Category:Transition states - Revision history

Csheldon at 11:43, 2 June 2025

2025-06-02T11:43:22Z

← Older revision		Revision as of 11:43, 2 June 2025
Line 8:		Line 8:
	<div style="clear:both;"></div>		<div style="clear:both;"></div>

	= Static methods =		== Static methods ==
	There are three static methods implemented in VASP for finding transition states (TS) using the static, or harmonic, approach: the improved dimer method (IDM), nudged elastic band (NEB), and following the intrinsic reaction coordinate (IRC). IDM takes a guess structure for the TS and then relaxes it along a trial unstable vibrational mode to a first-order saddle point, i.e. the TS. NEB takes the reactant and product structures and interpolates structures between the two to model the reaction pathway. IRC starts at the TS and follows the TS' unstable vibrational mode to reactant and product.		There are three static methods implemented in VASP for finding transition states (TS) using the static, or harmonic, approach: the improved dimer method (IDM), nudged elastic band (NEB), and following the intrinsic reaction coordinate (IRC). IDM takes a guess structure for the TS and then relaxes it along a trial unstable vibrational mode to a first-order saddle point, i.e. the TS. NEB takes the reactant and product structures and interpolates structures between the two to model the reaction pathway. IRC starts at the TS and follows the TS' unstable vibrational mode to reactant and product.

	== Improved dimer method ==		=== Improved dimer method ===
	[[File:idm_single_image.png\|250px\|thumb\|right\|A dimer consisting of two points on the potential energy surface (PES) is rotated to find the lowest curvature. The structure is updated and then a new dimer is defined. The IDM converges towards the desired saddle point, i.e. the TS.]]		[[File:idm_single_image.png\|250px\|thumb\|right\|A dimer consisting of two points on the potential energy surface (PES) is rotated to find the lowest curvature. The structure is updated and then a new dimer is defined. The IDM converges towards the desired saddle point, i.e. the TS.]]
	The dimer method{{Cite\|henkelman:jpc:1999}} is a technique for determining activated transitions without knowledge of the final state. Beginning with a trial transition state (TS) structure, a relaxed TS is obtained. In VASP, the improved dimer method (IDM) by Heyden et al. is implemented. The modification reduces the number of gradient calculations per cycle, improving algorithm performance. A detailed presentation of the method can be found in their paper{{Cite\|heyden:jpc:2005}}. The main tag is {{TAG\|IBRION}}=44.		The dimer method{{Cite\|henkelman:jpc:1999}} is a technique for determining activated transitions without knowledge of the final state. Beginning with a trial transition state (TS) structure, a relaxed TS is obtained. In VASP, the improved dimer method (IDM) by Heyden et al. is implemented. The modification reduces the number of gradient calculations per cycle, improving algorithm performance. A detailed presentation of the method can be found in their paper{{Cite\|heyden:jpc:2005}}. The main tag is {{TAG\|IBRION}}=44.
Line 19:		Line 19:
	<div style="clear:both;"></div>		<div style="clear:both;"></div>

	== Nudged elastic bands ==		=== Nudged elastic bands ===
	[[File:NEB.png\|300px\|thumb\|right\|The potential energy surface (PES) is described by a series of images (blue) connected by springs (green arrows). The initial (reactant) and final (product) states are shown in black and remain fixed. The TS is the maximum of the PES (red).]]		[[File:NEB.png\|300px\|thumb\|right\|The potential energy surface (PES) is described by a series of images (blue) connected by springs (green arrows). The initial (reactant) and final (product) states are shown in black and remain fixed. The TS is the maximum of the PES (red).]]
	The nudged elastic band (NEB) method{{cite\|mills:surf-sci:1995}}{{cite\|jonsson:book:1998}} involves the creation of an initial path connecting the system's initial and final states, utilizing a set of intermediate configurations ({{TAG\|IMAGES}}). Hence, as a prerequisite, the initial and final state (e.g., reactant and product) must be known. The ''images'' are interconnected by springs ({{TAG\|SPRING}}), forming a flexible ''band''. Through iterative adjustments (''nudging''), the positions of the images along the band are modified, minimizing energy until a minimum energy pathway, referred to as the ''nudged elastic band'', is attained.		The nudged elastic band (NEB) method{{cite\|mills:surf-sci:1995}}{{cite\|jonsson:book:1998}} involves the creation of an initial path connecting the system's initial and final states, utilizing a set of intermediate configurations ({{TAG\|IMAGES}}). Hence, as a prerequisite, the initial and final state (e.g., reactant and product) must be known. The ''images'' are interconnected by springs ({{TAG\|SPRING}}), forming a flexible ''band''. Through iterative adjustments (''nudging''), the positions of the images along the band are modified, minimizing energy until a minimum energy pathway, referred to as the ''nudged elastic band'', is attained.
Line 27:		Line 27:
	<div style="clear:both;"></div>		<div style="clear:both;"></div>

