Ab-initio molecular dynamics (= AIMD ) is too time-consuming and useless.

Home page
Molecular dynamics is useless

Ab-initio (or Car-Parrinello ) molecular dynamics (= AIMD, CPMD ) treating multiple electrons as one pseudo-electron DFT with pseudopotential is useless, unable to predict anything.

(Fig.1) Ab-initio MD just changes coefficients (= c ) of artificially-chosen fake electron's wavefunction lacking real atomic shape.

Ab-initio molecular dynamics (= MD ) is useless, unable to predict anything, more time-consuming than ordinary impractical MD.

Due to unphysical quantum mechanical wavefunction lacking real atomic shape, today's only molecular simulating method called ( classical ) molecular dynamics (= MD ) is impractical, too time-consuming to calculate each atomic position at short-time intervals many, many times, based on artificially-chosen force field potential.

Ab-initio or first-principle molecular dynamics (= AIMD, FPMD, CPMD ) changing fictitious electron's wavefunction based on DFT pseudo-potential is much more impractical and more time-consuming than the impractical classical MD ( this-p.2-left-1st-paragraph,  this-p.2-I-introduction,  thisp.19-3rd-paragraph ).

In Ab-initio MD, quantum mechanics has to treat the whole atoms as one pseudo-electron DFT (= Kohn-Sham ) model through artificially-chosen exchange pseudo-potential (= functional ) that is fake ab-initio (= empirical ), unable to predict any physical values ( this-p.2-2nd-paragraph,  this-p.2-left-last-paragraph~p.2-right,  this-p.23-lower ).

Ab-initio MD takes too much time in DFT calculating coefficients of artificially-chosen fictitious electron wavefunctions.

In this impractical ab-initio MD, only nuclei (= positive ions ) are treated like classical particles, moved gradually by differentiating the total fake energy E with respect to each nuclear coordinate R_I (= force = space derivative of total potential energy, this-p.6 ) at short-time intervals.

In Born-Oppenheimer (ab-initio) molecular dynamics (= BOMD ), they have to artificially choose fake one-pseudo-electron DFT wavefunction ( this-p.14-(40) ), and update their coefficients (= c ) by the extremely time-consuming self-consistent field (= SCF ) method at short-time intervals ( this-p.23,p.27,p.32 ). which is completely impractical ( this-p.35,  this-p.2 ).

Ab-initio Car-Parrinello MD relies on fictitious electron's mass, which is also too time-consuming, impractical.

So ab-initio MD often uses the slightly simpler Car-Parrinello molecular dynamics (= CPMD ) using fictitious electron's mass μ ( this-p.36,  this-p.9,  this-p.7,22 )

↑ In this CPMD, they update each coefficient (= c ) of the artificially-chosen pseudo-electron's wavefunction (= φ = basis set ) by differentiating the total energy E with respect to the coefficient c ( this-p.41,p.44,  this-p.12-(6.10),  this-p.5,p.7 ) at extremely short-time intervals (= 0.1 fs, this-p.12-22,  this-p.3,p.12 ) together with updating nuclear positions.

So ab-initio molecular dynamics just changes fictitious electron (fake) wavefunction's coefficients instead of moving real electrons.

And this ab-initio MD (= CPMD, BOMD.., old Ehrenfest MD is not used now, this-13-upper ) needing calculations of both nuclear and fictitious electron's states at much shorter time intervals (= 0.1fs ) is more impractical and more time-consuming than classical MD updating only atoms at the intervals of 2fs ( this-p.19-lower,  this-p.2 ).

Quantum mechanical mainstream density functional theory (= DFT ) treats the whole material as one pseudo-electron model with artificially-chosen spin pseudo-potential parameters.  ← No quantum mechanical prediction, so No evidence of spin

(Fig.S)  Quantum mechanical DFT just artificially choosing (fictitious) spin parameters cannot predict ferromagnet or antiferromagnet.  → No evidence of spin

Electron spin is unreal, irrelevant to (anti-)ferromagnet.

Ferromagnet or antiferromagnet is said to be caused by (fictitious) electron spin according to quantum mechanics.

But an electron is known to be Not actually spinning.
Furthermore, quantum mechanics just choosing free spin parameter J has No ability to predict ferromagnet or antiferromagnet.

Quantum mechanical Schrodinger equations are unsolvable and unable to predict any multi-electron atoms or molecules.

Quantum mechanics has to rely on the rough approximation called density functional theory (= DFT ) that tries to treat the whole material as one-pseudo-electron or quasiparticle model with artificially-chosen pseudo-potential (= different exchange-correlation functionals give different spin interaction parameter J as shown in this p.7-Table.III ).

Quantum mechanics just choosing free parameters cannot predict ferromagnet or antiferromagnet.

In this mainstream DFT, physicists have to choose various freely-adjustable parameters such as fictitious wavefunctions, pseudo-potential (= U, J,  this p.6-right-1st-paragraph ), (spin) magnetic moment (= m,   this p.4~p.7,   this p.136(p.128)-2nd~3rd-paragraphs ), and numbers of electrons (= n ) in different spin states ( this p.16-1st-paragraph ).  ← No quantum mechanical prediction

This p.11-3rd-paragraph says
"In practice treated as fitting parameters, i.e., adjusted to reach agreement with experiment: equilibrium volume, magnetic moment, band gap"

↑ Instead of (useless) quantum mechanical prediction, these artificially-chosen magnetic moment parameter (= m ), electron spin number (= n  or spin polarization parameter ζ ) determine whether the material is ferromagnet or antiferromagnet ( this p.53(p.45),p.60-63,  this p.8-p.12. p.21(or p.20) ).

DFT pseudo-potential parameters (= U, J ) must be chosen from experimental results in most cases (= instead of being predicted by quantum mechanics,  this p.4-right-3rd-last-paragraph,  this p.2-left-2nd-last-paragraph,  this p.3-1,  this p.3,  this p.22 chose t or J,  this p.6-right-1st-paragraph ).

This p.5-right says
"However, apart from the intentionally fitted LSDA+U result, all functionals overestimate the antiferromagnetic coupling J by at least a factor of two, and perform even worse for the interchain coupling J
 ( this p.16,  this p.8-left-1st-paragraph )"

So quantum mechanics (= unphysical spin model ) relying on various freely-chosen parameters ( this p.10,  this p.21 ) cannot predict whether the material is ferromagnetic or antiferromagnetic ( this p.22(or p.21)-2nd-paragraph ), which means there is No evidence that (anti-)ferromagnetism is caused by electron spin.

Stable ferromagnet is caused by the realistic electron's orbits (= instead of spin ) meshed with each other by Coulomb electric interaction.

Feel free to link to this site.