Best,
Elsaid


Eric Irrgang

unread,
Aug 17, 2021, 7:31:52 AMAug 17
to hoomd...@googlegroups.com
Hi All,

Please note that the code snippets in my previous email were not intended as solutions to any particular problem. I only intended to illustrate possibilities.

I apologize for the formatting of the code snippets. I was using three '>' marks to indicate the Python interpreter prompt but that is common email syntax for quoting.

Given the context,

* `frame.particles.position[:].shape` would be expected to produce `(2961, 3)`
* `frame.particles.typeid[:] == 0` would produce a NumPy array boolean values with the same shape as the original
* `frame.particles.position[frame.particles.typeid[:] == 0].shape` would give `(2940, 3)`

I apologize for any confusion.

Eric Irrgang

unread,
Aug 17, 2021, 7:39:39 AMAug 17
to hoomd...@googlegroups.com
Hi Elsaid.

You are providing the wrong type of argument to hoomd.md.integrate.nvt.

https://hoomd-blue.readthedocs.io/en/stable/module-md-integrate.html#hoomd.md.integrate.nvt says that the `group` parameter is a `hoomd.group` object. You have provided a list.

To create a `hoomd.group` based on particle type, see https://hoomd-blue.readthedocs.io/en/stable/module-hoomd-group.html#hoomd.group.type

Note that creating a `hoomd.group` does not give the user any access to particles. It is only used in the parameters of other hoomd commands.

Best,
Eric
> To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/CAO9cOeBPHjhzEPn-G38kgw0gO2BumYURG4gT-sg4H%2BwFQdt%2B3w%40mail.gmail.com.

Elsaid Younes

unread,
Aug 25, 2021, 2:49:12 PMAug 25
to hoomd...@googlegroups.com
Dear Eric,

I still cannot access type 'B'


hoomd.context.initialize('--mode=cpu')

system = hoomd.init.read_gsd('file.gsd')

with gsd.hoomd.open(name='file.gsd', mode='rb') as t: frame = t.read_frame(0)
frame.particles.position[:].shape
frame.particles.typeid[:] == 0
frame.particles.position[frame.particles.typeid[:] == 0].shape
groupB = []
for pos in frame.particles.position[frame.particles.typeid == frame.particles.types.index('B')]: groupB.append(frame.particles.types.index('B'))
print(groupB)

nl = hoomd.md.nlist.tree()

# NVT integration

groupB = hoomd.group.type('B');
hoomd.md.integrate.mode_standard (dt = 0.005)
nvt = hoomd.md.integrate.nvt (group = groupB, kT = 4.11433*(10**-24), tau = 1.0)
nvt.randomize_velocities (seed = 1)




pppm = hoomd.md.charge.pppm(groupB, nl)
pppm.set_params(Nx=150, Ny=150, Nz=250, order=6, rcut=450.0, alpha=0.007320808910601237)

# run the simulation
hoomd.run (100)

rdf = freud.density.RDF(bins=150, r_max=450)
def calc_rdf(timestep):
    hoomd.util.quiet_status()
    snap = system.take_snapshot()
    hoomd.util.unquiet_status()
    rdf.compute(system=snap, reset=False)

# Equilibrate the system a bit before accumulating the RDF.
hoomd.run(100)
hoomd.analyze.callback(calc_rdf, period=10)
hoomd.run(100)
# Store the computed RDF in a file
np.savetxt('rdf_273_100s_bins=150_period=10_N(x,y,z)=256_order=2.csv', np.vstack((rdf.bin_centers, rdf.rdf)).T,
delimiter=',', header='r, g(r)')
rdf_data = np.genfromtxt('rdf_273_100s_bins=150_period=10_N(x,y,z)=256_order=2.csv', delimiter=',')
plt.plot(rdf_data[:, 0], rdf_data[:, 1])
plt.title('Radial Distribution Function')
plt.xlabel('$r$')
plt.ylabel('$g(r)$')
plt.show()  


When I don't use hoomd.group.type(), it gives me errors like :

  File "/home/elsaid/untitled4.py", line 28, in <module>
    nvt = hoomd.md.integrate.nvt (group = groupB, kT = 4.11433*(10**-24), tau = 1.0)


  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/md/integrate.py", line 221, in __init__
    thermo = hoomd.compute._get_unique_thermo(group=group);

  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 282, in _get_unique_thermo
    res = thermo(group);

  File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 209, in __init__
    if group.name != 'all':

AttributeError: 'list' object has no attribute 'name'

Best,
Elsaid

Eric Irrgang

unread,
Aug 26, 2021, 10:36:17 AMAug 26
to hoomd...@googlegroups.com
Hi Elsaid.

