← Modern Nuclear Data Processing
Quote of the week · all quotes
The Cross Section Evaluation Working Group (CSEWG) has produced the ENDF/B libraries since 1966. Every Monday the home page shows one of the passages below, in this order. Each passage is given as it appears in the source and links to its page there, where the full context is.
The passages were chosen with the help of an AI assistant, Claude (Anthropic). The words are those of the cited authors and minutes; the selection, the short context notes and this page are not.
Week of 5 October 2026This week
The ENDF/B library is not directly used in design calculations. It must be processed first by programs such as NJOY to get either multigroup or Monte Carlo libraries.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 39 (PDF)
Week of 12 October 2026This week
However, the new library format Evaluated Nuclear Data File (ENDF) would be mathematically rigorous, e.g. specified interpolation schemes between tabulated points, so that cross section integrals, products, and ratios would yield well defined and repeatable results.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 17 (PDF)
Week of 19 October 2026This week
Without some level of data standardization, comparisons were difficult if not meaningless. The time was right for an activity such as CSEWG.
— Harry Alter, “Memories of CSEWG: 1966 - 1968”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 27 (PDF)
Week of 26 October 2026This week
This fundamental limit makes the production of files of "application independent" covariance data an unreachable goal.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 63 (PDF)
Week of 2 November 2026This week
Goldsmith is reported to have continually traveled around the country visiting one laboratory of the Project after another and he provided inter-laboratory information exchange by carrying the latest neutron cross section measurement results on various pieces of paper in his pockets.
— Norman Holden, “A Short History of Nuclear Data and Its Evaluation”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 11 (PDF)
On Hyman Goldsmith, the Manhattan Project's neutron cross section information coordinator.
Week of 9 November 2026This week
We had a very large piece of graph paper with all the measurements and their uncertainties plotted on it. We all stood around the table on which the graph paper was placed. We made our suggestions as to how we felt the curve should go, based on our understandings of the various experiments.
— Allan Carlson, “The ENDF Standards”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 53 (PDF)
On a “Task Force” evaluation of the ²³⁵U(n,f) cross section for ENDF/B-IV.
Week of 16 November 2026This week
Even if assigned, uncertainties would never be used. There simply was not sufficient interest to justify the enormous expense to implement uncertainties in reactor physics codes
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 43 (PDF)
CSEWG members in the debate on uncertainties, circa 1974, as recalled by Sol Pearlstein.
Week of 23 November 2026This week
The two main processing codes used by the ENDF community are the NJOY and PUFF systems. In principle, these two codes should yield comparable results if they both operate on the same evaluated file.
— CSEWG minutes, CSEWG and USNDP meetings, November 2009, p. 18 (PDF)
From the summary of R. Arcilla (BNL), “Covariance Processing Issues in ENDF/B-VII.0 and ENDF/A”.
Week of 30 November 2026This week
It was known that these libraries could give different answers when calculating the same reactor configuration but the dissimilarities in internal formats made it difficult to understand why the differences occurred.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 17 (PDF)
Week of 7 December 2026This week
In 1989, the chairs of the three evaluation projects, Massimo Salvatores (JEF), Sin-Iti Igarashi (JENDL), and I representing CSEWG, met in a small smoke-filled room at a hotel in Los Alamos. Many members of the NEANDC had gathered with us in that room. The smoke was courtesy of Alan Smith’s pipe.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 36 (PDF)
On the founding of WPEC.
Week of 14 December 2026This week
The basic problem is that coarse-group covariance data are already integrated over energy and angle, and this integration cannot be undone by a code.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 67 (PDF)
Week of 21 December 2026This week
In 1966, the plethora of databases and neutronics codes made methods of analysis seem mysterious, more art than science.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 41 (PDF)
Week of 28 December 2026This week
It was found that very unusual results can be obtained with discrepant correlated data. For example, combining two highly correlated discrepant data points can produce a result which is not between the two input values.
— Allan Carlson, “The ENDF Standards”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 58 (PDF)
Week of 4 January 2027This week
Of course, CSEWG had a File 6 before ENDF/B-VI, but anyone who ever tried to apply it realizes it was very restrictive and difficult to use. To my knowledge, no one ever succeeded in using it for an evaluation or, especially, in processing it.
— Phil Young, “Progression of ENDF/B-VI: 1990 to 2001”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 72 (PDF)
Week of 11 January 2027This week
To help overcome parochialism among organizations use of nuclear data, the sponsor had a powerful weapon-----FUNDS.
— Harry Alter, “Memories of CSEWG: 1966 - 1968”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 27 (PDF)
Week of 18 January 2027This week
Evaluated covariances must be reasonable, which implies that they must at least be positive semi-definite and be consistent with relevant experimental information.
— CSEWG minutes, 60th CSEWG meeting, Santa Fe, November 2010, p. 32 (PDF)
A recommendation of the IAEA Technical Meeting on Neutron Cross-Section Covariances (Vienna, September 2010), as summarised by D. Muir (ANL).
