Adams - Science topic

In machine learning, we’ve long been reliant on optimizers as fixed tools: SGD, Adam, RMSProp—tuned and trusted, but largely untouched.

But something new is happening.

We’re entering a new era, where the optimizer isn’t just a background mechanism. It’s learnable. Adaptable. Meta-aware.

This transition is being pioneered by approaches like NeuralGrok and Learnable Gradient Accumulation (LGA). Though they differ in implementation, both share a revolutionary idea: that optimization itself can—and should—learn.

These frameworks don’t just adjust weights. They change how updates happen. They invite the optimizer to evolve along with the model, ushering in a new generation of training systems that are not hand-crafted, but self-improving.

NeuralGrok: Evolving the Gradient, Not Just the Model

Zhou, X., Fan, S., Jaggi, M., & Fu, J. (2025). NeuralGrok: Accelerate Grokking by Neural Gradient Transformation. arXiv preprint arXiv:2504.17243. https://arxiv.org/abs/2504.17243

NeuralGrok introduces a simple but powerful concept: don’t just learn better representations—learn better updates.

At its core, NeuralGrok inserts a learnable module—a gradient transformer—into the training loop. This module reshapes the raw gradient signal before it is applied to the model’s parameters. Rather than treating gradients as sacred and immutable, NeuralGrok asks: what if the way we learn is as important as what we learn?

Don’t just learn better weights—learn better updates.

This idea is especially relevant in the context of “grokking,” where models linger in a memorization phase before suddenly snapping into generalization. NeuralGrok accelerates this transition by optimizing the flow of learning signals themselves.

It periodically meta-trains this transformer to predict and produce gradient directions that are more aligned with generalization objectives—shaping not just the landscape of learning, but the trajectory through it.

Learnable Gradient Accumulation (LGA): Memory / Context with Purpose

Stevens Jr., J. (2025). Learnable Gradient Accumulation. ResearchGate. https://www.researchgate.net/publication/391573797_Learnable_Gradient_Accumulation

Where NeuralGrok transforms gradients step by step, LGA focuses on the broader picture: how gradients are accumulated and applied over time.

Traditional optimizers like momentum or Adam implicitly use memory, but their mechanisms are rigid—decay rates and update rules are static, predefined. LGA breaks this mold by making memory dynamics learnable.

Instead of “when do I update?”, ask “what is the smartest way to apply what I’ve seen?”

LGA introduces a general, modular framework where accumulated gradient states are no longer manually filtered, but instead are governed by parameterized transformations. It provides a general framework for extending this to both temporal and spatial memory mechanisms, enabling updates that take context, history, and task-phase into account.

This function defines how to blend past accumulated gradients with the current gradient signal, using learnable parameters. The framework supports a spectrum of strategies—from simple affine updates, to GRU-style gating, to full recurrent memory systems like LSTMs.

From Memoryless to Meta-Aware

LGA formalizes gradient accumulation as a continuum—interpolating between:

Variants of LGA include:

LGA even supports spatiotemporal filtering, combining low-frequency emphasis (like in Grokfast) with spatial transformations such as projection, gating, or normalization. The goal: amplify what's meaningful, suppress what's not, and guide training with long-term signals that matter.

Perfect for cases where training needs to be smarter to be faster.

I have already fitted many column adsorption kinetic models linear and nonlinear fitting using Orifginpro, MS Excel, and Excel Solver and posted on my YouTube channel: youtube.com/@aminulchem. However, I could not fit the nonlinear Adam Bohart Model model yet. Could you please help me fit the nonlinear Adam Bohart Model model using Originpro?

Hello there

I was wondering about the correct procedure for disposing of a Pd complex catalyst (as PdCl2(dppf) or other) after it is used for a reaction in the lab. I do not feel confident about what I should do.

Thanks for helping!

Adam A Needle

Adam, F., Hassan, M., & Ahmed, N. (2019). Urban migration and housing crisis in Malé: A critical analysis. Island Studies Journal, 14(2), 124-136.

Dear Colleagues and Friends,

We need to get and use the right ruler to measure Health-Related Quality of Life. By then we only can improve their HRQOL.

The HRQ-6D is a free and standard tool to measure HRQOL of people especially patients. Please read our work in Journal of Clinical Medicine.

Thank you so much.

Regards,

Adam

Hi

Has anyone got any experience of using the above kit without vacuum extraction? I was talking to someone last week who was adamant they have done this and without any impact on DNA yield. Worried it seems too good to be true, but wondered if anyone was doing it and had troubleshooted this method?

FOI - i am vacuumless!

Thanks

Genesis

Koran

I need to a salt without any nitrogen in it

Experiments and studies have determined that Neanderthals existed before Adam and Eve. Darwin's theory of evolution is still used, isn't this a contradiction? According to religious teachings, the first humans were Adam and Eve, but studies show something else. What is the reason for this contradiction?Is Darwin right or religious teachings? Why?