> When I don't use hoomd.group.type(), it gives me errors like :
>
> File "/home/elsaid/untitled4.py", line 28, in <module>
> nvt = hoomd.md.integrate.nvt (group = groupB, kT = 4.11433*(10**-24), tau = 1.0)
>
> File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/md/integrate.py", line 221, in __init__
> thermo = hoomd.compute._get_unique_thermo(group=group);
>
> File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 282, in _get_unique_thermo
> res = thermo(group);
>
> File "/home/elsaid/anaconda3/lib/python3.8/site-packages/hoomd/compute.py", line 209, in __init__
> if group.name != 'all':
>
> AttributeError: 'list' object has no attribute 'name'

This is because you have assigned a list to `groupB` in the code above `groupB = hoomd.group.type('B');`, and a list is an invalid argument type for `hoomd.md.integrate.nvt (group = ...)`


> I still cannot access type 'B'

What do you mean by "access"?

Last time, I thought you meant that you wanted to inspect the particle data for particles of type 'B', so I suggested some syntax for exploring the particle data structure. But then it seemed like you were trying to define the group to be passed to the integrator, so we suggested `hoomd.group.type()`.

At a quick glance, your script looks fine. Are you getting an error?


> hoomd.md.integrate.mode_standard (dt = 0.005)
> nvt = hoomd.md.integrate.nvt (group = groupB, kT = 4.11433*(10**-24), tau = 1.0)

Depending on your other basic units, this temperature may be approximately equal to absolute zero. Also, if you are using the single-precision build (default), the equations of motion may evaluate to zero change due to floating point precision issues.

It's generally a good idea to choose units in HOOMD such that important values are as close to 1 in magnitude as possible. More than three orders of magnitude above or below 1 for parameters can lead to precision problems (too much truncation error).

See also https://hoomd-blue.readthedocs.io/en/stable/units.html

Best,
Eric

Joshua Anderson

unread,
Aug 26, 2021, 1:09:22 PMAug 26
to hoomd...@googlegroups.com
All,

I just want to clarify that single precision is not the default. All HOOMD binaries I provide are double precision, as is the default setting when you build from source.
------
Joshua A. Anderson, Ph.D.
Research Area Specialist, Chemical Engineering, University of Michigan
> --
> You received this message because you are subscribed to the Google Groups "hoomd-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hoomd-users...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/D7DC66BC-5F31-4C14-BA2C-9E4B7D50B6CA%40gmail.com.

Elsaid Younes

unread,
Aug 26, 2021, 2:21:54 PMAug 26
to hoomd...@googlegroups.com
Dear Eric,

When I write groupB = hoomd.group.type('B'), it does not pass that type to the NVT,   the r.d.f. calculated for the other type (( 
since I have two types of particles ['A','B'] with different diameters)) here are the calculated r.d.f.
For hoomd.group.type('A'):
HOOMD-blue 2.9.6 DOUBLE HPMC_MIXED TBB SSE SSE2 SSE3
Compiled: 03/17/2021
Copyright (c) 2009-2019 The Regents of the University of Michigan.
-----
You are using HOOMD-blue. Please cite the following:
* J A Anderson, J Glaser, and S C Glotzer. "HOOMD-blue: A Python package for
  high-performance molecular dynamics and hard particle Monte Carlo
  simulations", Computational Materials Science 173 (2020) 109363
-----
HOOMD-blue is running on the CPU
notice(2): Group "all" created containing 2961 particles
-----
You are using tree neighbor lists. Please cite the following:
* M P Howard, J A Anderson, A Nikoubashman, S C Glotzer, and A Z
  Panagiotopoulos. "Efficient neighbor list calculation for molecular simulation
  of colloidal systems using graphics processing units", Computer Physics
  Communications 203 (2016) 45--52
* M P Howard, A Statt, F Madutsa, T M Truskett, and A Z Panagiotopoulos.
  "Quantized bounding volume hierarchies for neighbor search in molecular
  simulations on graphics processing units", Computational Materials Science 164
  (2019) 139--146
-----
notice(2): Group "type A" created containing 2940 particles
-----
You are using PPPM. Please cite the following:
* D N LeBard, B G Levine, S A Barr, A Jusufi, S Sanders, M L Klein, and A Z
  Panagiotopoulos. "Self-assembly of coarse-grained ionic surfactants
  accelerated by graphics processing units", Journal of Computational Physics 8
  (2012) 2385-2397
-----
notice(2): -- Neighborlist exclusion statistics -- :
notice(2): Particles with 0 exclusions             : 2961
notice(2): Neighbors included by diameter          : no
notice(2): Neighbors excluded when in the same body: no
** starting run **
notice(2): charge.pppm: RMS error: 5.39052e-14


