Gaussian Processes - Science topic

I optimized the structure of Pd Acetate Trimer using HF/STO-3G and the B3LYP/LANL2DZ without any solvent. The optimization was successful without any negative frequency.

Then I added solvent IEFPCM/DCE to the system. This causes the optimization to show error link 9999. How Do I fix this problem ?

I have tried using the last optimized geometry and used that for next calculation but every time its showing same error.

I am adding the drive links to the files herewith:

It has the input geometries and log files for both without solvent and with solvent calculations.

Can anyone please help ?

n is time. and i want to know if the distribution of small scale fading is correct or not.

Three recent breakthroughs due to AI in arts and science serve as motivation: An award winning digital image, protein folding, fast matrix multiplication. Many recent developments in artificial neural networks, particularly deep learning (DL), applied and relevant to computational mechanics (solid, fluids, finite-element technology) are reviewed in detail. Both hybrid and pure machine learning (ML) methods are discussed. Hybrid methods combine traditional PDE discretizations with ML methods either (1) to help model complex nonlinear constitutive relations, (2) to nonlinearly reduce the model order for efficient simulation (turbulence), or (3) to accelerate the simulation by predicting certain components in the traditional integration methods. Here, methods (1) and (2) relied on Long-Short-Term Memory (LSTM) architecture, with method (3) relying on convolutional neural networks.. Pure ML methods to solve (nonlinear) PDEs are represented by Physics-Informed Neural network (PINN) methods, which could be combined with attention mechanism to address discontinuous solutions. Both LSTM and attention architectures, together with modern and generalized classic optimizers to include stochasticity for DL networks, are extensively reviewed. Kernel machines, including Gaussian processes, are provided to sufficient depth for more advanced works such as shallow networks with infinite width. Not only addressing experts, readers are assumed familiar with computational mechanics, but not with DL, whose concepts and applications are built up from the basics, aiming at bringing first-time learners quickly to the forefront of research. History and limitations of AI are recounted and discussed, with particular attention at pointing out misstatements or misconceptions of the classics, even in well-known references. Positioning and pointing control of a large-deformable beam is given as an example.

Hello, I'm a very new user of Gaussian 16. So, I started doing optimization and frequency of MIL-53(Fe) [construct with 3 atoms Fe and 4 bdc linker] cluster model using MN15L/genecp. The calculation was finished and I got inp.log. How do I know is the optimization and frequency calculation was success? And how do I get these data: (1) The optimized structure, (2) Thermodynamics (like enthalpy, Gibbs free energy, etc).

Please, advise me.

Thank you.

R. S.

This is my input

%mem=32Gb

%nproc=12

#p MN15L/genecp opt freq=NoRaman

test

-1 2

.....

****

O 0

6-31+G(d)

****

C H 0

6-31G(d)

****

Fe (lanl2ldz)

****

FE

FE-ECP (lanl2dz)

****

Note: Actually I wanna calculate the energy adsorption of CO2 using the energy that I got after opt and freq calculation with this formula: Eads= Esystem - (EMOFs + ECO2).

Can we do the computational studies of ZnO Nanoparticles using Gaussian Software? If yes, Which structure of ZnO can we use to compare Theoretical work with experimental work?

#computationalstudies #ZnO #ZnOnanoparticles #nanoparticles #Gaussiansoftware #ZnOstructure #theoreticalstudy

The code should be well documented with examples.

Hello everyone,

I am currently trying to run an IRC calculation for a RO2 structure isomerizing to the QOOH structure for the Ethanal +OH system. The RO2 and QOOH energies are agreeable with literature and the TS that was found seems to have the right imaginary frequency and movement, and it has an energy that is closer to the literature values. But whenever I run an IRC calculation, I see on GaussView that the hydrogen abstraction stops halfway and then goes back to the RO2 without ever forming a bond for the QOOH. I have tried reducing the step size, increasing the maxpoints, and even trying LQA. I have also tried doing separate IRC calculations for the forward and reverse reactions, to no fruition. Is this an indication that the TS is still wrong, or maybe the QOOH structure is incorrect? Or, is there a keyword in the route section that I am missing?

Please advise, thank you.