My mitsubishi P93D thermal printer has become landfill material because ther are no drivers available to support it. I have a Corning Axygen Gel Doc cabinet. I woudl reall yliek tohave ahtermal printer for documenting resutls. Any ideas / solutions that do not cost $1200? I coudl use a simple deskjust, I guess. Or can I buy a cheap thermal pritner peopel sue for receipts? thanks, Adam

Hello,

Do we have inhibitors specific for the ADAM and/or ADAMTS metalloproteinase family that do not inhibit MMPs?

I found ADAM 10 "specific" inhibitors but nothing on their effect on MMPs.

Thank you

Philippe

We have recently utilised Bambanker (BB01) from Nippon Genetics (Geneflow in UK), which is a serum free DMSO containing cryogenic storage medium for cells. PBMCs (5x10e5/ml) were not happy in this medium when stored at -80C (in an Isopropanol container, althouth I know this is not necessary) for ~12 months (viability dropped considerably from baseline and 1 month data, measured using flow cytometry) and thus we have moved them to Liquid N2. Since then I have found out that Nippon Genetics do not have data on PBMCs rather immortalised cell lines. Moreover I am still not sure whether they use viability and/or growth rate to validate its use.

I am writing to find out other peoples experiences of Bambanker. Going forward I plan to place PBMCs into Bambanker, freeze down at-80 then transfer to Liq N2. Does anybody else do this and have they experienced good (>90% viability?). Thanks Adam Wright

I understand that the following questions may seem basic in terms of Deep Learning, but I would like each of you to share your understanding of these concepts:

I look forward to reading your valuable responses.

In 1990 in Transfusion I published a paper intitled "doi: 10.1046/j.1537-2995.1990.30290162894.x.. 1990 Feb;30(2):109-13.Transfusion

The gel test: a new way to detect red cell antigen-antibody reactions

Y Lapierre , D Rigal, J Adam, D Josef, F Meyer, S Greber, C Drot.

I would like to know all the paers citing this article.

Thank you

Dr D RIGAL

I am trying to interpret some data from an ELISA of an immune response to a bacterial infection. In theory, these values will be useful for understanding how a host responds to its infection.

However, many of the negative control samples returned positive results. To me, this means one of two things. First, we cannot trust or use any of these results, as the values that we "know" are negative were not, in fact, negative. So our results will be inaccurate. Second, the maximum value of the largest false positive could be used as a "threshold", and any value less than or equal to that should be considered as "0".

Has anyone with experience in ELISA encountered similar issues? How did you address them?

Thanks in advance,

Adam

I have preformed a dehyrogenase assay on an arid Negev soil. the calibration curve highest values were about 0.5 but the values for some of my samples were even 2 or 3 times that based on the protocol Alef, K. "Dehydrogenase activity." Methods in applied soil microbiology and biochemistry. London: Academic Press, 1995. 228-231.‏

also the results were much higher on the FDA protocol based on:

Adam, G. and H. Duncan. 2001. Development of a sensitive and rapid method for the

measurement of total microbial activity using fluorescein diacetate (FDA) in a range of

soils. Soil Biol. Biochem. 33:943-951.

the curve values were 0.05 to 0.1 and the values were 0.2 -0.7. is this normal for soils? some of them were poluted 5 years ago

I am trying to detect TREM2 protein in THP-1. It is exposed to the external face of the cell membrane. This is proteolyzed by ADAM proteases, and external part of the protein seems to be released to the culture medium.

I have bought several antibodies and tried several times on monocytic THP-1, on stimulated with PMA to transform them to macrophages, fixing cell to avoid proteolysis and inhibiting ADAM proteases with GW28064X.

On the contrary, of antibodies Technical Data Sheets, I cannot see TREM2 with flow cytometry. I suspect that it is not possible with this technique. We´ve seen protein with difficulties in western blot and IF.

Anyone working in this issue? Any experience?

Is it possible to run an analysis in which the steering wheel angle at zero speed is an input in ADAMS Car software? If possible, could you please explain?

I am training a Beta-VAE using BDD-100k driving dataset. Here are my hyperparameters: Adam optimizer, 0.0001 learning rate, and my latent dimension is 16, loss function is reconstruction loss(MSE) and KLD loss multiplied to Beta factor. After a while of training, the model seems learned something, but with different samples the exact same model's performance is absolutely different. Can anyone give me some hint for how to understand what is going on there? Thanks! Here are examples of same model generating different results.

This superficial velocity is one parameter used in Adam Bohart Model for column studies. This has units cm/min. But it is not mentioned anywhere that which velocity they are talking about. Whether this is the velocity of water running through column but it must have units cm3/min. Kindly clarify.

Hello fellow abaqus users,

I try to solve the following problem:

I have 3 solid bodies (please see the attached file). The bodies are tied to their respective neighbor. Loads to be applied are a thermal load (thermal expansion) and a rotational load around the Z-Axis. The steel body is encastred.

I tried to model the adhesive as cohesive elements with elastic behavior type traction, but were not able to combine this element type with the thermal expansion. Is this even possible? If not, is there a workaround? I am really looking forward to any tips and hints.

Thanks in advance,

Adam

I'm trying to use some machine learning (autoencoder?) to help classify various data for pass/fail. This data will be RF data such as frequency response etc. Does anyone have any recommendations on methods to best accomplish this?

Regards,

Adam

I need to define an objective function that has several decision variables. This objective function needs to be minimised by using Adam optimiser from Tensorflow. There are a lot of examples on deep learning on internet. However, I am finding it difficult to get one that only optimises the objective function.

I am

Created a single stage speed reducer in Adams view. Applied a constant torque on input gear. After simulation, plotted the torque vs time for output shaft. But torque curve is fluctuating not a constant one. As per theory the output torque must be gear ratio times the input torque. I'm not getting it in adams view

I am a student, have a student version of MSC adams and i am doing Dynamic Simulation of vehicle, so, how to give contact force between tire and Slope road? please give me advice as soon as possible..........................

Error:

I did work on some solutions but unfortunately, I could not solve my problem.

The solutions:

1) Working on solvers

2)Making the new environment

3)changing time step

I have also checked these links:

1) http://www.4k8k.xyz/article/Robot_Starscream/105246155

2)https://www.mathworks.com/help/simulink/sfg/integrating-existing-c-functions-into-simulink-models-with-the-legacy-code-tool.html

3)http://www.dspace.com/shared/support/faqpdf/faq202.pdf

4)https://itectec.com/matlab/matlab-how-to-fix-error-in-simulink-adams-co-simulation-could-not-connect-to-adams-server/

5)https://www.youtube.com/watch?v=Wlps1dBc188

Please help and guide me on this problem

Thank you in advance

Hello Research Community,

I am in a bit of a dilemma with my common method bias test in AMOS. I am not sure which is more methodologically correct, whether to run the latent factor test with all the observed variables (from the entire survey), or to just use the variables/items remaining after good model fit?

I know that when running Harman's single latent factor method in SPSS you use all items, so does the same apply to AMOS common latent factor method?

I do know that it is even more accurate to run a marker variable method (however, I do not have this, unfortunately, so can only run a latent factor test)

If so, does anyone have any sources or citations to support this?

Your help is much appreciated

Kindest Regards,

Adam

Hello,

I'm having some problems running PROCESS v 4.0 on my Mac, for SPSS v27.0.

I can get to the stage where I run the Syntax and it looks like it works. But then the PROCESS tab isn't actually available.

There aren't error signs that appear at any stage, so I was wondering if anyone else was having this issue, and whether there are any solutions to this problem?

Any help would be much appreciated.

Adam

Hello everyone,

I am new to structural modelling and I currently have a number of models (2nd order) with a number of variables that are exhibiting very good model fit and validity.

Unfortunately, when running method bias tests I am having the following issues

(for the note I am running the common method factor) where all paths have regression weights as 1 and then 0 and need to do a comparison.

1. Maximum iterations are reached and model is not running

2. I am getting the following message at times "An error occured while attempting to fit the model. During the analysis of a bootstrap sample, an attempt was made to compute the correlation between two variables, one of whose estimated variances failed to be positive. This attempt was made because "Standardized estimates" in the "Analysis Properties" window was checked, or because the Standardized method was used or because a user-defined estimand requires calculating standardized estimates"

I cannot seem to understand why the model is not running and I really good use some advice. Any help is much appreciated

Thanks,

Adam

I completed the isothermal studies Thomas model, BDST, and adam and Bohart models, but don't know how to calculate "Kth, Qo for Thomas model, Kbdst, No for BDST, and Kab, No for the adam and Bohart models. Would you please help me find out these values and send the formulas for those?

I am having an issue while training a custom deep learning model. The training accuracy increases gradually until 0.82 then it drops back to 0.62 for every epoch. The model uses residual blocks with batch normalization and dropout probability of 0.1. The optimizer of the model is Adam and the learning rate is 1e-3. all the activation functions are Relu.

Hello everyone!

I imported a Hypermesh mesh into Adams. I ran the same simulations on both software but got different results. The mesh imported to Adams is a modal analysis with the extension ".mnf". How is it possible that the same mesh and the same simulation lead to two different results?

I applied a force in a master node and measured the displacement. I got two different values from the two Hypermesh and Adams simulations.