	== Intrinsic reaction coordinate ==		=== Intrinsic reaction coordinate ===
	[[File:IRC.png\|500px\|thumb\|right\|The potential energy surface (PES) is shown as a contour plot. The IRC begins at TS and follows the associated imaginary frequency <math>u_{\xi}</math> forward to the product P (red) or backward to the reactant R (blue). The energy <math>U(\chi)</math> is plotted against the complex coordinate <math>\chi</math> to illustrate the energy change as the PES is explored.]]		[[File:IRC.png\|500px\|thumb\|right\|The potential energy surface (PES) is shown as a contour plot. The IRC begins at TS and follows the associated imaginary frequency <math>u_{\xi}</math> forward to the product P (red) or backward to the reactant R (blue). The energy <math>U(\chi)</math> is plotted against the complex coordinate <math>\chi</math> to illustrate the energy change as the PES is explored.]]
	Following the intrinsic reaction coordinate (IRC) implies following the steepest descent path from the transition state to reactants and products. The IRC method employs a classical trajectory integration method with fixed velocity, i.e., the damped-velocity Verlet algorithm, incorporating an adaptive time step such that the dynamic reaction pathway resembles the IRC. Hence, as a prerequisite, the transition state must be known. The main tag is {{TAG\|IBRION}}=40.		Following the intrinsic reaction coordinate (IRC) implies following the steepest descent path from the transition state to reactants and products. The IRC method employs a classical trajectory integration method with fixed velocity, i.e., the damped-velocity Verlet algorithm, incorporating an adaptive time step such that the dynamic reaction pathway resembles the IRC. Hence, as a prerequisite, the transition state must be known. The main tag is {{TAG\|IBRION}}=40.
Line 35:		Line 35:
	<div style="clear:both;"></div>		<div style="clear:both;"></div>

	= Dynamic methods =		== Dynamic methods ==
	An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.		An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}}{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}}{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

	== Tutorials ==		== Additional resources ==
			=== Books and lectures ===
			* Tomáš Bučko has written a book, [https://stella.uniba.sk/texty/PRIF_TB_investigating_chemical_reactions_vasp.pdf Investigating chemical reactions with VASP - A practical guide], to guide in modeling chemical reactions in VASP with accompanying [https://doi.org/10.5281/zenodo.14531578 python scripts]. There are also lectures available on our YouTube channel for modeling chemical reactions (and transition states) using [https://youtu.be/mLK7CtDmw0A static approaches], [https://youtu.be/Rk0S-yvxFUo dynamic approaches], and [https://youtu.be/bzzHpTBwxbA) machine-learned force fields].

			=== Tutorials ===
	*Tutorial for [https://vasp.at/tutorials/latest/transition_states/part1/ NEB, IDM, elbow plot, and IRC calculations].		*Tutorial for [https://vasp.at/tutorials/latest/transition_states/part1/ NEB, IDM, elbow plot, and IRC calculations].
	*Tutorial for modeling reaction pathways and calculating free energies of reaction using [https://vasp.at/tutorials/latest/transition_states/part2/ static methods].		*Tutorial for modeling reaction pathways and calculating free energies of reaction using [https://vasp.at/tutorials/latest/transition_states/part2/ static methods].