Week of 25 January 2027This week
Most handbooks of data contain useful numbers but the probable errors for those numbers are seldom given because they are difficult to determine. Therefore, the range of confidence over which the numbers can be used is not known.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 22 (PDF)
Week of 1 February 2027This week
It took more than 10 years to get versions of the AMPEX (ORNL) and MC2 (ANL) processing systems operational.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 39 (PDF)
On processing the correlated energy-angle format (MF=6).
Week of 8 February 2027This week
In 1964, Henry Honeck at Brookhaven began work on the Evaluated Neutron Data File (ENDF) as a vehicle to simplify the exchange of evaluated data.
— Norman Holden, “A Short History of Nuclear Data and Its Evaluation”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 13 (PDF)
Week of 15 February 2027This week
Thus if one, for example, triples the number of data values in ENDF by such information splitting, then the number of data covariances will increase by a factor of nine! It appears that the data evaluation community has been more willing to accept a factor-of-three growth in the size of the main data files than it has a factor-of-nine increase in the size of the corresponding covariance data files.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 66 (PDF)
Week of 22 February 2027This week
The measurer was puzzled as to how the measurements could be that much in error. The reactor physicist thought the cross section measurements were wrong and the adjustment procedure was a clever way to determine differential cross sections.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 42 (PDF)
Week of 1 March 2027This week
Sublet further noted that the processing codes have to palliate the data format deficiencies, either because the format rules have not been well defined, have been interpreted differently, or are inconsistent/unphysical.
— CSEWG minutes, 60th CSEWG meeting, Santa Fe, November 2010, p. 36 (PDF)
From the summary of a presentation comparing processing codes for the unresolved resonance range.
Week of 8 March 2027This week
A few of us believe that this early rising and not the meeting content produced nodding heads and soft rhapsodies. A solution to this problem does not exist. Some of our finest heads have nodded in agreement.
— Harry Alter, “Memories of CSEWG: 1966 - 1968”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 27 (PDF)
Week of 15 March 2027This week
Artificial Intelligence. Placed in computer programs were steps previously performed by eye and hand operations.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 22 (PDF)
Week of 22 March 2027This week
The Version II library did not predict several integral benchmark measurements as well as the Version I library, even though much of the Version II data was believed to be appreciably “better” than the Version I data.
— Allan Carlson, “The ENDF Standards”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 47 (PDF)
Week of 29 March 2027This week
Hank’s concept was to develop an application-independent, well-documented, evaluated nuclear data library.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 29 (PDF)
Week of 5 April 2027This week
There were two major factors, which influenced this decision. The first one was that the dominant features of the uncertainties are the correlations in the data and these are not easily handled in the documentation. The second one was that the uncertainty information needs to be processed together with the differential data to obtain the uncertainties in the quantities of interest.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 65 (PDF)
Quoting F. G. Perey on the decision to put uncertainty information on the data tapes.
Week of 12 April 2027This week
One aspect of Uncertainty Quantification involves recognizing that nuclear data uncertainties are meaningless unless the impact on applications is quantified, and that there is a need for feedback from data users to the nuclear data community.
— CSEWG minutes, Nuclear Data Week 2012 (CSEWG/USNDP), p. 29 (PDF)
From the summary of M. E. Rising et al. (LANL), “PFNS: Consistent Evaluations, UQ, and Propagation of Uncertainties”.
Week of 19 April 2027This week
To this day, the Hansen-Roach 16 energy group set still gives good answers in many applications but the details of how the data were selected and tweaked by its many users are not documented which is a deficiency for today's regulatory requirements.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 41 (PDF)
Week of 26 April 2027This week
The 360-50 computer had a new IBM operating system, HASP, which could process TWO jobs simultaneously.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 37 (PDF)
Week of 3 May 2027This week
The formats were carefully defined so that data could be entered with no confusion about what physics was intended and computer programs could be written to provide unique answers.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 22 (PDF)
Week of 10 May 2027This week
In my view, ENDF/B-V.2 was a pretty good database for the time, and in fact, there are still a few areas that we have not greatly improved.
— Phil Young, “Progression of ENDF/B-VI: 1990 to 2001”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 71 (PDF)
Week of 17 May 2027This week
Consideration of cross section uncertainties while neglecting spectrum uncertainties can lead to too-small uncertainties in calculated integral quantities.
— CSEWG minutes, Nuclear Data Week 2011 (CSEWG/USNDP), p. 19 (PDF)
From the summary of M. Rising, P. Talou, T. Kawano and A. Prinja (LANL) on prompt fission neutron spectra.
Week of 24 May 2027This week
Such evaluations are difficult to document and it is not clear how to determine meaningful uncertainties and covariance information.
— Allan Carlson, “The ENDF Standards”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 53 (PDF)
Week of 31 May 2027This week
The battle over migrating to ENDF/C raged on for more than four years. The resistance of those organizations whose data needs were satisfied by existing formats was fierce.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 38 (PDF)
Week of 7 June 2027This week
The mathematical transformations needed to implement this concept within a practical sensitivity analysis system seem relatively straightforward, but someone needs to do the actual work.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 67 (PDF)
Week of 14 June 2027This week
If the evaluation of cross sections was to be considered a science instead of an art then the assignment of uncertainties to data was important.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 43 (PDF)
Week of 21 June 2027This week
They were a hard-working group, focused at the tasks on hand, competitively pushing their ideas, graciously (often begrudgingly) accepting ideas and concepts not their own, establishing a firm camaraderie and, perhaps most important, maintaining a sense of humor throughout.