How to find out predictors' contribution in Gaussian process regression (i.e., Predictor Importance by Permutation)? Is there any example for predictors' contribution in Gaussian process regression using the sklearn package in python?

Kindly if there is an applied example or an algorithm that explains the steps of this method

I am working with electricity time-series data collected at 15 minutes intervals. I am looking for a procedure/theory to find the pattern/sequence in the time-series data based on given features. As I am working with electricity time-series data and solving the problem of solar PV identification from these data, the given features would be:

See the attached figure to understand the above two features.

I have gone through the literature for the same. I got the following:

I am not looking to solve this problem with option 3. Can anyone guide me with the option 1 as it looks more relevant to my problem. I cannot understand how Gaussian prior can fit into the problem. Summarily, I want to utilize above two features as the prior knowledge and use the given data(electricity time-series) as the evidence to prove that the solar PV panel is present or not.

I want to optimize a structure with two different material of Phosphomolybdic acid (PMA) and 4,4′-bipyridine, but I can't run out with a chk point file.

Error message always show convergence failure error termination via I502 and no lower energy was found error termination via I508

I had tried with all possible changes such as adding instruction like SCF=QC, YQC and switch to a lower basis set like HF/STO-3G and DFT-B3LYP/STO-3G, but after runing about 20 days it still didn't work.

My version is gaussian16 32 bites and gaussianview06 32 bites.

Below is my .gif input file and calculation command # opt hf/sto-3g scf=qc %mem=2000 MB

I'm doing a reserch about heat and mass trasfer so, I tried to do the same research that has been by someone.

The research about vortex generator in a blocked channel for heat transfer enhancement.

The resercher used ((2D Desgin, using the hybrid genetic algorithm (GA) combining with Gaussian Process.

in Gaussian method are obtained by solving energy equation followed by the flow field using Navier–Stokes solver.

The flow is considered to be two dimensional, steady state, laminar and incompressible with constant fluid properties. Body forces and viscous dissipation are ignored. Also there is a negligible radiation heat transfer since there is no considerable temperature gradient.

The thermal properties of the fluid are assumed to be constant, and the buoyancy effects are negligible.))

I tried the same thing but the result was difference. I think I made a mistake in somewhere but I could not find it.

need some help.

I am trying to implement the Gaussian process-based model predictive control, where the Gaussian process is used to model the unmodelled dynamics of the system. I understood the theory behind it but faced difficulty in the implementation, basically in optimising the cost function in code. Please share any reading material or tutorial code with me.

I optimized a transition state using tight convergence criteria and all the four convergence criteria are yes; but when i run the IRC it stops after 3 or four steps ?

1-in the first attempt i added int=ultrafine

2-in the second attempt i added: LQA word, increased the stepsize from 5 to 10, and added recalc=5

Can someone help me to solve this problem?

I would like to check the possibility of implementation of BO using C language.

The implementation target is MCU like Tricore or further.

Is this possible?

Moreover, is there anyone know where the C-implementation sources is present at github or other sites?

Finally, what would be the most difficult parts while I try to implement BO to MCU starting from matlab toolbox or python code (@github) ?

I know that BO uses Gaussian Process Regression in order to estimate surrogate function from experiment data. Is this the most difficult part for implementation of BO?

Thanks in advance.

i restarted an opt +Freq job with the keywords opt=restart, pop=nboread... but the output file do not contain any NBO section as it should ( i an need the natural population analysis) ! can i get it from the chk file? is there any othar to avoid re run the process

We know that LU decomposition is an important method for stochastic simulation of 2D RF. Assume the covariance matrix of regionalized RVs is C, it can be decomposed as C=LU according to the LU algorithm, then a RF can be generated by X=L'*y, where y is a vector consisting of independent standard normal random numbers. I want to know whether can I use LU decomposition for simulation if n observations exist as the conditioning data? If so, how can it be demonstrated?

While reading a research paper I have found that to find the abnormality in a signal authors were using a sliding window and in each window he was dividing the mean^2 by variance. After searching in the internet I found a term called fano factor but it was inverse of that. Could anyone please give an intuitive idea behind

the equation mean^2/variance?

Many references are explained that kriging and GP regression is the same. However, in the formulation, they have differences from each other. Can you explain the difference between these two methods from the statistics domain?

