HFBTruncate¶

class HFBTruncate : public kotekan::Stage¶

Truncates absorber data and weight values.

Absorber values are truncated to a precision based on their weight.

Buffers

in_buf The input stream.
- Format: HFBBuffer.
- Metadata: HFBMetadata
out_buf The output stream with truncated values.
- Format: HFBBuffer.
- Metadata: HFBMetadata

Author

James Willis

Warning

Don’t run this anywhere but on the transpose (gossec) node. The OpenMP calls could cause issues on systems using kotekan pin priority threads (likely the GPU nodes).

Param err_sq_lim:: Limit for the error of absorber truncation.
Param weight_fixed_precision:: Fixed precision for weight truncation.
Param data_fixed_precision:: Fixed precision for absorber truncation (if weights are zero).

Public Functions

HFBTruncate(kotekan::Config &config, const std::string &unique_name, kotekan::bufferContainer &buffer_container)¶: Constructor; loads parameters from config.

~HFBTruncate() = default¶

virtual void main_thread() override¶: Main loop over buffer frames.