Csheldon at 13:19, 28 May 2025

2025-05-28T13:19:35Z

← Older revision		Revision as of 13:19, 28 May 2025
Line 39:		Line 39:

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}}{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}}{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

			== Tutorials ==
			*Tutorial for [https://vasp.at/tutorials/latest/transition_states/part1/ NEB, IDM, elbow plot, and IRC calculations].
			*Tutorial for modeling reaction pathways and calculating free energies of reaction using [https://vasp.at/tutorials/latest/transition_states/part2/ static methods].
			*Tutorial for modeling reaction pathways and calculating free energies of reaction using [https://vasp.at/tutorials/latest/transition_states/part3/ dynamic methods].

	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]

Csheldon: /* Intrinsic reaction coordinate */

2024-11-05T12:55:18Z

Intrinsic reaction coordinate

← Older revision		Revision as of 12:55, 5 November 2024
Line 28:		Line 28:

	== Intrinsic reaction coordinate ==		== Intrinsic reaction coordinate ==
	[[File:IRC.png\|500px\|thumb\|right\|The potential energy surface (PES) is shown as a contour plot. The IRC begins at TS and follows the associated imaginary frequency <math>u_{\xi}</math> forward to the product P (red) or backward to the reactant R (blue). The energy <math>U(\chi)</math> is plotted against the ~~reaction~~ coordinate <math>\chi</math> to illustrate the energy change as the PES is explored.]]		[[File:IRC.png\|500px\|thumb\|right\|The potential energy surface (PES) is shown as a contour plot. The IRC begins at TS and follows the associated imaginary frequency <math>u_{\xi}</math> forward to the product P (red) or backward to the reactant R (blue). The energy <math>U(\chi)</math> is plotted against the complex coordinate <math>\chi</math> to illustrate the energy change as the PES is explored.]]
	Following the intrinsic reaction coordinate (IRC) implies following the steepest descent path from the transition state to reactants and products. The IRC method employs a classical trajectory integration method with fixed velocity, i.e., the damped-velocity Verlet algorithm, incorporating an adaptive time step such that the dynamic reaction pathway resembles the IRC. Hence, as a prerequisite, the transition state must be known. The main tag is {{TAG\|IBRION}}=40.		Following the intrinsic reaction coordinate (IRC) implies following the steepest descent path from the transition state to reactants and products. The IRC method employs a classical trajectory integration method with fixed velocity, i.e., the damped-velocity Verlet algorithm, incorporating an adaptive time step such that the dynamic reaction pathway resembles the IRC. Hence, as a prerequisite, the transition state must be known. The main tag is {{TAG\|IBRION}}=40.

Csheldon: /* Dynamic methods */

2024-11-05T12:50:45Z

Dynamic methods

← Older revision		Revision as of 12:50, 5 November 2024
Line 38:		Line 38:
	An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.		An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}},{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}}{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]

Csheldon: /* Dynamic methods */

2024-11-05T12:47:32Z

Dynamic methods

← Older revision		Revision as of 12:47, 5 November 2024
Line 38:		Line 38:
	An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.		An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation, or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation{{Cite\|gesvandtnerova:rucco:bucko:2021}}, or metadynamics{{Cite\|klein:2006}},{{Cite\|tiwary:parrinello:2006}}, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

	~~{{Cite\|klein:2006}}~~
	~~{{Cite\|tiwary:parrinello:2006}}~~


	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]

Csheldon at 10:28, 5 November 2024

2024-11-05T10:28:56Z

← Older revision		Revision as of 10:28, 5 November 2024
Line 38:		Line 38:
	An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.		An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using a blue moon simulation or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using constrained MD, e.g. a blue moon simulation, or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

			{{Cite\|klein:2006}}
			{{Cite\|tiwary:parrinello:2006}}


	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]

Csheldon: /* Dynamic methods */

2024-11-05T08:41:24Z

Dynamic methods

← Older revision		Revision as of 08:41, 5 November 2024
Line 42:		Line 42:
	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[:Category:Advanced molecular-dynamics sampling]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[Category:Advanced molecular-dynamics sampling]]

Csheldon at 08:40, 5 November 2024

2024-11-05T08:40:37Z

← Older revision		Revision as of 08:40, 5 November 2024
Line 38:		Line 38:
	An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.		An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.

