René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"[Bug]: Corruption of 3D surface plot at log scale with negative or zero values (v3.11.0rc2)","articleBody":"### Bug summary\n\nStrange behavior of 3D surface plot axes in log scale if the data contains negative or zero values. With values \u003e0 surface plot output is correct. ( Also tried with PyQt5 UI backend )\n\n\n\n\n\n### Code for reproduction\n\n```Python\nimport os, sys, re\nimport math as m\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nfrom matplotlib.colors import LogNorm\n\n# path to test file with data\nfpath = r'D:\\Desktop\\MODES\\test NUMBER3 CL 0_03 -150 TO 120 STEP 5.TXT'\n\nfpath = os.path.normpath(fpath) \nfname = os.path.split(fpath)[1]\n\n# data search\npattern = r\"[-+]?(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:[eE][-+]?\\d+)?\"\n\nheadlines = 0\nwith open(fpath, 'r', encoding='ansi') as (f):\n    # head lines\n    while not re.match(pattern, f.readline().split()[0]):\n        headlines += 1\n            \ndata = pd.read_table(fpath, sep='\\t', skiprows=headlines, \n            names=['freq', 'temp', \"eps_real\", \"eps_imag\",\"sigma\"],                        \n            encoding = \"ISO-8859-1\") \n            \n# str to float\ntry:\n    data.applymap(lambda x: float(x))\nexcept:\n    # pandas ver. \u003e 2.1.0, map instead applymap !\n    data.map(lambda x: float(x))\n    \ndata.sort_values(by=[\"temp\", \"freq\"], ascending=[True, False], \n                                                   inplace=True)\n\n#----------------------------------------\n\nfig, axes = plt.subplots(2,3, figsize=(18, 12),subplot_kw={\"projection\": \"3d\"})\n\ny = data.temp.unique()\ny_len = len(y)\n\nfr = data.loc[data['freq'] == data['freq'][0]]\nx_len = (fr).index[1]\n\nx = (data['freq'].values)[0:x_len]\n\nz = data['eps_imag'].values\n\ncolormap = 'coolwarm' #,  \nlog_norm = LogNorm(vmin=0.1, vmax=z.max())\n\ndef Plot_3D(ax):\n    X, Y = np.meshgrid(x, y)   \n    Z = np.reshape(z, (y_len, x_len))\n    ax.set_xscale('log')\n    ax.set_yscale('log')\n    ax.set_zscale('log')\n    ax.set_xlabel(\"Freq\")\n    ax.set_ylabel(\"Temp\")\n    ax.set_zlabel(\"Eps\")\n    ax.plot_surface(X, Y, Z, cmap= colormap, norm =log_norm)\n\naxes[0,0].remove()\n\naxes[0,1].set_title(\"X,Y,Z log scale\")\nPlot_3D(axes[0,1])\n\naxes[0,2].set_title(\"y \u003e= 0\")\ny = np.maximum(y, 0)\nPlot_3D(axes[0,2])\n\naxes[1,0].set_title(\"y and z \u003e= 0\")\nz = np.maximum(z, 0)\nPlot_3D(axes[1,0])\n\naxes[1,1].set_title(\"y \u003e= 1, z \u003e= 0\")\ny = np.maximum(y, 1)\nPlot_3D(axes[1,1])\n\naxes[1,2].set_title(\"y and z \u003e= 1\")\nz = np.maximum(z, 1)\nPlot_3D(axes[1,2])\n```\n\n### Actual outcome\n\n\n\u003cimg width=\"7598\" height=\"5040\" alt=\"Image\" src=\"https://github.com/user-attachments/assets/9e7e5461-2832-40ce-a307-2107f568190d\" /\u003e\n\n### Expected outcome\n\nCorrect 3d surface plot even with negative of zero input values. (Probably with discarded negative data). And/Or warning about inappropriate data . \n\n### Additional information\n\nI created this test example to observe an issue.\n\nHi, I wrote a program for output experimental 3D dielectric spectra from text data (with a PyQt5 interface). To make the data more understandable, they should be displayed in log scale. Matplotlib v3.11.0rc2 just added this feature. I would like to thank everyone who contributes to matplotlib.\n\nThe input data file is a Eps values ​​versus frequency and temperature. Temperature values ​​in the test data (attached to report) range from -150 to 120°C. Eps values ​​can be negative (although this is incorrect \"overshoot\", I used it to observe an issue).\n\nI noticed that, 3D surface plot axes are behaving strangely, in log scale if the data contains negative or zero values. \n\n\n\n[test NUMBER3 CL 0_03 -150 TO 120 STEP 5.TXT](https://github.com/user-attachments/files/28119242/test.NUMBER3.CL.0_03.-150.TO.120.STEP.5.TXT)\n\n### Operating system\n\nWindows 10 / 11\n\n### Matplotlib Version\n\n3.11.0rc2\n\n### Matplotlib Backend\n\nmodule://matplotlib_inline.backend_inline\n\n### Python version\n\n3.13 / 3.14.2\n\n### Jupyter version\n\n-\n\n### Installation\n\npip","author":{"url":"https://github.com/mrARGJ","@type":"Person","name":"mrARGJ"},"datePublished":"2026-05-21T20:51:13.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":5},"url":"https://github.com/31726/matplotlib/issues/31726"}