Actually need to compare for different kriging methods, for example, simple kriging assumed that the mean of the random function is known and constant as GP regression. However ordinary kriging assumed the mean of the function is known and changing locally (E(x)).

I want to perform a checkerboard resolution test of my tomographic inversion. What kind of function do people use to make the checkerboard model of low/high anomalies? Bell-shaped (gaussian) anomalies?

Hi, I am doing an optimisation job for the molecule, but after a long wait it comes out with operation on file out of range. The job runs just alright for another molecule in the same homologous series. Anyone knows what is going on?

Dears

I am doing QM/MM calculations and trying to optimize a Gaussian input file of about 35000 atoms. QM region contains 60 atoms. But after a long time, I got an error that Max no. of micro-iterations restarts exceeded. Can anyone help me out?

Format of input file;

%chk=/home/SKD/work/gaussian/calculations/g23.chk

%mem=32GB

%nproc=16

#n oniom(hf/6-31g:UFF) opt=ReadFreeze scf=(xqc,maxcycle=1000)

That's a graph for Intensity-2Theta .

Hello everyone,

I want to perform some gaussian calculations and want to use Materials studio for my research work. Can anyone guide me how to get access to supercomputer resources in India for these jobs? I am ready to pay charges for the usage.

Thank You.

Greetings,

I have a bit of problem in setup the basis set of my DFT study for ZnO.. As you can see in the image attached below, I use B3LYP 6-31G (d,p) for geometrical opt. of ZnO.. But sadly, I still cannot get the energy gap of 3.4 eV for ZnO... At best I can get is 3.6 eV when using B3LYP 6-311G... Can someone make a suggestion on how to improve my calculation to get an accurate energy gap of ZnO using B3LYP.... Thank you !!

I am working with the algorithm Gaussian Process Regression. I used gaussian process regression module from scikit learn library of python. But I couldn't correlate the hyperparameters of Gaussian Process Regression(GPR) implemented in sklearn with that in actual mathematical equation. So, if anyone can give any suggestion about this it will be very helpful for me. In addition to that, I am also trying to optimise the hyperparameters of GPR. If anyone can do any favour for this specially provide any codes for this I will be really grateful. Any suggestions are highly appreciated. Thanks in advance.

I have done docking a ligand molecule with a receptor. After the docking, I separated the ligand from the receptor. Now I need to add partial charges to the ligand without minimizing its energy.

Dear all,

I've read some literatures that say Kriging works only for single output(scalar output) .But the DACE toolbox handbook mentions that it can handle multiple response case(P15, chapter5.1).So I am puzzled . If the output is high-dimensional (hundreds of thousands), how can i deal with it? (i.e.the output of an input sample is a vector, or a curve).Especially when combined with adaptive filling criteria ,like EGO etc. ?

Many thanks!

Dear all. The Normal distribution (or Gaussian) is mostly used in statistics, natural science and engineering. Their importance is linked with the Central Limit Theorem. Is there any ideas how to predict the numbers and parameters of thos Gaussians ? Or any efficient deterministic tool to decompose Gaussian to a finite sum of Gaussian basic functions with parameter estimations ? Thank you in advance.

Dear all. The Gaussian function is mostly used in statistics, physics and engineering. Some examples include:

1. Gaussian functions are the Green's function for the homogeneous and isotropic

2. The convolution of a function with a Gaussian is also known as a Weierstrass transform

3. A Gaussian function is the wave function of the ground state of the quantum harmonic oscillator

3. The atomic and molecular orbitals used in computational chemistry can be linear combinations of Gaussian functions called Gaussian orbitals

4. Gaussian functions are also associated with the vacuum state in quantum field theory

5. Gaussian beams are used in optical systems, microwave systems and lasers

6. Gaussian functions are used to define some types activation function of artificial neural networks

7. Simple cell response in primary visual cortex has a Gaussian function modeled by a sine wave

8. Fourier transforms of Gaussian is a Gaussian

10. Easy and efficient approximation for signals analysis and fitting (Gaussian process, gaussian mixture model, kalman estimator , ...)

