The example input files in this directory calculate binding energies for actin dimers. This is an example of a large biomolecule binding energy calculation that often requires parallel focusing.

This example was contributed by Dave Sept.

Input File	Description	APBS Version	Results (kJ/mol)	UHBD (kJ/mol)
apbs-mol-auto.in	Sequential, 2-level focusing to ≤ 0.725 A, NPBE, srfm mol	3.0	104.868	106.7 (1.00 A res., NPBE)
		1.5	104.868
		1.4.2	104.868
		1.4.1	104.8683
		1.4	104.8683
		1.3	104.8683⁸
		1.2.1	104.867
		1.2	104.867
		1.1.0	104.867⁵
		1.0.0	104.868
		0.5.1	104.868³
		0.5.0	105.0338²
		0.4.0	104.8895
apbs-smol-auto.in	Sequential, 2-level focusing to ≤ 0.725 A, NPBE, srfm smol	3.0	109.5841	106.7 (1.00 A res., NPBE)
		1.5	109.5841
		1.4.2	109.5841
		1.4.1	109.5841
		1.4	109.5841
		1.3	109.5841⁸
		1.2.1	109.5829
		1.2	109.5829
		1.1.0	109.5829⁵
		1.0.0	109.5841
		0.5.1	109.5841³
		0.5.0	109.7518²
		0.4.0	109.6043¹
		0.3.2	90.8704
		0.3.1	88.6101
		0.3.0	88.6101
		0.2.6	88.6101
		0.2.5	88.6101
		0.2.4	88.6101
		0.2.3	88.6064
		0.2.2	90.829
		0.2.1	90.829
		0.2.0	90.829
		0.1.8	90.84
apbs-mol-parallel.in	Parallel with 8 processors, focusing to ~0.9 A, LPBE, srfm mol	1.5	98.1746	106.7 (1.00 A res., NPBE)
		1.4.2	98.1746
		1.4.1	98.1746
		1.4	98.1746
		1.3	98.1746⁸
		1.2.1	98.1733⁷
		1.2	98.1635⁶
		1.1.0	98.1630⁵
		1.0.0	98.1643⁴
		0.5.1	98.1654³
		0.5.0	98.3530²
		0.4.0	98.1834
apbs-smol-parallel.in	Parallel with 8 processors, focusing to ~0.9 A, LPBE, srfm smol	1.5	115.542	106.7 (1.00 A res., NPBE)
		1.4.2	115.542
		1.4.1	115.5421
		1.4	115.5421⁹
		1.3	115.5422⁸
		1.2.1	115.5409⁷
		1.2	115.5563⁶
		1.1.0	115.5560⁵
		1.0.0	115.5573⁴
		0.5.1	115.5584³
		0.5.0	115.7492²
		0.4.0	115.5751¹
		0.3.2	87.1121
		0.3.1	87.1121
		0.3.0	90.2573
		0.2.6	90.2573
		0.2.5	90.2573
		0.2.4	90.2573
		0.2.3	90.2543
		0.2.2	91.9450
		0.2.1	91.945
		0.2.0	91.939
		0.1.8	91.67

¹ The discrepancy in values between versions 0.4.0 and 0.3.2 is most likely due to three factors:

A bug fix in Vacc_molAcc which removed spurious regions of high internal dielectric values
A switch in the algorithm used to compute the dielectric smoothing for srfm smol.
The addition of the Vacc sphere density (sdens keyword) as a variable and a change in the default sdens value from 3.0 to 10.0

² The discrepancy in values between versions 0.5.0 and 0.4.0 is most likely due to the following factor(s):

A change in the autofocusing routine for APBS

³ The discrepancy in values between versions 0.5.1 and 0.5.0 is most likely due to the following factor(s):

Bug fix regarding multipole behavior for neutral proteins

⁴ The discrepancy in values between versions 0.5.1 and 1.0.0 was due to the execution of the previous APBS tests on a PowerPC platform with the XLC/XLF compilers. Running with binaries compiled with gcc/gfortran or the Intel compilers gives identical results between versions 0.5.1 and 1.0.0.

⁵ The discrepancy in values between versions 1.0.0 and 1.1.0 is due to a bugfix in the implementation of the boundary conditions. This bug introduces a very small error (generally less than 1%) the calculated results.

⁶ APBS 1.2 has switched the multigrid smoothing algorithm from standard Gauss-Seidel to Gauss-Seidel red/black in order to facilitate parallelization. This switch has caused small differences in individual calculation energies which, when combined to the final answer, create larger errors (up to 0.04%). These errors can be reduced by resetting the APBS error tolerance to 1e-9 or smaller values. For a more detailed explanation, please see the FAQ, [here](http://www.poissonboltzmann.org/docs/apbs-faq/#sources error calculation).

⁷ The discrepancy in values between versions 1.2 and 1.2.1 is most likely due to the following factor(s):

Fixed a bug in Vpmg_fillcoCoefMolIon which causes npbe based calculations to return very large energies

⁸ The discrepancy in values between versions 1.2.1 and 1.3 is most likely due to the following factor(s):

Fixed a bug in Vpmg.c which causes zero potential values on boundaries in non-focusing calculations.

⁹ The discrepancy in values between versions 1.3 and 1.4 is most likely due to the following factor(s):

Translation of contrib/pmgZ library from FORTRAN to C
Differences in numerical implementations between FORTRAN and C compilers result in small round-off discrepencies
Small margins due to these round-off discrepencies acumulate in the computations

Please see the ChangeLog or the APBS website for more information.