The MATXS File Format

MATXS is a generalized, flexible multigroup interface file based on the CCCC (Committee on Computer Code Coordination) standard conventions and produced by NJOY's matxsr module. It stores cross-section vectors and group-to-group transfer matrices for an arbitrary set of particles, materials and reactions.

Radiant reads and writes a MATXS file through the "matxs" cross-section source and the cs.write(path,"matxs") method (see Section 4.5 of the User's Guide). This page describes the record structure, the reaction-name keywords and the particle codes used by Radiant, so that the files can be inspected or exchanged with other tools.

File structure

A MATXS file is a sequence of records. The first four records describe the whole file; the group structures are then given once per particle; and the data of each material follows as a material-control record and, per submaterial, a set of vector and matrix records.

file identification        (always)
file control               (always)
set hollerith id           (always)
file data                  (always)

  (repeat per particle)
    group structure         (always)

  (repeat per material)
    material control        (always)
      (repeat per submaterial)
        vector control      (if the submaterial holds vector reactions)
        vector block(s)
        matrix control      (if the submaterial holds a matrix reaction)
        matrix sub-block(s)

In the BCD (ASCII) encoding each record begins with a 4-character tag. The records written by Radiant are:

npart is the number of particles, nmat the number of materials, and ntype the number of data types. Radiant defines one data type per ordered (incident, outgoing) particle pair, so ntype = npart²; the type of submaterial s carries the matrix for its pair and, on the "self" pairs (incident = outgoing), the vector reactions of the incident particle.

Particle codes

The hollerith particle codes (hprt, and the prefix of every reaction name) are:

Data-type names (htype) are the concatenation of the incident and outgoing codes, e.g. gg (photon→photon), gb (photon→electron), bb (electron→electron).

Reaction-name keywords

Each reaction name is the incident-particle code followed by a reaction suffix. The vector reactions (record 6d/7d) are written only on the self pairs; the matrix reaction (record 8d/9d) is written for every pair. Using the photon prefix g as an example:

For every incident particle, *tot0 has length Ng, *abs has length Ng, and *edep has length Ng+1. Charge deposition is represented by either *cdep with length Ng+1 or *char with length Ng; the latter receives a zero appended at the lower boundary. *momt and *stpw are optional and default to zero vectors when absent. The legacy private names *momt, *stpw, and *cdep remain accepted on input but are not emitted.

On reading, *heat is ignored as an authoritative value because the full Ng+1 energy- deposition vector is recovered from *edep.

Scattering-matrix banding

The transfer matrix is stored in the compact banded form used by MATXS. For each outgoing group, jband gives the number of contributing incident groups and ijj the highest incident group number of the band. Within a band the data are ordered by Legendre moment (P₀ band, then P₁ band, …), and the incident groups within each moment are listed in descending order.

Encodings

Radiant supports two encodings, selected by the binary argument of cs.write. The reader auto-detects the encoding from the first bytes of the file.

• BCD / ASCII (binary = false, default) — tagged text records. Reals are written as fixed-width e12.5 fields (6 per line), integers as i6 (12 per line) and hollerith identifiers as a8 (8 per line).
• Energy units — Radiant group boundaries are stored internally in MeV; MATXS group boundaries are written and read in eV, with a factor of 10⁶ applied at the format boundary.
• Binary (binary = true) — each record is written as a single FORTRAN-style unformatted record: a 4-byte Int32 length marker, a payload of native-endian Int32 integers, Float64 reals and 8-byte hollerith fields, and a trailing length marker.

Example

# Write a built library, then read it back into a fresh Cross_Sections
cs.write("./library.matxs","matxs")        # or "matxs", true for the binary encoding

cs2 = Cross_Sections()
cs2.set_source("matxs")
cs2.set_file("./library.matxs")
cs2.set_materials([water])                  # same materials, in the same order
cs2.set_particles([electron,photon])        # same particles
cs2.build()