I'm working to refine a mass candy making process which requires batches of 12kg be produced in rotation over the course of 24 hours per day, 5 days per week. The current process works in the assumption that by achieving a temperature of 245°F, we're achieving an approximate brix of 86. This brix, combined with our activated gelling agent (Pectin) appears to provide the best results for our purposes; however, a new consultant reviewing the process is adamant that temperature is not a sufficient indicator of Brix for consistent mass production confection manufacturing. I'm hoping to get additional insight from other professionals with experience in the process whom can better clarify whether the cook temp of a sugar syrup mixture is an accurate indicator of Brix.

Hey Everyone! I'd love to have a discussion and see results on flexible thin-film transistors and discuss the direction of this research!

Regards,

Adam

Hi. Everyone. I'm trying to code hyperparameter optimization by using keras tuner.

At first I made layer, node and activation function optimization code

for i in range(hp.Int('num_layers',2,9)):

model.add(keras.layers.Dense(units = hp.Int('units_'+str(i),32,512,16), activation = hp.Choice('a_'+str(i), values=['softmax','relu','tanh','sigmoid'])))

like this.(Of course I made model first)

And then I want to optimize 'Optimizer' and 'learning_rate'

But I can't make the code that optimizes two parameter at the same time.

model.compile(optimizer = hp.Choice('optimizer',['Adam','sgd','rmsprop']),

loss = keras.losses.MeanSquaredError(reduction="auto"),

metrics = ['mae'])

this is my optimizer optimization code.(learning_rate is not added)

I want to add learning_rate optimization code so optimizer and learning_rate parameter is optimized simultaneously

learning_rate = hp.Choice('learning_rate', values = [1e-1, 1e-2])

model.compile(optimizer = keras.optimizers.Adam(learning_rate=hp_learning_rate),

loss = keras.losses.MeanSquaredError(reduction="auto"),

metrics = ['mae'])

Like this, i can make each single parameter optimization.

Could you help me to make double parameter optimization?

Thank you so much!

Please help me know about SDTM (Study Data Tabulation Model) required to submit clinical data to FDA and how can we convert clinical data to SDTM followed by ADaM?

Can we use R studio for STDM, ADaM formatting, and TLF's creation?

I want to know feature importance for the same reason i want to give my model as input to permutation_importance( ) function, which wants model to give output as 0 or 1 only.

My dataset outcome is binary, I want Output of ANN to be either 0 or 1, but i am getting probabilities in the range 0 to 1.

I tried to customise step activation function for output layer , but not much luck.

This is my ANN

model = keras.models.Sequential() model.add(keras.layers.Dense(32,input_dim=a,activation='relu')) ## input layer(8 neurons) --> hidden layer(32) model.add(keras.layers.Dropout(0.2)) model.add(keras.layers.Dense(64,activation="relu")) ## second hidden layer model.add(keras.layers.Dropout(0.2)) model.add(keras.layers.Dense(2,activation="softmax")) ## output layer

model.compile(loss="binary_crossentropy",optimizer='adam',metrics=["accuracy"])

Kindly suggest how can i get output as 0 or 1 for this ANN

Hello Everyone,

I want to use Pre-trained models such as Xception, VGG16, ResNet50, etc for my Deep Learning image recognition project to quick train the model on training set with high accuracy. I am having trouble to find exact code to implement my model. I tried to fit the model with the following code with modifications in output layer. I am getting these errors after running on the Jupyter Notebook.

ValueError: Error when checking input: expected image_input to have 4 dimensions, but got array with shape (224, 224, 3)

ValueError: The shape of the input to "Flatten" is not fully defined (got (None, None, 512). Make sure to pass a complete "input_shape"

ValueError : Error when checking target (output shape error)

Can any one please help to find the right codes? Thank you so much for help in advance :)

The paper below has Madison Brazell as an author. It should be Michael Brazell. I cannot find Madison Brazell listed as faculty at Georgia College. Does this mean that someone is claiming a paper that they had no part in or just a honest mistake? Not being able to find the person where they are working is odd. Any answers to this?

Electrocoating carbon fiber microelectrodes with Nafion improves selectivity for electroactive neurotransmitters

Good day! I am Zachary Origenes, a mechanical engineering student in the Philippines and I am currently doing my thesis which involves validating my simulation model of an in-pipe inspection robot made in MSC Adams with the physical model of a previous study.

I have searched on what range the error can be acceptable when validating simulation results with the physical results of multibody systems, but I've only found that it's 5-20% only when it comes to FEA (Finite Element Analysis).

Hi,

We all know the dispersion relation for an ideal graphene in hexagonal lattice. Since graphene is polycrystalline by nature, assuming a perfect graphene crystals with average size La, nm connected by pentagons and heptagons randomly distributed in their edges, what will be the dispersion relation of such material? Can we theoretically quantify the effect of the domain size on the dispersion relation? Is there experimental evidences ?

