cpptraj ideas

[hainm]

some of ideas might be better for pytraj/cpptraj api. (or might be not)

dump coordinates directly to xyz 3d or 1d array without creating Frame object.

Motivation: pytraj might use numpy Trajectory for in-memory trajectory. In the current implementation, cpptraj need to read coords from disk to Frame, then pytraj make copy to numpy array.

[hainm]

Status: implemented

Add more constructor for cpptraj's Frame class

low priority

Frame(int n_atoms, double* ptr)
Frame::Frame(int natomIn, double* ptr) :
  natom_(natomIn),
  maxnatom_(natomIn),
  ncoord_(natomIn*3), 
  T_(0.0),
  time_(0.0),
  X_(0),
  V_(0),
  Mass_(natomIn, 1.0)
{
  if (ncoord_ > 0)
    X_ = ptr;
}

Motivation : We can create contigous memory block for numpy array. Whenever creating a Frame for cpptraj's action, we just make a "view" without copying data. This saves a lot of memory.

Dan's comment This could be done, but then Frame needs to be re-worked so it doesn't try to free memory, and any routines that make use of this need to be responsible for freeing memory.

Hai: may be give the ownership to numpy.

[hainm]

list of cpptraj's Actions needed to add more Dataset

• Action_Hbonds Add Dataset_Integer to store atom indies, like mdtraj?

In [20]: md.baker_hubbard(traj)
Out[20]:
array([[ 13,  14,   9],
       [ 47,  48,  50],
       [167, 168, 170]])

But I am not sure where we need those indices. Currently cpptraj us 0 and 1 to say 'no', 'yes' for the existence of hbond for each frame. This is what I am expecting.

In [23]: pt.search_hbonds(traj)
Out[23]:
<pytraj.DatasetList with 10 datasets>
total_solute_solute
[5 5 5 4 4 5 6 5 5 4]

LYS12_O-SER1_N-H3
[1 1 1 1 0 1 0 1 0 1]

LYS8_O-GLU5_N-H
[1 1 0 1 1 0 1 1 1 1]
...

TRP2_O-THR3_OG1-HG1
[0 0 0 0 1 1 1 0 0 0]

THR3_O-THR10_N-H
[0 0 0 0 0 1 1 1 1 0]

But it's also great if we can take the atom indices and inspect distances, angles later (example below)?

In [31]: traj = pt.iterload("./data/tz2.nc", "./data/tz2.parm7")

In [32]: pt.angle(traj, indices)
Out[32]:
array([[ 128.26687651,  140.4319339 ,  130.87185144, ...,   98.56379937,
         107.85608806,   89.34456598],
       [  17.55336985,  118.77424219,   27.46849459, ...,    7.05500858,
          88.28587392,   81.54357284],
       [  52.47597634,   49.17080732,   45.59776004, ...,   47.63029318,
          71.20751101,   80.74681557]])

[hainm]

delay data processing in cpptraj for clustering?