I take it you have the 32bit libraries installed. They are not normally
installed
by default. Then you need to make sure the app is linked with these 32bit
libraries.
I suspect that you may have linked to the 64-bit libraries.
Adam
Hi all,
So, I know this probably isn’t the right place,
Can it be parallelized? That is how you reduce run-time. One of the tests
matrix-matrix multiplication has been successfully speeded up by using GPUs.
CUDA is the language used for this, which is a derivative of C. To be fair you
only see the benefit for really large matrices, smaller ones
Dear Chang,
Some CCP4 progs can be used with a multi-core machine,
using OpenMP threads (including refmac it would appear). You will
need to compile the code from source rather than taking the binary
versions, which is much less convenient.
Even if the CCP4 apps are not parallel
Dear Sergei,
Second point is probably easier to do. An alternative to NFS is sshfs. The
advantage is that it uses SSH which is installed by default and configured the
same way. If you generate key pairs you can use ssh or sshfs without a password.
Check this page below;
Hi Ed,
You can have both types of error in a single experiment, however you
cannot determine
statistical (precision or as Ian says uncontrollable) error with one
experiment. The manufacturer
will usually give some specs on the pipette, 6ul +/- 1ul. In order to verify
the specs
you would
Dear Ed,
It is not usual to find small discrepancies in floating point
calculations. A simple test to demonstrate this is to calculate
sum 1.0/n n=1-100 and n=100-1. You might find
that the results differ. For 4-byte reals will tend to differ in the
6th significant digit and
Dear Ed,
The error does indeed happen in ccp4lib. One of the first routines called by
CCP4
progs is ccp4fyp. This initialises the CCP4 environment. See
lib/cctbx/cctbx_sources/ccp4io/lib/src/ccp4_general.c.
If you look at the code you can see that $CINCL is determined at run-time. You
are
Following what Marcin said, if you have a compiler other than gcc/gfortran
then I would definitely recommend compiling from source. Generally they
give much better performance and in addition there
might be optimised maths functions.
Adam
One reason to compile refmac on Linux is that
Hi Ed,
It looks as though you have not sourced $CCP4/include/ccp4.setup.
This needs to be customized and sourced before you configure and make
CCP4.
Adam
You can just connect two machines together with an ethernet cable, no switches
or
routers. However you need a crossover cable to do this. Most likely you have
standard
cables only in the lab.
Adam
Dear Michael,
In sftools you can generate a set of FreeR flags on only a subset of
the reflections. If you have a combined dataset (high and low relsoution), then
select the high res refls, construct the FreeR set (unselected refls should not
be changed) and write out.
sftools
READ
Those people considering running faster code should consider using GPGPUs.
Advantages of GPUs are that they have many more cores than CPU. The
disadvantages are that the communication between the CPU and GPU is slow and
memory management is tricker. Thus there is no guarantee that code will
Dear Brigitte,
Looking at the formulae it could be possible to get those results. Take an
example
below
Rho_cal = -0.11, 0.0, 0.05, 0.05
Rho_obs = -0.08, 0.01, 0.04, 0.04
R-fac = 0.02/0.0 = undefined
Correl = 0.0032 - (-0.0025*0.0025)
Dear All,
There is a free alternative to MS Office, OpenOffice from Oracle. It
can read and write MS Word files and save as PDF. There are some issues
with names of spreadsheet functions when moving from OO to MS Office.
If you use latex and beamer then there is no need to either ;-). I
Dear James,
You can get around this by having an artificial SIGFP. Two ways,
you can assign an existing column to SIGFP i.e.
LABIN FP=FP SIGFP=FP
I am not sure if refmac will complain about this because MTZ data columns
have a type 'F' for magnitudes and 'Q' for errors.
Secondly you
Hi Dirk,
If you have a N points of a 1D real discrete function, there will be Fourier
coefficients indexed h=0,1,2,...,N-1. Taking N as odd, there will be
int(N/2)+1 independent Fourier coefficients but your h(max) will in fact
be 'N-1'. In crystallography we write h(N-1) as h(-1) etc and
Dear Leila,
The CCP4 program FFT, which generates 3D maps from reflection data
can also produce projections. Use the keyword PROJECTION. It will only
give projections down an axis but if the X-ray SG is trigonal then you
will be OK. Another CCP4 prog MAPSLICER can be used to view the map.
Dear Lijun Liu,
You can use the CCP4 prog sftools to determine if there are
just a few strong reflections.