For example in CVD graphene transferred onto SiO2 substrates, the bulk graphene resistivity ρ, was measured by the four-point probe technique and correlated to crystallite size La by Vlassiouk et al. [1] through the following equation:

ρ(Ω)=6.7105/La (nm)

Another method of estimating the conductivity of arbitrarily stacked graphene sheets was proposed from semi-classical Boltzmann transport theory could be found here [2-4].

Thanks,

{1} Vlassiouk, I.; Smirnov, S.; Ivanov, I.; Fulvio, P.F.; Dai, S.; Meyer, H.; Chi, M.; Hensley, D.; Datskos, P.; Lavrik, N.V. Electrical and thermal conductivity of low temperature CVD graphene: The effect of disorder. Nanotechnology 2011, 22, 275716. [Google Scholar] [CrossRef] [PubMed]

{2} Min, H.; Jain, P.; Adam, S.; Stiles, M.D. Semiclassical Boltzmann transport theory for graphene multilayers. Phys. Rev. B 2011, 83, 195117. [Google Scholar] [CrossRef]

{3} Chen, J.H.; Cullen, W.G.; Jang, C.; Fuhrer, M.S.; Williams, E.D. Defect scattering in graphene. Phys. Rev. Lett. 2009, 102, 236805. [Google Scholar] [CrossRef] [PubMed]

{4} Stauber, T.; Peres, N.M.R.; Guinea, F. Electronic transport in graphene: A semiclassical approach including midgap states. Phys. Rev. B 2007, 76, 205423. [Google Scholar] [CrossRef]

In 2015, I upload a working paper that is now published in another version (publication in 2020). Between these two version, authorship changed. Is it possible to remove the following items (the DOI associated with this items is the one 2020 not of 2015).

Thank you very mmuch,

Laurie Bréban

A missing touch of Adam Smith in Amartya Sen's Public Reasoning : the Man Within for the Man Without

I have tried different pretrained networks with Adam optimizer, and obtained high performance. However, when I tried it with Alex network, the performance was poor. Could you please help me understanding the reason?

Thanks

I am working on comparing the performance of evolutionary algorithms in optimizing a suspension model. The suspension model is built using ADAMS/Car. I need the best way to implement these algorithms to work in fatest way with our model.

I am using google colab to train a model for sign language recognition using this dataset: https://www.kaggle.com/feronial/turkish-sign-languagefinger-spelling

I tried different architecture for my CNN model but this one gives me the best results so far:

l2 = regularizers.l2(0.001) model = Sequential([     Conv2D(16, 3, padding='same', activation='relu',kernel_regularizer=l2,            input_shape=(224, 224 ,3)),     MaxPooling2D(),     Dropout(0.35),     Conv2D(32, 3, padding='same', activation='relu',kernel_regularizer=l2),     MaxPooling2D(),     Dropout(0.35),     Conv2D(64, 3, padding='same', activation='relu',kernel_regularizer=l2),     MaxPooling2D(),     Dropout(0.35),     Conv2D(128, 3, padding='same', activation='relu',kernel_regularizer=l2),     MaxPooling2D(),     Dropout(0.35),     Flatten(),     Dense(1024, activation='relu',kernel_regularizer=l2),     Dropout(0.35),     Dense(512, activation='relu',kernel_regularizer=l2),     Dropout(0.35),     Dense(23,activation="softmax") ])

i am using an ADAM optimizer with lr=0.001 and batch size of 32

i tried training for 50,100,200 epochs but the results weren't so much different.

I am getting 99-100% accuracy on training,

but only 70-75% accuracy on validation,

and a very low 30-40% accuracy on test.

can anyone point in a direction to improve these numbers?

This discussion is part of my Dissertation of the history of Microsoft and MSDOS 1.0.

For 10 years i downloaded the book of WPF4 Unleashed. The Author of this great book is Adam Nathan.

By the time i was a student.

So currently i'm narrating the book to investigate it for historical purpose.

Here are some of my conclusions.

1. Adam Nathan was in the core of the Windows Presentation Fondation WPF.

2. His FotoGallery Example was not completed and could have benefit to billions of people by simply visualizing fotos. I am one of the few people that use Picasa3 and i am currently developping a new version with WPF : Picasa 4 Photo Gallery

1. Author 1: Adam Nathan

2. Author 2: Kais el Kara

- Also i also developped a Radioplayer and this example was not present in the book of Adam.

- Also I build a new Notepad with Multitabs and numerated lines.

I used TabControl.

This examples should be completed and be available in a new version of the book 2020.

- More of this Microsofts News in the next Days Months.

Ps. Dont Forget Picasa For Archeology.

Google was too smart to end its developpment in 2015.

And Microsoft was not active enough to buy the product from Google to integrate it in Windows2020-PicassaPicassoEdition

Dissertation started back in 2019.

