Spectra: Fast Estimation of the Pattern Frequency Spectrum

This is a 100% JavaScript implementation of the Spectra algorithm described in "Fast estimation of the pattern frequency spectrum" (van Leeuwen & Ukkonen, 2014) [pdf]. Some details about the implementation, as well as a performance comparison between different JavaScript engines can be found in a short technical report.

The example datasets are automatically loaded from the server. Please be patient, loading Connect and Pumsbstar may take a few seconds. On these datasets running the estimation using default parameters may take up to one minute. Lower the number of samples if you don't want to wait. If your browser says that the script has stopped responding, just tell it to keep going. The algorithm will finish eventually.

For the example datasets we provide the exact frequency spectrum (computed offline). This is shown in black. The estimated frequency spectrum is shown in blue, while the expected frequency spectrum in data having the same column marginals is shown in green.

Load example data:
Or load data (in FIMI format) from local filesystem:
num rows:
num cols:
min sigma:
max sigma:
num samples: