Path integral #144

ajschult · 2023-01-23T23:05:43Z

placeholder

unclear why results are sensitive to shift, but they are and shifting based on ring COM doesn't work

avoid trouble when ring0 is split across PBC, so just use atom0

2. computed dt=1/omegaMax (real, nm, staging)

…M-EC and Stage-EC as expected!

…e always get the infinite-beads with harmonic model, even with a single-bead.

Added HO-PI LT integrators. All meters are consistent for HO/AO.

…M/Stage data are more correlated.

…or simple NM to have same relative mi's as LAMMPS.

…aled up by expX

…c needs fix.

…ich also has an exact solution.

ajschult and others added 30 commits January 20, 2023 09:41

need to include shift0

58cfda5

add En writer

f0a79f8

need to compute shift for previous config too

811effd

shift so that all stage-0 beads have no net drift

e3f91d5

unclear why results are sensitive to shift, but they are and shifting based on ring COM doesn't work

try to get NM HMA working for LJ

fd6f42a

cleanup

41cefdf

fix up NM HMA (working now, at least for n=1)

0b7336c

pass appropriate parameter (still not working for n>1)

7999674

get meters working again for 1D AO

2ca3b3e

fix Jacobian and COM terms of HMA and HMA2

8b24839

exclude COM term for AO

462b8eb

be more careful about computing shift0

09dbc45

avoid trouble when ring0 is split across PBC, so just use atom0

small fix for COM shift (result still wrong)

10e0ffb

just do what NM does because NM is good and wholesome.

a6007a4

PIMD 1D AO - needs fix

9da7f98

fixed to work with PIMD

9bc133e

Working PIMD with NM coord

0668dc5

1. omegaN->omegaN/nBeads

aeb29ff

2. computed dt=1/omegaMax (real, nm, staging)

COM of NM-EC is now like real and Staging-EC

4ab3ff2

fixed fictitious masses and timestep for PIMD (all 5 coords)

b542385

corrected dt-real to be c/(sqrt(n)*w0)

a295368

PIMC using all 5 coord. Simple NM needs fix.

94aebc2

make q array of vectors for PIMC

efe0e47

adjusted meters to work with 1D AO. Simple stage needs fix.

e8fdb80

Added err of Cv. HMAc and CentVir are not consistent with true HO.

dfa4edd

Fixed meters for Cvn. All working finr for HO, with zero error from N…

9b06e74

…M-EC and Stage-EC as expected!

fixed alpha=0 case. Simple_stage for n>1 needs fix.

a19a0b8

Try to avg Cv over nShifts

941eb36

adjusted dt for simple stage using analytic scaling s.

3112965

All meters are consistent in 3D LJ

14a3011

sabrygad and others added 30 commits January 4, 2024 19:50

Added Cv (prim, vir, cvir) based on the HO-PI action. Just like En, w…

75dadc7

…e always get the infinite-beads with harmonic model, even with a single-bead.

Added Cv (NM-HMA) based on the HO-PI action.

8aa208b

Added Cv (EC Staging) based on the HO-PI action.

899482e

Added HMAc with HO-PI.

bfc0371

Added HO-PI LT integrators. All meters are consistent for HO/AO.

Get all E meters consistent for LJ, for all coordTypes. However, EC N…

f2cda95

…M/Stage data are more correlated.

Added Cv for lj.

f87f4af

Did Cv LJ, Staging needs fix

0b58c53

Staging Cv LJ is now working!

62a8409

All PI Cv meters are done and consistent, and AO/HO are still working.

8098236

Used EnShift with LJ Cv PIMD meters to avoid roundoff, and fixed m0 f…

900e055

…or simple NM to have same relative mi's as LAMMPS.

First try of HO-PI LJ.

32b5d91

Try fix COM for real LT HO-PI LJ.

ddfdc15

actually fix COM

3e0c20a

include full velocity to prevent roundoff from accumulating and/or sc…

47fa4ef

…aled up by expX

Revisiting PIMC LJ.

6c94f9e

Add COM to PIMD HO/AO.

3b4ce0c

1.Corrected mi ring NM/stage, 2. Corrected Eigenvectors.

c11e29a

Corrected NM uhNew.

2e580bf

1. Added BAOAB PIMD integrator using exact A. 2. using gamma_LT=omega.

2c935d2

Using NM as used in literature and cleaned up mi masses.

c48a4d0

added onlyCentroid.

00ae014

for loops are faster

a573c9b

add faster move hardcoded for HO model

a12d387

added primitive and CentVir pressure estimators to LJ PIMD

10faa6a

Started Cv using finite difference of d2Udb2. Vir and CVir works. HMA…

7cd34ae

…c needs fix.

FD Cv using HMAc and HMAq-NM. No match still.

25dab91

Include RP/RP+HO forces in the B part of the BAOAB PIMD splitting, wh…

c46f97f

…ich also has an exact solution.

Add Cayley A propagator to the PIMD (BAOAB).

7780d55

take optional constant to subtract

ec19f53

Finite difference for Cv LJ. Staging needs fix,

c547642

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Path integral #144

Path integral #144

ajschult commented Jan 23, 2023

Path integral #144

Are you sure you want to change the base?

Path integral #144

Conversation

ajschult commented Jan 23, 2023