Experience 20 years.

xoxo

iooi

iooi

oxox

22:59KartagoTZ

I am currently working on Co-simulation of ADAMS-simulink environment. After modelling full vehicle in MSC ADAMS, I need to create state variables to define them as input and output variables. The output variables from the adams model will be Speed of all 4 wheels, vehicle slip angle and yaw rate. The input controlled state variables will be brakes force to all four wheels. It will be done in ADAMS view. But I dont know how to create the state variables. So, anybody please help me out with the required guidance or any tutorial or required simulink files, it will be great help.

Thsi is my article: I made simmilar research in Poland, if you are interested I will be really honored if we can cooperate:

sfile:///C:/Users/ThinkPad/Downloads/79-Article%20Text-218-1-10-20180306.pdf

Sonia Dzierzyńska- Breś

Adam MIckiewicz University in Poznań

The narratives in Genesis are silent about God warning Adam, Eve or both not to touch the tree of the knowledge of good and evil until Eve's claim that God says so: but of the fruit of the tree which is in the midst of the garden, God has said, ‘You shall not eat it, nor shall you touch it, lest you die (Gen. 3:3). Did Eve tell a lie, exaggerate, say what she had heard from the husband, an angel or God Himself? Please give reason(s).

This is not a question but a 'cut and paste' that the computer I'm using right now decided was a question. So going with the 'flow' so to speak I will try to turn this 'cut and paste' into a question.

Why buy the above when one already has an earlier version of it?

Answer: Because it was there and because I can and because in the words Adam West )who played the best Batman: 'Poor Deluded Woman/Girl!!!'

I have access to a LiveCyte Phasefocus (without heat, CO2 capability) that has a slide adaptor able to take ibidi mu chemotaxis slides. I wish to monitor monocyte migration in real time using these two pieces of equipment. Has anyone been able to achieve this ? Thanks Adam

I have been culturing with M199 with 10% FBS and 1%P/S encountering many problems for long sometimes after resuscitation or after passage and even after glass slide implantation the cell dies , try to figure out the problem but still couldn't find out the reason .

please i need some suggestion and advice to deal with this issue

Best regards.

Shafiu Adam Shinge .

I have a multivariate time-series forecasting problem which I am trying to solve using LSTM models. I have transformed my data such that it looks back last 20 timestamps to reconstruct the records at current timestamp, using deep LSTM model. Following is the model architecture :

This question is for Principal Investigators. How long should we retain old physical laboratory notebooks from our former students, postdocs, and technicians? This is assuming that all data have been published and that there are no intellectual property concerns. I have notebooks now approaching 20 years old and would appreciate any feedback from other PIs as to how they deal with this issue. Thanks, Adam

I have been culturing with M199 with 10% FBS and 1%P/S encountering many problems for long sometimes after resuscitation or after passage and even after glass slide implantation the cell dies , try to figure out the problem but still couldn't find out the reason .

please i need some suggestion and advice to deal with this issue

Best regards.

Shafiu Adam Shinge .

I have started a research on the modeling of multibody systems dynamics, and since dynamics mean to determine the Differential Equations of Motion, i am called to make the resolution of these OEM.

I need to know how to do so, so i started working on an easy example, but i find it difficult because i have no background on mathematics especially numerical integration.

My question is how to proceed to solve a DAE numerically using one of these methods ; Runge kutta, Adams - Bashford or Moulton, Central differences or Gaussian method?

Hello Everyone,

I am wanting to PCR a whole vector (except a region which I am trying to delete). The size of my desired product is around 6kb. My primers are as follows:

1. 5 - TCCAATTTGTTAAAGACAGG

2. 5 - GGTTGTGGCCATATTATCATC

Using phusion polymerase, I have been trying touchdown PCR with a protocol as follows:

30sec - 98

30sec - 65 to 55 (0.5 increments)

10min - 72

--- 20 cycles ---

30sec - 98

30sec - 55

10min - 72

I keep getting terrible smearing and an unclear band. I am running a control which is suppose to amplify 10kb (from the phusion kit made with costume DNA and primers), and it is performing this no problem. Thus, I am not sure what is causing the problem.

Can anyone provide any suggestions on how to get this to work?

Adam

Can anyone recommend a good anti-human DP1 (type-1 prostanoid receptor) antibody for either flow cytometry use? I appreciate there might be antibodies out there not validated for flow cytometry so if you know of one that works well in western blot, immunofluorescence etc then please let me know....

I am looking to confirm DP1 cell surface expression on Flag-tagged DP1 transfected HEK cells and primary human Basophils.

Thanks

Adam Wright

Dear coleagues,

I'm trying to use R to perform geometric morphometrics and I'm facing difficulties in inputing my data.

