https://pfspipe.ipmu.jp/jira This file is an XML representation of an issue en-us 8.3.4 803005 13-09-2019 https://pfspipe.ipmu.jp/jira/browse/PIPE2D-460 DRP 2-D Pipeline <p>Evaluate the accuracy of interpolations of the AMBRE model spectra. The interpolations were provided in <a href="https://pfspipe.ipmu.jp/jira/browse/PIPE2D-365" title="Interpolate between model templates " class="issue-link" data-issue-key="PIPE2D-365"><del>PIPE2D-365</del></a>, where 4D radial basis functions were used for interpolations.</p> PIPE2D-460

Evaluate the interpolations between model templates

Task Normal Done Done Takuji Yamashita Takuji Yamashita flux-calibration model-templates Fri, 27 Sep 2019 06:09:12 +0000 Mon, 11 Nov 2019 20:59:22 +0000 Fri, 8 Nov 2019 06:56:28 +0000 0 3 <p>To evaluate the accuracies, the interpolations were made at each parameter point that we have. <br/> An RBF fitting was performed for every spectrum. A document is in preparation. Target value of the accuracy is &lt;1%.<br/>  <br/> <b>Model templates</b><br/> All the AMBRE templates with 6000 K &lt; Teff &lt; 8000 K</p> <ul> <li>2590 models of the plan-parallel model (p-model)</li> <li>2125 models of the spherical model (s-model)</li> </ul> <p><b>Computer</b><br/> 3.8GHz CPU, 62GB memory, CentOS7, 7 cores are used.<br/>  <br/> <b>Processing time</b><br/> ~10 days for 2590 p-model spectra; ~5 days for 2125 s-model spectral. On average, 5 min for p-model, 3 min for s-model. Most of the interpolations takes only ~10 sec. Interpolations which use many neighbor spectra take much time. <br/> <span class="image-wrap" style=""><a id="12209_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12209/12209_numspec_time.png" title="numspec_time.png" file-preview-type="image" file-preview-id="12209" file-preview-title="numspec_time.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12208/_thumb_12208.png" style="border: 0px solid black" /></a></span> <span class="image-wrap" style=""><a id="12211_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12211/12211_pmodel_rms_4param.png" title="pmodel_rms_4param.png" file-preview-type="image" file-preview-id="12211" file-preview-title="pmodel_rms_4param.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12209/_thumb_12209.png" style="border: 0px solid black" /></a></span><br/>  <br/> <b>RMS of residuals</b><br/> For each interpolation spectrum, a residual is calculated as a function of wavelength.<br/> Residual (%) == (Interpolation - Reference) / Reference * 100, <br/> where Reference is an original AMBRE spectrum.<br/> Overall, the accuracies is good enough. On average, they are 0.03 (<ins>/-0.03)% and 0.02 (</ins>/-0.04)% in p-mode and s-model, respectively. I found that accuracies become worse a bit at lower Teff and high metallicity in both p/s-models.<br/>  <br/> P-model<br/> <span class="image-wrap" style=""><a id="12214_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12214/12214_pmodel_rms_numspec.png" title="pmodel_rms_numspec.png" file-preview-type="image" file-preview-id="12214" file-preview-title="pmodel_rms_numspec.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12211/_thumb_12211.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12215_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12215/12215_pmodel_rms_Teff.png" title="pmodel_rms_Teff.png" file-preview-type="image" file-preview-id="12215" file-preview-title="pmodel_rms_Teff.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12214/_thumb_12214.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12213_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12213/12213_pmodel_rms_logg.png" title="pmodel_rms_logg.png" file-preview-type="image" file-preview-id="12213" file-preview-title="pmodel_rms_logg.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12215/_thumb_12215.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12216_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12216/12216_pmodel_rms_Z.png" title="pmodel_rms_Z.png" file-preview-type="image" file-preview-id="12216" file-preview-title="pmodel_rms_Z.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12213/_thumb_12213.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12216_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12216/12216_pmodel_rms_Z.png" title="pmodel_rms_Z.png" file-preview-type="image" file-preview-id="12216" file-preview-title="pmodel_rms_Z.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12216/_thumb_12216.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12217_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12217/12217_smodel_rms_4param.png" title="smodel_rms_4param.png" file-preview-type="image" file-preview-id="12217" file-preview-title="smodel_rms_4param.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12212/_thumb_12212.png" style="border: 0px solid black" /></a></span><br/>  <br/> S-model<br/> <span class="image-wrap" style=""><a id="12220_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12220/12220_smodel_rms_numspec.png" title="smodel_rms_numspec.png" file-preview-type="image" file-preview-id="12220" file-preview-title="smodel_rms_numspec.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12217/_thumb_12217.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12221_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12221/12221_smodel_rms_Teff.png" title="smodel_rms_Teff.png" file-preview-type="image" file-preview-id="12221" file-preview-title="smodel_rms_Teff.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12220/_thumb_12220.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12222_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12222/12222_smodel_rms_Z.png" title="smodel_rms_Z.png" file-preview-type="image" file-preview-id="12222" file-preview-title="smodel_rms_Z.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12221/_thumb_12221.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12218_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12218/12218_smodel_rms_alpha.png" title="smodel_rms_alpha.png" file-preview-type="image" file-preview-id="12218" file-preview-title="smodel_rms_alpha.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12219/_thumb_12219.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12210_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12210/12210_rms_table.png" title="rms_table.png" file-preview-type="image" file-preview-id="12210" file-preview-title="rms_table.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12222/_thumb_12222.png" style="border: 0px solid black" /></a></span><br/> <span class="image-wrap" style=""><a id="12210_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12210/12210_rms_table.png" title="rms_table.png" file-preview-type="image" file-preview-id="12210" file-preview-title="rms_table.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12218/_thumb_12218.png" style="border: 0px solid black" /></a></span><br/>  </p> <p><span class="image-wrap" style=""><a id="12210_thumb" href="https://pfspipe.ipmu.jp/jira/secure/attachment/12210/12210_rms_table.png" title="rms_table.png" file-preview-type="image" file-preview-id="12210" file-preview-title="rms_table.png"><img src="https://pfspipe.ipmu.jp/jira/secure/thumbnail/12210/_thumb_12210.png" style="border: 0px solid black" /></a></span></p> <p> <b>P and S-models</b><br/> Both models reproduce with high accuracies. The results show that the accuracy of p-model is systematically higher than that of s-model. This is because the parameter grid of the p-model templates is dense like a regular grid while one of s-model is sparse. </p> <p>The results of the evaluations was provided and a code for this work was uploaded at github. </p> Blocks PIPE2D-365 Development Rank 0|02qpt6:508yl0chzzy0000pa Reviewers hassan Sprint 2DDRP-2019 H 2DDRP-2019 I Story Points 4.0