🍦 Cone Sampling

📖 In the Cone Sampling step, NANR systematically explores the branching plane around each optimized MECI to understand how different displacement directions affect photochemical outcomes.

---

Concept Overview

• The branching plane is defined by two orthogonal vectors, g (gradient difference) and h (nonadiabatic coupling).
• New geometries — cone points — are generated by displacing the MECI geometry along linear combinations of g and h.
• Each cone point represents a possible direction for the molecule to evolve following nonadiabatic transitions.
• These cone points serve as starting geometries for the Relaxation stage.

---

Step-By-Step Cone Sampling

1. Create cone_sampling/ inside each MECI directory

   mkdir <molecule>/seam_sampling/clustering/ciopt0/cone_sampling
   

2. Copy in your optimized MECI geometry

   cp ../ci0.xyz cone_sampling/
   

3. Create sampling.yaml

   engine:
     type: terachem
     gpus: 0
     cwd: .
   
   options:
     method: rhf
     basis: 6-31gs
     precision: mixed
     scf: diis+a
     casscf: 'yes'
     closed: 20
     active: 5
     cassinglets: 4
     casguess: 'c0.casscf'
     castarget: 2
     castargetmult: 1
     casscftrustmaxiter: 1000
     casscfmaxiter: 1000
     casscforbnriter: 1000
     cphfiter: 1000
     cpsacasscfmaxiter: 1000
   
   cone:
     sampling: radial
     lower_state: 2
     upper_state: 3
     samples: 60
     radii_list: [0.10]
   

4. Create a submission script submit.sh

   #!/bin/bash
   #SBATCH -p l40-gpu
   #SBATCH -N 1
   #SBATCH -n 4
   #SBATCH -J cone_sample
   #SBATCH --mem=50G
   #SBATCH -t 1:00:00
   #SBATCH --qos gpu_access
   #SBATCH --gres=gpu:1
   
   module load tc/25.03
   python3 /<path>/NonadiabaticNanoreactor/scripts/gh_sampling.py sampling.yaml
   

5. Submit your cone sampling job

   cd <molecule>/seam_sampling/clustering/ciopt0/cone_sampling
   sbatch submit.sh
   

---

Notes

The radii_list keyword specifies the displacement magnitude (in Å) for each cone point. Multiple values can be supplied, e.g. [0.05, 0.10, 0.15].

Ensure that your MECI optimization finished successfully before running this step.

---

Outputs

• ci0_radial_XXXX.xyz → individual cone-point geometries
• sampling.yaml → input record for reproducibility
• submit.sh → TeraChem submission script

---

Continue to 🏖️ Relaxation →