11. Discribe the Shape of the UV−Visible Absorption

12. Used in Time-frequency analysis (Gabor Transform)

13. Central Limit Theorem (CLT) : Sum of independent random variables tends toward a normal distribution

14. The Gaussian function serves well in molecular physics, where the number of particles is closed to the Avogadro number NA = 6.02214076×1023 mol−1 ( NA is defined as the number of particles that are contained in one mole)

15. ...

Why Gaussian is everywhere ?

I would like to use CLT (Central limit theory) to substitute the linear combination of sincs with random coefficients C_k drawn according to chosen constellation with a complex Gaussian process with the same mean and covariance.

What is the way to do it?

What's the difference between a multivariant Gaussian and a Gaussian mixture model? Do they refer to the same thing?

I'm dealing with a supervised problem which includes three features. My main purpose is to find the relationship among these features which is complicated. I'm looking for a straight relationship, not a multistage one. Support Vector Regression and Gaussian Process Regression include multiple stages but they are not proper for me. Which algorithm do you recommend to me?

I have read multiple articles that have used machine learning algorithms (convolutional neural network, random forest, support vector regression, and gaussian process regression) on cross-sectional MRI data. I am wondering whether it is possible to apply these same methods to longitudinal or clustered data with repeated measures? If so, is there an algorithm that might be better to use?

I would be interested in seeing how adding longitudinal data could improve the performance of these types of machine learning models. So far, I am only aware of using mixed effect-models or generalized estimating equation on longitudinal data, but I am reading books and papers to learn more. Any advice or resources would be greatly appreciated.

Additionally, can someone comment on the following:

I am generating a 1D data using a squared exponential kernel. If I use the data to learn the hyperparameters using maximum likelihood approach, then what are the conditions under which I will get the same hyperparameters as I have used for generating the data.

Hey everyone,

currently, I am working with Gaussian process latent variable based models. In literature, the model likelihood is discussed for model selection.

Unfortunately, this does not work for my application. Currently I am using the log likelihood and the reconstruction error. The model likelihood increases and the reconstruction error decreases with increasing dimension/inducing points. BIC doesn't make sense in this context (and behave similar...).

Are there better parameters for model selection?

All the best,

Will

Hello everyone

I have successfully drawn CuO structure in Gauss view through PBC, I need to know what are the appropriate parameters in functional and basis set in order to successfully optimize the crystalline structure. Please refer to some literature, websites and forums regarding this

thank you

what are the new upgrades in Kriging methods ? I am thinking about variable distributions for the Kriging predictor? and I want to know that is it rational and could it be a proper upgrade to the method for learning/predicting applications?

I'm using fitrgp toolbox for Gaussian process Machine learning on MATLAB. But till now all my work has been in real numbers. Now I need to include complex numbers in my project. But fitrgp does not accept complex training data. Is there any way to get around this situation, like a substitute toolbox or some way to include complex numbers in the same toolbox?

I am trying to simulate the molten metal flow of TIG welding in ANSYS Fluent by giving 30 iterations per time step. I observed that initially at each time step, convergence occurred at around 16th or 17th iteration. But when the simulation progresses, the convergence shifted towards larger iteration (like 20th, 25th, 30th..) and finally started diverging after reaching half of the total time steps.

What are the possible reasons of this and how can I keep the simulation running with convergence at around same iteration?

Specifically, TA spectra in solution.

Consider a single mode Gaussian beam entering a 0.75 NA objective. I made measurements of beam's FWHM around the focus. It looked like before the focus, the beam converges according to NA=0.75. But after the focus, NA is somewhat larger - I saw this in 3 attempts. Is this normal? NA=0.75 is not associated with a single mode beam usually.

I am doing DFT calculation on metal complex. I got this error: "Bad data into FinFrg.

Error termination via Lnk1e in /usr/local/g09l/g09/l122.exe at Wed Nov 20 00:00:58 2013.

Job cpu time: 0 days 0 hours 1 minutes 0.9 seconds."

Please suggest some solutions.

I have recently been working on applying ML algorithms for time series prediction (forecasting demand of item). I am a bit confused somehow. So, my questions is as follows

1.Does the performance of these algorithms usually depends of the nature of the dataset i.e descriptive characteristics like mean, max, min etc. The machine learning algorithms i applied includes, Gaussian processes, backpropagation neural network and support vector machine.