image.png

For hoomd.group.type('B'):
--
notice(2): Group "all" created containing 2961 particles
-----
You are using tree neighbor lists. Please cite the following:
* M P Howard, J A Anderson, A Nikoubashman, S C Glotzer, and A Z
  Panagiotopoulos. "Efficient neighbor list calculation for molecular simulation
  of colloidal systems using graphics processing units", Computer Physics
  Communications 203 (2016) 45--52
* M P Howard, A Statt, F Madutsa, T M Truskett, and A Z Panagiotopoulos.
  "Quantized bounding volume hierarchies for neighbor search in molecular
  simulations on graphics processing units", Computational Materials Science 164
  (2019) 139--146
-----
notice(2): Group "type B" created containing 21 particles
-----
You are using PPPM. Please cite the following:
* D N LeBard, B G Levine, S A Barr, A Jusufi, S Sanders, M L Klein, and A Z
  Panagiotopoulos. "Self-assembly of coarse-grained ionic surfactants
  accelerated by graphics processing units", Journal of Computational Physics 8
  (2012) 2385-2397
-----
notice(2): -- Neighborlist exclusion statistics -- :
notice(2): Particles with 0 exclusions             : 2961
notice(2): Neighbors included by diameter          : no
notice(2): Neighbors excluded when in the same body: no
** starting run **
notice(2): charge.pppm: RMS error: 7.54673e-12

image.png
For units:
I am very confused. How shall I write:
"columb"? charge = -140 columb
"Temperature"? T = 298 K, boltzmann constant* T = kT = 4.11433*(10**-24)

Best,.
Elsaid

--
You received this message because you are subscribed to the Google Groups "hoomd-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hoomd-users...@googlegroups.com.

Joshua Anderson

unread,
Aug 26, 2021, 2:28:47 PMAug 26
to hoomd...@googlegroups.com
Elsaid,

The group that you pass to the NVT integration method sets which particle that method acts on. It DOES NOT set what particles freud uses when computing RDFs.
------
Joshua A. Anderson, Ph.D.
Research Area Specialist, Chemical Engineering, University of Michigan

> To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/CAO9cOeAq3fCng8GcVON%3Du85t2%3DLg9Uz5%2BT_uGux58RXEM7goqA%40mail.gmail.com.

Eric Irrgang

unread,
Aug 27, 2021, 6:20:18 AMAug 27
to hoomd...@googlegroups.com
Following on Josh's note, it doesn't look like you are applying any integrator to particles of type 'A'. This is fine if you intend for them to be constrained or if they are part of a rigid body, but I thought I should point that out.

P.S. Thanks to Josh for correcting me on my mistake on default precision.

Eric Irrgang

unread,
Aug 27, 2021, 6:55:54 AMAug 27
to hoomd...@googlegroups.com

> For units:
> I am very confused. How shall I write:
> "columb"? charge = -140 columb
> "Temperature"? T = 298 K, boltzmann constant* T = kT = 4.11433*(10**-24)

This depends on your choice of fundamental units. It looks like you may be assuming time units of picoseconds, but you need to choose three fundamental units with which to derive a coherent system. Then you can solve the equations at https://en.wikipedia.org/wiki/Coherence_(units_of_measurement)#Dimension-related_coherence (or https://hoomd-blue.readthedocs.io/en/stable/units.html) to get other units. Then Coulomb's Law https://en.wikipedia.org/wiki/Coulomb%27s_law will give you the appropriate conversion factor for your units of charge.

Regardless of floating-point precision, it is usually convenient to choose at least a few units such that many quantities are 1.0 (whether that is particle mass, charge, size, temperature...).

You might get more feedback from HOOMD users doing electrostatics simulations if you start a new thread to ask what combinations of units people find most useful for simulations like yours (similar temperature, particle size, charge magnitude, etc)

Best,
Eric
Reply all
Reply to author
Forward
0 new messages