I've de Geomorph publication (Adams & Otárola-Castillo 2013), and in their example the array has already the two different groups.

I have 2 different .tps files of the groups I want to test, how do I work with 2 different files in R ? Or should I create a single .tps file with both groups, if so, how do I tell in .tps which image belongs to each group ??

Thanks in advance !

es gibt soviel ich weiß keine genauen empirisch belegten Angaben zur Zahl der Wörter im Kroatischen. Diese wird auf cca. 400 000 geschätzt, davon etwa 20 bis 30 % Fremdwörter (auch nicht belegt - nur Schätzungswert). Nach meinen Untersuchungen gibt es cca. 2000 Germanismen, wobei es noch viel mehr gibt, wenn man die indirekten Entlehnungen (z.B. kr. bojler < dt. Boiler < eng. boiler) hinzuzählt. Die frühesten Entlehnungen gehen noch auf die slawisch-germanischen Sprachkontakte zurück (z.B. hljeb). Betrachtet man den Sprachkontakt seit der Zeit, als sich das Deutsche und Kroatische als Einzelsprachen herausbildeten, dann geht es um Germanismen, die die fränkischen Missionare im kroatischen Sprachraum verbreitet haben (7. bis 9. Jh., z.B. crkva, kloštar u.Ä.).

Diachronische und synchronische Aspekte des deutsch-kroatischen Sprachkontaktes habe ich ziemlich ausführlich in der Monographie Deutsch-kroatische Sprachkontakte, die im Narr Verlag erschienen ist, dargestellt (mit Wortlisten). Dort sind auch weiterführende Literatur (zu bestimmten Aspekten dieses Sprachkontaktes).

Hi everybody,

I would like to convert a pre-trained Caffe model to Keras format. If I understood correctly I have to extract the output shape and the weights of the Caffe model, build a model in Keras with the same shape and then upload the weights.

This is the shape and weights of the pretrained model

The output shape of layer data is (3, 400, 400) The output shape of layer conv1 is (64, 98, 98) The output shape of layer norm1 is (64, 98, 98) The output shape of layer pool1 is (64, 49, 49) The output shape of layer conv2 is (256, 49, 49) The output shape of layer norm2 is (256, 49, 49) The output shape of layer pool2 is (256, 24, 24) The output shape of layer conv3 is (256, 24, 24) The output shape of layer conv4 is (256, 24, 24) The output shape of layer conv5 is (256, 24, 24) The output shape of layer pool5 is (256, 12, 12) The output shape of layer conv6 is (4096, 2, 2) The output shape of layer conv7 is (4096, 2, 2) The output shape of layer conv8 is (3, 2, 2) The output shape of layer pool6 is (3, 1, 1) The output shape of layer prob is (3, 1, 1) conv1 (64, 3, 11, 11) (64,) conv2 (256, 64, 5, 5) (256,) conv3 (256, 256, 3, 3) (256,) conv4 (256, 256, 3, 3) (256,) conv5 (256, 256, 3, 3) (256,) conv6 (4096, 256, 6, 6) (4096,) conv7 (4096, 4096, 1, 1) (4096,) conv8 (3, 4096, 1, 1) (3,)

This is the shape I programmed in Keras using as reference this link

data = Input(shape=(3,400,400), name='DATA') conv1 = Convolution2D(64,11,11, subsample=(2,2))(data) conv1 = Activation('relu', name='CONV1')(conv1) norm1= BatchNormalization(name='NORM1')(conv1) pool1 = MaxPooling2D(pool_size=(3,3), strides=(2,2), name= "POOL1")(norm1) conv2 = Convolution2D(256,5,5)(pool1) conv2 = Activation('relu', name='CONV2')(conv2) norm2= BatchNormalization(name='NORM2')(conv2) pool2 = MaxPooling2D(pool_size=(3,3),strides=(2,2), name='POOL2')(norm2) conv3 = Convolution2D(256,3,3)(pool2) conv3 = Activation('relu', name='CONV3')(conv3) conv4 = Convolution2D(256,3,3)(conv3) conv4 = Activation('relu', name='CONV4')(conv4) conv5 = Convolution2D(256,3,3)(conv4) conv5 = Activation('relu', name='CONV5')(conv5) pool5 = MaxPooling2D(pool_size=(3,3),strides=(2,2), name='POOL5')(conv5) conv6 = Convolution2D(4096,6,6)(pool5) conv6 = Activation('relu', name='CONV6')(conv6) conv7 = Convolution2D(4096,1,1)(conv6) conv7 = Activation('relu', name='CONV7')(conv7) conv8 = Convolution2D(3,1,1)(conv7) conv8 = Activation('relu', name='CONV8')(conv8) pool6 = MaxPooling2D(pool_size=(3,3),strides=(2,2), name='POOL6')(conv8) prob = Activation('softmax', name='PROB')(pool6) model = Model(input=[data],output=[prob]) model.compile(optimizer='adam', loss='categorical_crossentropy')

But the model.summary() looks very different from the original I don't understand how to obtain the exact same shape.

