.. _transition_state_optimization :

Transition State Optimization
=============================

Overview
--------

Transition State Optimization calculations involve locating the first-order saddle points on the potential energy surface.  
This is required for calculating energy barriers for chemical reactions, reaction rates and analyzing reaction mechanism.

Check the :ref:`visualiser-input` section for the allowed input types and how to upload the files.

Modules Available
-----------------

Three modules are currently available:

1. **Hybrid ML model** - Fastest and DFT-accurate using machine learning
2. **DFT** - Density Functional Theory (highest accuracy)

Hybrid ML Module Input Fields
-----------------------------
Upon selecting the **Hybrid ML** module, following inputs have to be provided: 

.. grid:: 1 2 2 2
   :gutter: 2

   .. grid-item-card:: **Charge**
      :text-align: left 

      Total charge of the molecule (e.g., 0)

   .. grid-item-card:: **Multiplicity**
      :text-align: left 

      Spin mulitplicity = 2S+1 (e.g., 1 for singlet)

   .. grid-item-card:: **Maximum Force (fmax)**
      :text-align: left 

      Maximum force to be applied for transition state optimization

   .. grid-item-card:: **Energy Change Threshold**
      :text-align: left 

      Threshold parameter for energy change 

DFT Module Input Fields
-----------------------

Upon selecting the **DFT** module, the following inputs must be provided:

Optimization Methods
--------------------

Any of the methods listed below can be chosen for the transition state optimization:

1. **Quadratic Steepest Descent** - A second order method that computes the hessian at every step during the optimization process. 
2. **geomeTRIC method** - A backend library which provides algorithms for optimizing transition states. It can perform gradient calculations and optionally include vibrational analysis to confirm the transition state. 
3. **Berny Algorithm** - An optimizer of molecular geometries with respect to the total energy. It takes energy and Cartesian gradients as inputs in each step and returns a new equilibrium structure estimate. An amalgamation of several techniques such as quasi-Newton method, redundant internal coordinates, iterative Hessian approximation, trust region scheme and linear search is implemented in the algorithm. 
4. **Eigenvector Following Method** - Another quasi-Newton like algorithm. 
5. **Nudged Elastic Band** - Finds the transition state structure only from the geometries of the reactants and products. The method converges to the minimum energy path and optimizes the saddle-point in the same job. 

The first three methods require only the initial guess transition state structure in XYZ format as input. Following inputs must accompany the input structure:

.. grid:: 1 2 2 2
   :gutter: 2

   .. grid-item-card:: **Charge**
      :text-align: left

      Total charge of the molecule (e.g., 0)

   .. grid-item-card:: **Multiplicity**
      :text-align: left

      Spin multiplicity = 2S+1 (e.g., 1 for singlet)

   .. grid-item-card:: **Basis Set Category**
      :text-align: left

      Select the basis set family (e.g., Pople, Dunning)

   .. grid-item-card:: **Basis Set**
      :text-align: left

      Select the basis set (e.g., 6-31G, 6-31+Gss)

   .. grid-item-card:: **Exchange Functional**
      :text-align: left

      Choose an exchange-correlation functional (e.g., M06)

.. note::
    We have LDA, PBE, PBE0, M06, B3LYP, CAM-B3LYP, and WB97X functionals
    available right now for the GPU calculations.

If **Add Dispersion** box is ticked, following additional inputs must be provided: 

.. grid:: 1 2 2 2
   :gutter: 2

   .. grid-item-card:: **Dispersion Method**
      :text-align: left

      Select the dispersion correction to be added (e.g., D3ZERO, D3BJ)

If **Add Solvent** is ticked, following input parameters for solvation models must be provided: 

.. grid:: 1 2 2 2
   :gutter: 2

   .. grid-item-card:: **Solvent Model**
      :text-align: left

      Choose the solvation model (e.g., PCM, SMD)

   .. grid-item-card:: **Solvent**
      :text-align: left

      Select a solvent (e.g., water, ethanol)

In case of **Nudged Elastic Band** or **Eigenvector Following Method**, the structures of the reactant, product, and the initial guess transition state (if applicable) must be provided. 

Finally, click the **Run Calculation** button to start the simulation.

Output Details
--------------

The following options are available to explore and save the results of your geometry optimization:

.. grid:: 1 1 1 1
   :gutter: 2

   .. grid-item-card:: **Optimized TS**
      :text-align: left

      To see the optimized transition state structure, select "Molecular Structure" button at the top 
      left and then select **Optimized TS** from the dropdown above the visualizer.


   .. grid-item-card:: **Trajectory**
      :text-align: left

      Click the **tsopt_trj**  to view the trajectory of the transition state optimization by selecting 
      **Molecular Structure** from the dropdown on the left and **tsopt_trj** from the dropdown on the right.

   .. grid-item-card:: **Vibrational Spectrum**
      :text-align: left

      Select **Vibrational Spectrum** from the dropdown on the left above the visualizer to view 
      the vibrational spectrum of the optimized transition state. Imaginary frequencies, if found 
      will be listed above the spectrum. 

   .. grid-item-card:: **Save Results**
      :text-align: left

      In the bottom right corner, you can save the "Metadata".