2.I found Gaussian processes performs better than the rest but i am not too convinced with my results since the RMSE is still high for all algorithms i tried. Below is a summary for data i have.

item_1

Min = 100

Max = 40200

Mean = 11849

RMSE = 11756.5

I used WEKA forecasting package i have attached a time series plot

Dear Altruistic

I am new with Gaussian, I have optimized a Graphene structure by using Gaussian with DFT method and 6-31G basis set. But after optimization the bond length between carbons have changed from original  one and in the optimize structure different carbon carbon bond length also different.

could any one help me to resolve this ? 

I have read about What Bhat have done about this MDCEV model but I am eager to know more about this.

Hi there, I could not find a transition state for some reason. I have provided my input and output file. For some reason the output did not have an imaginary frequency but seemed to progress like a reaction.

Please not that the OH radical is the reason why I have an 0 2 charge and multiplicity.

Can someone explain why using Gaussian function for fitting the photoluminescence spectrum, we must change the unit of wavelength (nm) to unit of energy (eV)? I see someone doesn't change unit of wavelength (nm) in to unit of energy (eV). I wonder about their method?

Thank you so much for your answer.

I have one drug candidate in diastereomeric mixture. Is it possible to identify the number of isomers in diastereomeric mixture by using the second derivative calculations and Gaussian curve fitting analysis of DTG (Differential Thermogravimetric analysis) curve by using OriginPro 8 software?

What are some advantages of using Gaussian Process Models vs Neural Networks?

An example explain that's advantages.

what are the advantage of Gaussian process regression in hydrology compare to other data driven techniques,

  

Hi,

I have a big dataset with mat format (used for Matlab). It is a giant matrix. Its size is about 280 MB. I am trying to input it into R using the function 'readMat' in package 'R.matlab'. However, it took soooo long time (about 20 minutes) and then it threw out an error 'cannot allocate vector of size 10.7Gb'. Though it can be easily inputed in Matlab (about 1 minute).

Does anyone know how to input a large mat dataset into R quickly? And why the memory stuff is OK in Matlab but not in R?

Thanks,

Lei

I am trying to use "volutil" to combine ils outputs (.dx occupancy maps). The error I got is "No weight detected in DX file."

Any ideas?  

For a relative large event, it usually has several sub-events and we get a large set of data. In order to do stacking or for visualization purpose, it's better to flip the traces with opposite polarity. Could you tell me some simple and effective method to detect it. Some crude method like finding the maximum on displacement record usually failed if the waveform is complex. Thanks!

Hello, I am very new user of Gaussian. I started the optimization of the connected two amino acids. Calculation was finished, and I would like to get the optimized structure. How can I know success of optimization? and get the optimized structure?

Please, advice me.

Thank you.

B.M.

I would like to simulate ambient vibration in a numerical model such as a MDOF shear building. In some cases, the use of random Gaussian white noise can be a solution. Is it a plausible approach?

Thank you

Hi,

I am trying to fit Gaussian process to learn a distribution from input sequence to output sequence. I am relatively new and have several questions:

1) For time series prediction problems, if you have a sequential input you can feed them without stacking inputs into a vector while using RNNs or LSTMs. Do Gaussian processes have such a structure? If no, is it still make sense to use GPs to learn output sequence from input sequence?

2) When I try to fit GP to training data with 400 samples and 18 features it has really difficult time. Is there a way to batch the data and use the learned parameters from one batch to another as in neural networks.

Thanks in advance.

If the amplitude distribution is given how can we calculate the total energy? 

I am using mixed z-matrix and Cartesian coordinates in ONIOM with Gaussian 16 (z-matrix for high layer and Cartesian for low), and the freeze code for the Cartesian atoms is not working. Do I need to specify something in the route card to direct the program to adhere to this coordinate specification? I have no constants in my internal coordinates, but I need to hold one atom fixed in Cartesian coordinates.

Thanks in advance.

I have a little problem with small frequencies (<50-100 cm-1) in optimization jobs when using Gaussian. Obviously the problem is when comparing free energies. In a recent paper I found this line: "Truhlar’s quasiharmonic correction was applied

in order to reduce error in estimation of entropies arising from the

treatment of low frequency vibrational modes as harmonic oscillations

by setting all frequencies less than 100 cm-1 to 100 cm-1".

