diff --git a/scripts/covmatrix_plot.py b/scripts/covmatrix_plot.py
index 79d7ff5658a0792b862b7fe27e01d3781e853fa8..50d6eb62faeacaa6d0cf287669af672a88985894 100755
--- a/scripts/covmatrix_plot.py
+++ b/scripts/covmatrix_plot.py
@@ -8,6 +8,7 @@ from typing import TYPE_CHECKING
from h5py import File, Group
from matplotlib import pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
+from matplotlib.colors import SymLogNorm
from numpy import arange, diagonal, isnan, quantile
from numpy.typing import NDArray
@@ -26,7 +27,6 @@ def main(opts: Namespace) -> None:
model = group["model"][:]
edges = group["edges"][:]
- blocksize = edges.size - 1
sfile, sgroup, smodel = None, None, None
if opts.subtract:
@@ -48,6 +48,11 @@ def main(opts: Namespace) -> None:
except KeyError:
elements = None
+ blocksize = edges.size - 1
+ block_selected = 1
+ block_selection = slice(blocksize * block_selected, blocksize * (block_selected + 1))
+ block_name = elements[block_selected]
+
if opts.output is not None:
ofile = opts.output and opts.output[0] or opts.input.replace(".hdf5", ".pdf")
assert ofile != opts.input and ofile.endswith(".pdf")
@@ -156,8 +161,8 @@ def main(opts: Namespace) -> None:
ylabel="bin",
title=f"Covariance matrix {name}{title_suffix}",
)
- pcolor_with_blocks(matrix_cov, blocks=elements, cmap=cmap)
- if pdf:
+ hm = pcolor_with_blocks(matrix_cov, blocks=elements, cmap=cmap)
+ if pdf and hm is not None:
pdf.savefig()
if vmax is not None:
@@ -166,12 +171,25 @@ def main(opts: Namespace) -> None:
111,
xlabel="",
ylabel="bin",
- title=f"Covariance matrix (truncated) {name}{title_suffix}",
+ title=f"Covariance matrix {name} (truncated){title_suffix}",
)
- pcolor_with_blocks(matrix_cov, blocks=elements, cmap=cmap, vmax=vmax)
- if pdf:
+ hm = pcolor_with_blocks(matrix_cov, blocks=elements, cmap=cmap, vmax=vmax)
+ if pdf and hm is not None:
pdf.savefig()
+ plt.figure(figsize=figsize_2d)
+ ax = plt.subplot(
+ 111,
+ xlabel="",
+ ylabel="bin",
+ title=f"Covariance matrix {name} (symlog){title_suffix}",
+ )
+ hm = pcolor_with_blocks(
+ matrix_cov, blocks=elements, cmap=cmap, symlog=True
+ )
+ if pdf and hm is not None:
+ pdf.savefig()
+
plt.figure(figsize=figsize_2d)
ax = plt.subplot(
111,
@@ -179,8 +197,8 @@ def main(opts: Namespace) -> None:
ylabel="bin",
title=f"Covariance matrix {name} (blocks){title_suffix}",
)
- pcolor_with_blocks(bmatrix_cov, blocks=elements, cmap=cmap)
- if pdf:
+ hm = pcolor_with_blocks(bmatrix_cov, blocks=elements, cmap=cmap)
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -188,12 +206,39 @@ def main(opts: Namespace) -> None:
111,
xlabel="",
ylabel="bin",
- title=f"Covariance matrix {name} ({elements[0]}){title_suffix}",
+ title=f"Covariance matrix {name} (blocks, symlog){title_suffix}",
)
- pcolor_with_blocks(
- matrix_cov[:blocksize, :blocksize], blocks=elements[:1], cmap=cmap
+ hm = pcolor_with_blocks(bmatrix_cov, blocks=elements, cmap=cmap, symlog=True)
+ if pdf and hm is not None:
+ pdf.savefig()
+
+ plt.figure(figsize=figsize_2d)
+ ax = plt.subplot(
+ 111,
+ xlabel="",
+ ylabel="bin",
+ title=f"Covariance matrix {name} (block {block_name}){title_suffix}",
)
- if pdf:
+ hm = pcolor_with_blocks(
+ matrix_cov[block_selection,block_selection], blocks=elements[:1], cmap=cmap
+ )
+ if pdf and hm is not None:
+ pdf.savefig()
+
+ plt.figure(figsize=figsize_2d)
+ ax = plt.subplot(
+ 111,
+ xlabel="",
+ ylabel="bin",
+ title=f"Covariance matrix {name} (block {block_name}, symlog){title_suffix}",
+ )
+ hm = pcolor_with_blocks(
+ matrix_cov[block_selection,block_selection],
+ blocks=elements[:1],
+ cmap=cmap,
+ symlog=True,
+ )
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -203,8 +248,8 @@ def main(opts: Namespace) -> None:
ylabel="bin",
title=rf"Relative covariance matrix {name}, %虏{title_suffix}",
)
- pcolor_with_blocks(matrix_cov_rel, blocks=elements, cmap=cmap)
- if pdf:
+ hm = pcolor_with_blocks(matrix_cov_rel, blocks=elements, cmap=cmap)
+ if pdf and hm is not None:
pdf.savefig()
if vmax_rel is not None:
@@ -213,12 +258,12 @@ def main(opts: Namespace) -> None:
111,
xlabel="",
ylabel="bin",
- title=rf"Relative covariance matrix {name}, %虏{title_suffix}",
+ title=rf"Relative covariance matrix {name} (truncated), %虏{title_suffix}",
)
- pcolor_with_blocks(
+ hm = pcolor_with_blocks(
matrix_cov_rel, blocks=elements, cmap=cmap, vmax=vmax_rel
)
- if pdf:
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -226,12 +271,41 @@ def main(opts: Namespace) -> None:
111,
xlabel="",
ylabel="bin",
- title=f"Relative covariance matrix {name} ({elements[0]}){title_suffix}",
+ title=rf"Relative covariance matrix {name} (symlog), %虏{title_suffix}",
)
- pcolor_with_blocks(
- matrix_cov_rel[:blocksize, :blocksize], blocks=elements[:1], cmap=cmap
+ hm = pcolor_with_blocks(
+ matrix_cov_rel, blocks=elements, cmap=cmap, symlog=True
)
- if pdf:
+ if pdf and hm is not None:
+ pdf.savefig()
+
+ plt.figure(figsize=figsize_2d)
+ ax = plt.subplot(
+ 111,
+ xlabel="",
+ ylabel="bin",
+ title=f"Relative covariance matrix {name} (block {block_name}){title_suffix}",
+ )
+ hm = pcolor_with_blocks(
+ matrix_cov_rel[block_selection,block_selection], blocks=elements[:1], cmap=cmap
+ )
+ if pdf and hm is not None:
+ pdf.savefig()
+
+ plt.figure(figsize=figsize_2d)
+ ax = plt.subplot(
+ 111,
+ xlabel="",
+ ylabel="bin",
+ title=f"Relative covariance matrix {name} (block {block_name}, symlog){title_suffix}",
+ )
+ hm = pcolor_with_blocks(
+ matrix_cov_rel[block_selection,block_selection],
+ blocks=elements[:1],
+ cmap=cmap,
+ symlog=True,
+ )
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -241,8 +315,8 @@ def main(opts: Namespace) -> None:
ylabel="bin",
title=f"Correlation matrix {name}{title_suffix}",
)
- pcolor_with_blocks(matrix_cor, blocks=elements, cmap=cmap)
- if pdf:
+ hm = pcolor_with_blocks(matrix_cor, blocks=elements, cmap=cmap)
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -250,16 +324,16 @@ def main(opts: Namespace) -> None:
111,
xlabel="",
ylabel="bin",
- title=f"Correlation matrix {name} ({elements[0]}){title_suffix}",
+ title=f"Correlation matrix {name} (block {block_name}){title_suffix}",
)
hm = pcolor_with_blocks(
- matrix_cor[:blocksize, :blocksize],
+ matrix_cor[block_selection,block_selection],
blocks=elements[:1],
pcolormesh=True,
cmap=cmap,
)
# heatmap_show_values(hm, lower_triangle=True)
- if pdf:
+ if pdf and hm is not None:
pdf.savefig()
plt.figure(figsize=figsize_2d)
@@ -276,7 +350,7 @@ def main(opts: Namespace) -> None:
logger.info(f"Plot {name}")
- if pdf:
+ if pdf and hm is not None:
pdf.savefig()
if opts.show:
@@ -285,7 +359,7 @@ def main(opts: Namespace) -> None:
if not opts.show:
plt.close("all")
- if pdf:
+ if pdf and hm is not None:
pdf.__exit__(None, None, None)
logger.info(f"Write output file: {ofile}")
@@ -394,28 +468,41 @@ def pcolor_with_blocks(
pcolormesh: bool = False,
sep_kwargs: Mapping = {},
vmax: float | None = None,
+ symlog: bool = False,
**kwargs,
):
from numpy import fabs
- from numpy.ma import array
dmin = data.min()
dmax = data.max()
+ if dmin==dmax==0:
+ return None
+
bound = max(fabs(dmin), dmax)
if vmax is not None:
vmin, vmax = -vmax, vmax
else:
vmin, vmax = -bound, bound
+ if symlog:
+ pcolor_kwargs = dict(
+ norm=SymLogNorm(
+ linthresh=bound * 0.01, linscale=0.05, vmin=vmin, vmax=vmax
+ ),
+ **kwargs,
+ )
+ else:
+ pcolor_kwargs = dict(vmin=vmin, vmax=vmax, **kwargs)
+
# fdata = fabs(data)
# data = array(data, mask=(fdata < 1.0e-9))
ax = plt.gca()
ax.set_aspect("equal")
if pcolormesh:
- hm = ax.pcolormesh(data, *args, vmin=vmin, vmax=vmax, **kwargs)
+ hm = ax.pcolormesh(data, *args, **pcolor_kwargs)
else:
- hm = ax.pcolorfast(data, *args, vmin=vmin, vmax=vmax, **kwargs)
+ hm = ax.pcolorfast(data, *args, **pcolor_kwargs)
hm.set_rasterized(True)
if colorbar:
add_colorbar(hm, rasterized=True)