	It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using a blue moon simulation or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular dynamics\|advanced MD]] page.		It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using a blue moon simulation or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular-dynamics sampling\|advanced MD]] page.

	== References ==		== References ==

	[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[:Category:Advanced molecular dynamics]]		[[Category:VASP\|Transition States]][[Category:Ionic minimization]][[:Category:Advanced molecular-dynamics sampling]]

Csheldon: /* Dynamic methods */

2024-11-05T08:39:53Z

Dynamic methods

@@ Line 38: / Line 38: @@
 An alternative to the static approach for modeling the TS, is to use a dynamic (or anharmonic) approach, that is to use molecular dynamics (MD). The static approach models a single structure for the reactant R, TS, and product P. An MD simulation treats each of these states as an ensemble, a combination of many possible structures.
-There is difficulty in trying to find uncommon states or rare events, such as the TS, using MD. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using a blue moon simulation or metadynamics, finding even rare events like the TS becomes feasible.
+It is difficult using MD to find uncommon states or rare events, such as the TS. An MD simulation is for a short period of time, typically only femto or picoseconds, while seconds would be required to encounter a TS. However, using a blue moon simulation or metadynamics, finding even rare events like the TS becomes feasible. More detail for each of these methods, as well as other advanced MD approaches can be found on the relevant [[:Category:Advanced molecular dynamics|advanced MD]] page.
 == References ==
-[[Category:VASP|Transition States]][[Category:Ionic minimization]]
+[[Category:VASP|Transition States]][[Category:Ionic minimization]][[:Category:Advanced molecular dynamics]]

Csheldon: /* Metadynamics */

2024-10-30T16:05:56Z

Metadynamics

← Older revision		Revision as of 16:05, 30 October 2024
Line 50:		Line 50:

	=== Metadynamics ===		=== Metadynamics ===
	[[File:Metadynamics.png\|300px\|thumb\|right\|Metadynamics can be used to model rare events, such as a transition state (TS). The structure begins in the left-hand potential well. Over time, Gaussian potentials are added to the well and it gradually "fills" up, cf. lines 1-2. The energetic barrier over the saddle point (TS) decreases until in can be crossed, cf. line 3, at which point the well on the left begins to fill, cf. lines 4-7. Finally, both wells are filled and the metadynamics simulation finishes, cf. line 8.]]		[[File:Metadynamics.png\|300px\|thumb\|right\|A plot of potential V against position x. Metadynamics can be used to model rare events, such as a transition state (TS). The structure begins in the left-hand potential well. Over time, Gaussian potentials are added to the well and it gradually "fills" up, cf. lines 1-2. The energetic barrier over the saddle point (TS) decreases until in can be crossed, cf. line 3, at which point the well on the left begins to fill, cf. lines 4-7. Finally, both wells are filled and the metadynamics simulation finishes, cf. line 8.]]
	A specific biased potential, is [[Metadynamics\|metadynamics]], where a series of Gaussian potentials are applied periodically over the course of the simulation. Fictitious particles "roll" in the bottom of the potential energy well{{Cite\|klein:2006}}. After a time interval, a small Gaussian-shaped repulsive potential is added to the positions of the particles. Over time, these additional potential fills up the potential wells allow the fictitious particles to escape from the first well via a saddle point to a nearby local minima. By finding the saddle point, the transition state is also modeled. The resulting time-dependent biased potential is, at infinite time, related to the free energy{{Cite\|tiwary:parrinello:2006}}.		A specific biased potential, is [[Metadynamics\|metadynamics]], where a series of Gaussian potentials are applied periodically over the course of the simulation. Fictitious particles "roll" in the bottom of the potential energy well{{Cite\|klein:2006}}. After a time interval, a small Gaussian-shaped repulsive potential is added to the positions of the particles. Over time, these additional potential fills up the potential wells allow the fictitious particles to escape from the first well via a saddle point to a nearby local minima. By finding the saddle point, the transition state is also modeled. The resulting time-dependent biased potential is, at infinite time, related to the free energy{{Cite\|tiwary:parrinello:2006}}.