The still open 5 000 Euro challenge
Posted: Wed Jun 18, 2014 11:47 pm
Unfortunately, an attempt to claim the 10 000 Euro prize for a successful attempt before June 11 failed at the adjudication stage. The challenge is still open, but now only 5 000 Euro is on offer.
http://www.sciphysicsforums.com/spfbb1/viewtopic.php?f=6&t=52#p1898
In the light of experience the following clarifications are added.
The number N of directions in the two files is preferably around 10 000. It must not exceed 100 000.
Directions may be represented by polar coordinates (theta, phi) (azimuth, zenith) in degrees or angles; or in Cartesian coordinates (x, y, z) of vectors in R^3. The third Cartesian coordinate z, or the polar coordinate phi, may be omitted. The submitter should make clear what is his or her choice.
In the formulas for the correlations, the same two sets of N directions u_k, v_k, k = 1, ..., N, is used in each separately calculated correlation. The formulas use the representation of directions in R^3 by real vectors in R^3; "dot" is the usual dot product of vectors in R^3, "sign" is the usual sign function; addition, multiplication and division are all by ordinary real number arithmetic.
The challenge is open to all. The number of submissions per person is unlimited, but at most one appeal to adjudication is allowed. Files should be posted on internet and the submission announced on this forum, and to me by email. A computer script which generates the data files with pre-set random seed and two explicit "write" statements is an allowed form of submission, subject to availability on my side of the computer language used. However the two files still need to be posted to internet, to allow easy independent verification.
If adjudication is required, we will together decide on and then approach suitable adjudicators and request their assistance. Their working procedure is up to them e.g. majority vote, chairman decides, or something else. The letter inviting adjudicators to take on this job can be found elsewhere on this forum. On request, I can also post it here again.
http://www.sciphysicsforums.com/spfbb1/viewtopic.php?f=6&t=52#p1898
In the light of experience the following clarifications are added.
The number N of directions in the two files is preferably around 10 000. It must not exceed 100 000.
Directions may be represented by polar coordinates (theta, phi) (azimuth, zenith) in degrees or angles; or in Cartesian coordinates (x, y, z) of vectors in R^3. The third Cartesian coordinate z, or the polar coordinate phi, may be omitted. The submitter should make clear what is his or her choice.
In the formulas for the correlations, the same two sets of N directions u_k, v_k, k = 1, ..., N, is used in each separately calculated correlation. The formulas use the representation of directions in R^3 by real vectors in R^3; "dot" is the usual dot product of vectors in R^3, "sign" is the usual sign function; addition, multiplication and division are all by ordinary real number arithmetic.
The challenge is open to all. The number of submissions per person is unlimited, but at most one appeal to adjudication is allowed. Files should be posted on internet and the submission announced on this forum, and to me by email. A computer script which generates the data files with pre-set random seed and two explicit "write" statements is an allowed form of submission, subject to availability on my side of the computer language used. However the two files still need to be posted to internet, to allow easy independent verification.
If adjudication is required, we will together decide on and then approach suitable adjudicators and request their assistance. Their working procedure is up to them e.g. majority vote, chairman decides, or something else. The letter inviting adjudicators to take on this job can be found elsewhere on this forum. On request, I can also post it here again.