The source of information of the pre-trained model is https://github.com/nealjean/predicting-poverty

_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= DATA (InputLayer) (None, 3, 400, 400) 0 _________________________________________________________________ conv2d_7 (Conv2D) (None, 64, 195, 195) 23296 _________________________________________________________________ CONV1 (Activation) (None, 64, 195, 195) 0 _________________________________________________________________ NORM1 (BatchNormalization) (None, 64, 195, 195) 780 _________________________________________________________________ POOL1 (MaxPooling2D) (None, 64, 97, 97) 0 _________________________________________________________________ conv2d_8 (Conv2D) (None, 256, 93, 93) 409856 _________________________________________________________________ CONV2 (Activation) (None, 256, 93, 93) 0 _________________________________________________________________ POOL2 (MaxPooling2D) (None, 256, 46, 46) 0 _________________________________________________________________ conv2d_9 (Conv2D) (None, 256, 44, 44) 590080 _________________________________________________________________ CONV3 (Activation) (None, 256, 44, 44) 0 _________________________________________________________________ conv2d_10 (Conv2D) (None, 256, 42, 42) 590080 _________________________________________________________________ CONV4 (Activation) (None, 256, 42, 42) 0 _________________________________________________________________ conv2d_11 (Conv2D) (None, 256, 40, 40) 590080 _________________________________________________________________ CONV5 (Activation) (None, 256, 40, 40) 0 _________________________________________________________________ POOL5 (MaxPooling2D) (None, 256, 13, 13) 0 _________________________________________________________________ conv2d_12 (Conv2D) (None, 4096, 8, 8) 37752832 _________________________________________________________________ CONV6 (Activation) (None, 4096, 8, 8) 0

Recently we have synthesized several CPP (cell penetrating peptide) and one of them appeared to be located straight into nucleus of the cell (fibroblast and keracynotes cell line). As we are chemist/biochemist sitting on the edge of cell biology we are looking for cooperation to investigate properties of this peptide and possible mechanism and so on. Please send me a private message if interesting.

best

Adam

,,

Or as Londoners say-''Would you Adam and Eve it'.

We must redefine ourselves based on the growing reality of species diversity, but also in terms of why only one has survived.

https://www.livescience.com/65201-newfound-ancient-human-relative-homo-luzonensis.html?utm_source=notification

Hello erudite personalities,

I would like to ask if anyone has worked in the past with trying to form Candida krusei biofilm. How cloudy and how adhering was the biofilm? I used 96-well plate and anytime I try to wash the biofilm before XTT assay, all the biofilm will most entirely wash off.

Also what strain of Candida krusei did you use in the past?. I have tried different protocols in various articles, but my strains here are just adamant. That is why I decide to request for a more physical and scholarly advice here.

Thanks all

I am attempting to run mixing models with isotopic data of various consumer groups collected in upland pine systems in Texas. Does anyone know of any references to TEF values of sources in terrestrial environments that I can use to run the model? My sources consist of three groups:

C4 grasses (Bluestem grasses, Sedges, ect.)

C3 group (American beauty berry, Ilex spp.)

Detritus (Pine straw, Quercus spp. leaves)

Thanks everyone!

Connor Adams

Simulating the driver assistance system with IPG car maker and MATLAB-Simulink, or can ADAMS be used in place of that?

ok , so i am doing an adams simulation of a drag chain carrying several stiff cable of bending stiffness k, to model it what should be the value of torsional stiffness i should be using for the joints of the dragchain.

I am making a disc cutter Modell of a TBM in Msc Adams to analyse the vibration behaviour of the disc cutter. To make the Tunnel surface deformable, so that I get penetration, I need to make a mesh. I tried to make the box flexible, but it did not work. I need also to get the diagrams of the different forces that affect the disc cutter in contact area with the Rock. Thank you very much.

I don't really know if it is a problem with designing my subsystems ( I ran subsystem simulation analysis before assembling and no errors came up ) or problem with my full-car simulation test inputs ?

FYI : this is a suspension project (Front : Double Wishbone and Rear : Trailing arm )

my error and inputs are uploaded .

Be delighted to hear you views and get rid of this plight .

thanks

نمیدونم مشکل از زیر سیستم هست و یا اینکه داده های تست شبیه سازی ؟ چون تست روی زیر سیستم های جلو و عقب ، در ادمز کار ، انجام دادم ولی هیچ اروری نداشت ولی در تست فول کار ارور میده که عکسش را گذاشتم .

I would like to know where can I find out, whether the scholer gave me access to the requested article.

Thanks a lot,

Adam