Pmf-integrate

Navigation: Documentation / Utilities / pmf-integrate

---

pmf-integrate: version 6.1090.a7cf688 (2026-06-18 21:15)

Name

pmf-integrate

Synopsis

pmf-integrate [OPTIONS] [--] ACCU ENE

Description

The program numerically integrates data from the PMF calculations. The integration is performed either by the reverse finite difference (RFD) method, radial basis functions (RBF), or Gaussian process (GPR).

Arguments

ACCU Name of the file containing the input PMF accumulator.

ENE Name of file where the resulting energy surface will be printed. If the name is '-' then the output will be written to the standard output.

Options

-r, --realm NAME Requested realm for the integration. The list of supported realms can be obtained by --listrealms. (Default: dA/dx)

--listrealms List supported realms.

-m, --method NAME Integration method. Supported methods are: rfd (reverse finite differences via csparse), rbf (radial basis functions), and gpr (Gaussian process). (Default: rfd)

--emethod NAME Integration method for the energy cut-off. Supported methods are: rfd (reverse finite differences via csparse), rbf (radial basis functions), and gpr (Gaussian process). (Default: rfd)

-a, --lmethod NAME Linear algebra method for LLS solution or matrix inversion. Supported algorithms are: default, svd (SVD - singular value decomposition, divide-and-conquer driver), svd2 (SVD - singular value decomposition, simple driver), qr (QR factorization), lu (LU factorization), ll (LL - Cholesky factorization). Possible combinations are: RFD(LU, default), RBF(QR, SVD, default), and GPR(LU, SVD, SVD2, LL, default). (Default: default)

--rcond NUMBER RBF+GPR: Rank condition for SVD. The selected value must be carefully tested. Calculation at computer precision is requested with -1 (not recommended). (Default: 1e-06)

-l, --limit NUMBER Only bins containing more samples than NUMBER are considered properly sampled. (Default: 1000)

--skipfftest Skip flood-fill test for discontinuous regions.

-q, --energylimit NUMBER Integrate data only if the free energy is below NUMBER. This limit enforces two integration runs. A negative value disables the limit. (Default: -1)

--rmnegene Remove regions with negative free energy.

--skiplastelimit Skip energy limit filtering after the final integration.

--mfmaxzscore NUMBER RBF+GPR: Reject mean forces whose prediction errors have a z-score above NUMBER. It is assumed that mean force errors have zero mean and follow a normal distribution. This limit is applied in each pass. A negative value disables the limit. (Default: -1)

--mfnumofztests NUMBER RBF+GPR: Repeat the z-score test for mean force errors NUMBER times. (Default: 1)

-s, --sigmaf2 NUMBER GPR: Variance of the reconstructed free energy surface (signal variance). (Default: 15.0)

-w, --wfac SPEC RBF+GPR: Factors influencing widths of RBFs or squared exponential kernels. The width is the distance between adjacent squared exponential functions, multiplied by these factors, in the form WFac1[xWFac2x...]. The last value pads the rest. (Default: 3.0)

-n, --sigman2 SPEC GPR: Values of noise sigma squared for each CV, in the form SigmaN2(1)[xSigmaN2(2)x...]. The last value pads the rest. (Default: 0.0)

--loadhyprms NAME GPR: Name of file containing GPR hyperparameters.

-t, --rfac SPEC RBF: Reduction factor for the number of RBFs. The number of RBFs in a given direction is the number of bins in that direction divided by this factor, in the form RFac1[xRFac2x...]. The last value pads the rest. (Default: 1.0)

-g, --overhang NUMBER RBFs overhang to properly integrate areas near sampled edges. Ignored for periodic CVs. (Default: 2)

--includeglued RBF+GPR: Explicitly include glued regions. This option is set ON when --glueing > 0.

--glueing NUMBER RBF+GPR: Calculate energy also for unsampled bins in close vicinity to sampled ones. (Default: 0)

--glueholes RBF+GPR: Calculate energy also for unsampled regions inside the FES.

-p, --periodic RFD: Enable periodicity for collective variables that are periodic.

--globalmin SPEC RFD+RBF+GPR: Position of the global minimum provided as a single string in the form CV1xCV2x...xCVn (relevant for error determination). If not set, the position is determined automatically.

-o, --offset NUMBER Specify an integration constant. (Default: 0)

-e, --witherror GPR: Estimate free energy errors.

--noenergy GPR: Skip calculation of energy and errors (can save time when only logML is required).

--saveaccu NAME Save the final PMF accumulator to the file NAME.

--mfinfo NAME RBF+GPR: Name of file containing input and predicted mean forces.

--output FORMAT Output FORMAT to print the free energy surface. Supported formats are: plain, gnuplot. (Default: gnuplot)

--printall NAME Print results for all bins, even if they are not properly sampled. (Default: 0)

--unsampledasmax Set energy values in unsampled regions to the maximum energy from the sampled region, or to the value provided by --maxenergy.

--maxenergy NUMBER If set, this is the energy used for unsampled regions. (Default: 0)

--noheader Do not print a header in the output file.

--includebinstat Include bin statuses (1 = sampled, 0 = unsampled, -1 = glued) in the resulting FES.

--oldrfd RFD: Use the old RFD implementation.

--fdpoints NUMBER RFD: Number of points used in the differentiation scheme (three or four points are supported) in the RFD method. (Default: 3)

--fx FORMAT Output FORMAT to print values of collective variables. (Default: %15.7e)

--fe FORMAT Output FORMAT to print values of free energy. (Default: %15.7e)

--kernel NAME GPR: Kernel type. Supported types: ardse (ARD squared exponential), ardmc52 (ARD Matern class 5/2), default (= ardse). (Default: default)

--numdiff GPR: Use numerical differentiation of the kernel function (for testing only).

--useinv GPR: Use the matrix inversion pathway (for testing only).

--calclogpl GPR: Calculate logPL.

--inczpe GPR: Include zero-point energy at the position specified by --globalmin.

--nofasterror GPR: Do not use the faster algorithm for error calculation.

--slevel VALUE Sigma-level for the confidence interval. (Default: 1)

--keepcvs SPEC Which CVs should be kept during statistical reweighting of the FES. Flags are specified in the form CV1[xCV2x...] with F and T for skipped and kept CVs, respectively. The last value pads the rest. (Default: T)

--reducedfes NAME Name of file for FES reduced by statistical reweighting, containing only kept CVs.

-v, --verbose Increase output verbosity.

--version Output version information and exit.

-h, --help Display this help and exit.