Adam
On Tue, 20 Jul 2010, Lijun Liu wrote:
Hi all,
I solved a structure and the refinement is close to complete.
However, some residual density puzzles me.
The SG is P212121,
I believe that Francois is correct in saying that a
feature of intel compilers is that they do not work
well on non-intel CPUs. Can't imagine why that would be.
If you have a an intel CPU then using an intel compiler
would be an advantage. Code optimization is usually
architecture specific, so
Hi James,
If you use
P(A=5,B=3|p) = p^5 x (1-p)^3 x Choose(8,3)
which is the probability of A winning 5 times and B winning 3 times
given a known value for p. You have a combinatoric term 8 choose 3
which will cancel out in the final equation.
Substitute this into the equation bottom of
I forgot to say the assumption is that P(p) is uniform. Adam
On Wed, 28 Apr 2010, Adam Ralph wrote:
Hi James,
If you use
P(A=5,B=3|p) = p^5 x (1-p)^3 x Choose(8,3)
which is the probability of A winning 5 times and B winning 3 times
given a known value for p. You have a combinatoric
I think that you should suggest a new unit of 10^(-11) m, a JHm
perhaps. If it is convenient to have B in A^2 then u^2 should be
in JHm^2.
Adam
On Thu, 19 Nov 2009, James Holton wrote:
Many textbooks describe the B factor as having units of square Angstrom
(A^2), but then again, so does
Adam
On Thu, 1 Oct 2009, Vellieux Frederic wrote:
Thanks Adam,
Do you know if it's possible to get hold of cctbx subroutines (for free)
in order to incorporate them into the program that requires them?
Thanks again,
Fred.
Adam Ralph wrote:
Hi Fred,
It is part of cctbx in CCP4
Dear Claudia,
I use Ubuntu and I am happy with it, cannot say if it is
the best. It is relatively easy to maintain which is what you
want.
As for HD, try and determine how much space a typical project
would need and how many concurrent projects there might be. Data
from images would be
What's the date today? It must be Christmas I have lots of stuff to
publish. A
You can use the CCP4 program sftools. You will need
to use the program twice if you want to keep the differences
positive ie
F1 - F2 if F1F2
and
F2 - F1 if F2F1
Adam
On Thu, 29 Jan 2009, Rana Refaey wrote:
Hi,
I was wondering if anyone knows what programme I need to use to
Dear Jan,
I am using the 64-bit version of Ubuntu plus CCP4 without
any porblems. The only application that needed the 32-bit
libraries was the flash plugin for firefox. It seems that not
all useful applications have been converted to 64-bit. Using
the two libraries together has not been a
Dear Xie,
In order to run DISTANG for the command line you must type something
like this
distang xyzin your_PDB.file
If you just type `distang` the message below will appear.
Adam
On Wed, 6 Aug 2008, Xie Jiabao wrote:
Hello,
I am trying to run the CCP4 program DISTANG from the
Hi Pietro,
It looks as though SFALL thinks you have fractional coords and
has coverted them into orthog ones. Does the PDB have a CRYST1 and
SCALEi cards?
Adam
On Thu, 24 Apr 2008, Pietro Roversi wrote:
Dear everyone,
I am using sfall in MODE ATMMAP RESMOD to
Dear Shivesh,
I believe that Fedora 8 has a more updated version of
libstdc++ in /usr/lib. You can install libstdc++.so.5 from
compat-libstdc++-33-3.2.3-62 using yum (if you have it). An
easier approach would be to make a symbolic link between
libstdc++.so.5 and libstdc++.so.6 but there maybe
Dear Flip,
I have had trouble with video drivers for various flavours
of Linux. If if driver is not working then, as you say, you just
cannot do anything after booting normally. You need to boot the
machine in a safe mode or single user mode (init 1). You can then
edit the /etc/X11/XF86Config
There is also a CCP4 prog that compares phase sets,
phistats. Don't know if this does what Phil suggests.
Adam
On Wed, 29 Aug 2007, Phil Evans wrote:
Kevin
Does this give the correlation between two (weighted) complex Fs,
which is arguably (ref Bricogne) the best measure, as it
MAPMAN from Uppsala will do this. Adam
On Thu, 2 Aug 2007, Bernhard Rupp wrote:
Dear All -
before I code ahead - is there a simple program/script available where I can
read
in a map and it gives me a basic 1d density histogram (or the tabular
data at least) in return?
Thx, br
33 matches
Mail list logo