Does anyone know how to do this and which commands are required in Gaussian. Otherwise if anyone as an other option to remove small frequencies it would be very welcome.

Thanks!

Hi,

I need to calculate gradient of energy at a given geometry without doing the full optimization. In GAMESS-US, one can set a "gradient" type of calculation for that. In G09, I usually do Opt (MaxCycles =1) and. The G09 does the gradient calculation but naturally produces .chk and .rwf files complaining about non optimized structure.

Is there is a way to do it more conveniently? Sure, I could just delete all unnecessary files after the calculation is finished but may be you have better ideas...

I want to perform calculation in solvent 2,2,2-Trifluoroethanol in gaussian but its not defined in gaussian. Please help me out how should I proceed for the same.

Hello, I am trying to running a frequency calculation with gaussian for a dimer that is made by benzene and bromine fluoride at T-shape. After finding a stationary point in the optimization, the software starts the frequency calculation, but then the latter one aborts without any error message... The last line in the output file is:

Using OV2 memory method for fx*t*t/D, MaxI= 43 DoOO2=F NP= 8.

Attached there are the input and the output files.

Anyone can help me, please?

I want to calculate with different temperature on gaussian 09 and I do not know how to do it

Help me please

What is the best method for approximating a Gaussian mixture as a single Gaussian in the sense of accuracy? Also it can be used in onlinevapplications.

I have passed the white Gaussian noise from a low pass filter in labview and resulting it should be colored noise. I get the signal and I saved random number in Excel file. Could someone suggest that how can I make sure that the random number saved in Excel file is actual colored noise.

Any help/suggestion would be welcome.

Consider the following expression x^HAx, where x is a known vector, i.e. a determinisitc parameter, and x^H its hermitian, and A is a hermitian, positive semi-definite matrix constructed as the summation of D rank-one matrices of the form d_id_i^H for i=1,...,D. In this case, the vectors d_i are distributed as a standard complex Gaussian (zero mean and unit variance).

The question is which is the statistical model of x^HAx?

Thanks in advance.

I am looking for a derivation of conditional simulation for Kriging also known as Gaussian Process Regression. Conditional simulation allows to simulation realization based on the modelled Gaussian process conditioned to the given data.

Can someone help?

I have read many of papers in which they have carried out calculation of potential energy scan for molecule in singlet and triplet states and have investigated the crossing point of singlet and triplet potential energy diagrams. Are these rigid or relaxed scans?

I want to investigated the potential energy diagram for HXeXeF molecule in singlet state and triplet state (at the singlet state geometry) along the H-Xe coordinate to obtain crossing point of them. I fixed distances of Xe-Xe and Xe-F and used scan keyword in rout section.Is it right?

I attached gaussian output files.

The Fourier transform of the Guassian signal is again a Guassian signal. Is it from the duality property of the Fourier transform?

I want to compare between blending data using Regression Kriging and Bayesia Kriging. What is the advantage of Bayesian Kriging compare to Regression Kriging? Anyone has a recommended link/journal for learning Bayesian Kriging? Many thanks in advance

Gaussian09D: After an excited state computation has finished, the requested excited states are listed along with the atomic orbital transitions that are supposedly taking place during the excitation. How does one determine what type of transition is taking place?...in other words, is/are the transition/s pi-pi*, sigma-sigma*? How do you determine which one it is? I am using the following scheme: td=(50-50,nstates=20) rb3lyp/6-311g(d,p). Thank you for your time! :)

The space grid dimension in FDTD depends on the frequency we choose to operate on. When considering sine wave excitation, we can find space grid value from the frequency chosen. But, how to find the space grid value when the excitation is taken as guassian pulse. the excitation is given for a microstrip line with width(along x axis)=2mm, length of the line(along y axis)=20mm and height of the substrate(along z axis)=0.8mm. with these specifications, i had found dx, dy and dz as one tenth of the values and the time step as (least dimension)/(2*speed of light). but the field values extend upto infinity after a few time steps. can anybody tell were have i gone wrong.

thanks.

I have some multivariate Gaussian random variables which I want to whiten them. During decorrelation procedure the means of variables change. Is there any way for whitening the correlated variable that does not change the mean?