Added three files. One for MACE model, one for python script, and one…

… for the Slurm script.

src/gpu_a100_job.sh

@@ -0,0 +1,21 @@
+#!/bin/bash -l
+#SBATCH --job-name=test
+#SBATCH --account=pc_chippschao
+#SBATCH --partition=es1
+#SBATCH --qos=es_lowprio
+#SBATCH --nodes=1
+#sBATCH --ntasks=1
+#SBATCH --cpus-per-task=2
+#SBATCH --gres=gpu:V100:1
+#SBATCH --time=24:00:00
+
+source activate quacc
+
+mkdir -p a100
+
+cp test1.py a100
+
+cd a100
+  python test1.py
+cd ..
+

src/test1.py

@@ -0,0 +1,125 @@
+import os
+from ase.io import read
+from ase.neb import NEB
+from ase.neb import NEBTools
+from ase.optimize import BFGS
+import matplotlib.pyplot as plt
+from ase.calculators.emt import EMT
+from ase.mep.neb import NEBOptimizer
+from ase.optimize import BFGS, ODE12r
+
+
+def calc():
+    from mace.calculators import MACECalculator
+    mlcalculator = MACECalculator(
+        model_paths='/global/home/users/kumaranu/Documents/gpu_jobs/MACE_model.model',
+        device='cuda',
+        default_dtype="float64",
+    )
+    return mlcalculator
+
+
+def setup_images():
+    N_intermediate = 40
+    # xyz_r_p = '/global/home/users/kumaranu/trash/inputs/abcd/A_B.xyz'
+    # xyz_r_p = '/global/home/users/kumaranu/trash/inputs/abcd/A_C+D.xyz'
+    xyz_r_p = '/global/home/users/kumaranu/trash/inputs/abcd/B_C+D.xyz'
+    reactant = read(xyz_r_p, index='0')
+    reactant.calc = calc()
+    product = read(xyz_r_p, index='1')
+    product.calc = calc()
+    os.makedirs('/tmp/geodesic_calc_dir', exist_ok=True)
+    os.chdir('/tmp/geodesic_calc_dir')
+    os.system('geodesic_interpolate ' + str(xyz_r_p) + ' --output output.xyz --nimages ' + str(N_intermediate))
+    intermediate = read('output.xyz', index=':')
+
+    qn = ODE12r(reactant)
+    qn.run(fmax=1e-3, steps=1000)
+
+    qn = ODE12r(product)
+    qn.run(fmax=1e-3, steps=1000)
+
+    images = [reactant]
+    for image in intermediate:
+        image.calc = calc()
+        images.append(image)
+    images.append(product)
+    return images, None, None
+
+
+def run_neb_method(method, optimizer, precon, optmethod, testdir='trash/'):
+    images, _, _ = setup_images()
+
+    fmax_history = []
+
+    def save_fmax_history(mep):
+        fmax_history.append(mep.get_residual())
+
+    k = 0.1
+    if precon == 'Exp':
+        k = 0.01
+    mep = NEB(
+              images,
+              k=k,
+              method=method,
+              climb=False,
+              precon=precon,
+              remove_rotation_and_translation=False,
+              parallel=True,
+             )
+
+    os.makedirs(testdir, exist_ok=True)
+    log_filename = f'neb_band_{method}_{optimizer.__name__}_{precon}.txt'
+    logfile_path = os.path.join(testdir, log_filename)
+
+    if optmethod is not None:
+        opt = optimizer(mep, method=optmethod, logfile=logfile_path)
+    else:
+        opt = optimizer(mep, logfile=logfile_path)
+
+    opt.attach(save_fmax_history, 1, mep)
+    opt.run(fmax=1e-2, steps=1000)
+
+    nebtools = NEBTools(images)
+
+    Ef, dE = nebtools.get_barrier(fit=False)
+    print(f'{method},{optimizer.__name__},{precon} '
+          f'=> Ef = {Ef:.3f}, dE = {dE:.3f}')
+
+    # Save the NEB band plot
+    output_filename = f'neb_band_{method}_{optimizer.__name__}_{precon}.png'
+    output_path = os.path.join(testdir, output_filename)
+
+    # Plot and save the NEB band
+    fig, ax = plt.subplots()
+    nebtools.plot_band(ax=ax)
+    plt.savefig(output_path)
+    plt.close(fig)
+
+
+if __name__ == "__main__":
+    # Define multiple sets of inputs
+    input_sets = [
+        #('aseneb', BFGS, None, None),
+        #('improvedtangent', BFGS, None, None),
+        #('spline', BFGS, None, None),
+        #('string', BFGS, None, None),
+        ('aseneb', NEBOptimizer, None, None),
+        ('improvedtangent', NEBOptimizer, None, None),
+        ('spline', NEBOptimizer, None, None),
+        ('string', NEBOptimizer, None, None),
+    ]
+
+    # Set the test directory (you can modify this based on your needs)
+    testdir = '/global/home/users/kumaranu/Documents/gpu_jobs'
+
+    # Iterate over input sets and call run_neb_method
+    for input_set in input_sets:
+        run_neb_method(
+            method=input_set[0],
+            optimizer=input_set[1],
+            precon=input_set[2],
+            optmethod=input_set[3],
+            testdir=testdir,
+        )
+