— Harry Alter, “Memories of CSEWG: 1966 - 1968”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 28 (PDF)
Week of 28 June 2027This week
The role of covariances is critical since they actually determine which quantities are modified and how much they are changed during adjustment of the library to the integral experiments.
— CSEWG minutes, Nuclear Data Week 2012 (CSEWG/USNDP), p. 8 (PDF)
From the meeting summary’s paragraph on covariance data.
Week of 5 July 2027This week
The data evaluators were being overwhelmed. The only solution was to increase the evaluators output by introducing computer mechanized evaluation of the cross sections.
— Norman Holden, “A Short History of Nuclear Data and Its Evaluation”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 13 (PDF)
Week of 12 July 2027This week
The NNDC had to lock the master tapes in a large safe, which now sits empty in the NNDC library.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 35 (PDF)
On restricting the distribution of ENDF/B-V.
Week of 19 July 2027This week
Somehow, these subcommittees displayed wisdom beyond their experience. The objectives chosen were modest in scope.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 21 (PDF)
Week of 26 July 2027This week
It should be emphasized that the evaluations for ENDF/B-I were basically used to check out the processing codes which had been written.
— Allan Carlson, “The ENDF Standards”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 47 (PDF)
Week of 2 August 2027This week
Moreover, information that is not part of the existing format cannot be stored in the evaluated files without the processing codes being instructed to ignore the information.
— CSEWG minutes, Nuclear Data Week 2011 (CSEWG/USNDP), p. 22 (PDF)
From D. Brown’s (BNL) summary of the meeting on an XML data structure for nuclear data (GND).
Week of 9 August 2027This week
In this way the resulting discrepancies between calculation and experiment remained a gadfly to spur greater understanding of the issues.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 44 (PDF)
Week of 16 August 2027This week
We conclude that the successes achieved in this pioneering effort constitute a real success, but that the work is far from finished.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 63 (PDF)
Week of 23 August 2027This week
He was the consummate nuclear data evaluator. He combined knowledge of theory, experiments, computers and work ethic to be the most prolific evaluator in the history of CSEWG.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 32 (PDF)
On Phil Young.
Week of 30 August 2027This week
Mindful of the impact of changes on the cost and schedule of reprogramming codes, the Subcommittee's conservative response helped stabilize the ENDF/B effort.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 20 (PDF)
Week of 6 September 2027This week
The main issue: Everyone agrees in principle that evaluated covariances files ought to be of good quality to be included in ENDF/B (a QA requirement).
— CSEWG minutes, CSEWG and USNDP meetings, November 2009, p. 21 (PDF)
From the summary of D. Smith (ANL), “Quality Assurance Requirements for ENDF Covariance Evaluations”.
Week of 13 September 2027This week
People were the engine that drove CSEWG.
— Harry Alter, “Memories of CSEWG: 1966 - 1968”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 27 (PDF)
Week of 20 September 2027This week
The covariances (uncertainty matrix) are application independent. The sensitivity coefficient matrix consisted of the relative change in a calculated result to changes in the cross sections. The sensitivity matrix is application dependent. The product of these two matrices was the predicted uncertainty in the result.
— Sol Pearlstein, “Cross Section Evaluation Working Group Physics”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 43 (PDF)
Week of 27 September 2027This week
Today, the ENDF format is the de facto international standard for storage and exchange of evaluated nuclear data. The inventors of the ENDF system and the ENDF format probably did not foresee the international impact of their work.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 35 (PDF)
Week of 4 October 2027This week
Any discussion about QA issues related to covariances should not be approached without also considering a comparable review of ENDF files in general.
— CSEWG minutes, Nuclear Data Week 2012 (CSEWG/USNDP), p. 32 (PDF)
From the session on the covariance quality-assurance document (D. L. Smith, ANL).
Week of 11 October 2027This week
Prior to the new LANSCE measurements, there were very few angular distribution measurements of ¹⁶O gamma rays. Consequently, evaluators were forced to infer the integrated cross sections from measurements at 1 or 2 angles, using assumptions about the angular distributions that were often erroneous.
— Phil Young, “Progression of ENDF/B-VI: 1990 to 2001”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 74 (PDF)
Week of 18 October 2027This week
It appears that an investment in further development of the processing tools now would yield substantial benefits, especially in terms of stimulating new work by data evaluators in this field.
— D. W. Muir, “The Covariance Files of ENDF/B”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 68 (PDF)
Week of 25 October 2027This week
The concepts we were investigating seem to be pretty primitive by today’s standards. But at the time, the approach seemed to offer a better, more efficient way to evaluate nuclear data.
— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 37 (PDF)
Week of 1 November 2027This week
International cooperation has helped shore up the available resources. No one person has proved indispensable.
— Sol Pearlstein, “Cross Section Evaluation Working Group History”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 23 (PDF)